71255 2 About the Health Equity and Financial Protection Reports The Health Equity and Financial Protection reports are short country-specific volumes that provide a picture of equity and financial protection in the health sectors of low- and middle-income countries. Topics covered include: inequalities in health outcomes, health behavior and health care utilization; benefit incidence analysis; financial protection; and the progressivity of health care financing. Data are drawn from the Demographic and Health Surveys, World Health Surveys, Multiple Indicator Cluster Surveys, Living Standards and Measurement Surveys, as well as other household surveys, and use a common set of health indicators for all countries in the series. All analyses are conducted using the health modules of the ADePT software. Also available are Health Equity and Financial Protection datasheets that summarize key measures of equity and financial protection. The most recent versions of the Health Equity and Financial Protection reports and datasheets can be downloaded at www.worldbank.org/povertyandhealth. Full citation: World Bank. 2012. Health Equity and Financial Protection Report – Malawi. Washington, D.C.: World Bank. i Acknowledgements This report was produced and written by a task team consisting of Caryn Bredenkamp (TTL, Health Economist, HDNHE), Adam Wagstaff (Research Manager, DECHD), Marcel Bilger (consultant), Leander Buisman (consultant) and Leah Prencipe (consultant) under the overall supervision of Nicole Klingen (Sector Manager, HDNHE) and Cristian Baeza (Sector Director, HDNHE). The authors would also like to thank John Paul Clark (AFTHE) for advice on data and documentation, Ellen Van de Poel (consultant) for advice on data, Andrea Thoumi (consultant) for assistance with editing, Emiliana Gunawan (HDNHE) and Daniela Hoshino (HDNHE) for administrative support, and Devon Rohr (consultant) for graphic design. The financial contributions of the Rapid Social Results Trust Fund (RSR-MDTF) and the Trust Fund for Environmentally and Socially Sustainable Development (TFESSD) are gratefully acknowledged. This version: May 17, 2012 ii List of Acronyms and Acronyms ARI Acute respiratory infection BIA Benefit-incidence analysis CHAM Christian Health Association of Malawi CPI Consumer price index DHS Demographic and Health Survey GDP Gross domestic product GHE Government health expenditures HSSP Health Sector Strategic Plan IHS Integrated Household Survey LCU Local currency units MCH Maternal and child health MICS Multiple Indicator Cluster Survey NHA National Health Accounts PPP Purchasing power parity VAT Value added tax VCT Voluntary counseling and testing WHO World Health Organization WHS World Health Survey iii HEALTH EQUITY AND FINANCIAL PROTECTION IN MALAWI Contents Executive Summary...................................................................................................................................... vi 1 Malawi’s health system ........................................................................................................................ 1 1.1 Equity and financial protection as policy goals ............................................................................. 1 1.2 Health financing system ................................................................................................................ 1 1.3 Health care delivery system .......................................................................................................... 4 2 Inequalities in health............................................................................................................................. 5 2.1 Data availability............................................................................................................................. 5 2.2 Inequalities in health..................................................................................................................... 5 3 Inequalities in health care utilization .................................................................................................... 9 3.1 Data availability............................................................................................................................. 9 3.2 Inequalities in health care utilization ............................................................................................ 9 4 Benefit incidence of government spending ........................................................................................ 12 4.1 Data availability........................................................................................................................... 12 4.2 Inequalities in benefit incidence ................................................................................................. 13 5 Financial protection in health ............................................................................................................. 16 5.1 Data availability........................................................................................................................... 16 5.2 Catastrophic out-of-pocket payments ........................................................................................ 16 5.3 Impoverishing out-of-pocket payments ..................................................................................... 17 6 References .......................................................................................................................................... 21 7 Annexes ............................................................................................................................................... 22 7.1 Measurement of indicators ........................................................................................................ 22 7.2 Methodological notes ................................................................................................................. 25 iv Figures Figure 1.1: Health care financing mix, 2005-2009 ........................................................................................ 3 Figure 1.2: Composition of out-of-pocket health spending.......................................................................... 3 Figure 5.1: The impoverishing effect of out-of-pocket spending ............................................................... 20 Tables Table 1.1: Health expenditure data, 2009 .................................................................................................... 2 Table 2.1: Inequalities in child health ........................................................................................................... 6 Table 2.2: Inequalities in adult health .......................................................................................................... 7 Table 2.3: Inequalities in health behaviors ................................................................................................... 8 Table 3.1: Inequalities in maternal and child health interventions ............................................................ 10 Table 3.2: Inequalities in adult preventive care.......................................................................................... 10 Table 3.3: Inequalities in adult curative care .............................................................................................. 11 Table 4.1: Inequalities in use of publicly financed facilities........................................................................ 13 Table 4.2: Distribution in fees paid ............................................................................................................. 14 Table 4.3: Inequality in the incidence of government health spending (shares) ....................................... 15 Table 5.1: Incidence of catastrophic out-of-pocket spending .................................................................... 17 Table 5.2: Impoverishment through out-of-pocket health spending ......................................................... 18 v Executive Summary This report analyses equity and financial protection in the health sector of Malawi. In particular, it examines inequalities in health outcomes, health behavior and health care utilization; benefit incidence analysis; and financial protection. Data are drawn from the 2004 Malawi Demographic and Health Survey, the 2006 Malawi Multiple Indicator Cluster Survey, the 2004 Malawi Integrated Household Survey, the 2003 Malawi World Health Survey and the 2002-03 Malawi National Health Accounts. All analyses are conducted using original survey data and employ the health modules of the ADePT software. Is ill health more concentrated among the poor? Yes. In general, ill health is more concentrated among the poor in Malawi. This includes virtually every one of the selected indicators of child health: infant and under-five mortality, stunting, underweight, diarrheal diseases, acute respiratory infection (ARI) and fever. Malaria alone appears to be slightly more prevalent among the better-off. With respect to measures of adult health, some conditions are concentrated among the poor (such as tuberculosis, angina, arthritis, and measures of difficulty with work and household activities) whereas HIV positive prevalence and obesity (among non-pregnant women) are more common among the better-off. Results for all other indicators are not statistically significant. With respect to risky health behaviors, the results suggest that the poorer populations are significantly more likely to exhibit unhealthy behaviors, such as smoking, insufficient physical activity, and having multiple sexual partners (at least for one of two available data sources for this last indicator). It is the wealthy who are more likely to have healthy behaviors, namely using condoms during concurrent partnerships and using mosquito nets. Do the poor use health services less than the rich? Yes, for almost all services. All three household surveys used – the 2003 WHS, 2004 DHS, and the 2006 MICS – showed that utilization of most types of care are significantly more concentrated among the better-off. Of the selected maternal and child health (MCH) interventions, all – childhood immunization, treatment of diarrhea and ARI, antenatal care take-up, skilled birth attendance and contraceptive prevalence – are more concentrated among the better-off part of the population. Among adult preventive services, screening for tuberculosis and voluntary testing and counseling for HIV are more likely to be utilized by the wealthy. For adult curative care, almost all indicators of inpatient and outpatient care suggest that health services are disproportionately consumed by the wealthy as well, Is the distribution of government spending on health pro-rich or pro-poor? Mildly pro-rich. Government subsidies to all public facilities do not appear to be particular pro-rich or particular pro-poor when using two of the three methods of conducting benefit-incidence analysis. Moreover, the results are not statistically significant. Only when the third (alternative) assumption of proportional cost is invoked does government spending at outpatient clinics, health centers, vi dispensaries or maternity centers, and most strongly inpatient hospitals, become significantly pro-rich – a result that is driven by the fact that the fee burden lies heavily on the wealthy. What is the effect of out-of-pocket payments on household financial well-being? Not too severe. According to the 2003 WHS, about 11.5 per cent of households spend more than 10 per cent of total household consumption on out-of-pocket health payments and only 3 per cent spend more than 40 per cent. Using the alternative nonfood measure, approximately 39 per cent of households spend more than 10 per cent of nonfood consumption on out-of-pocket payments and around 21 per cent spend more than 40 per cent. The 2003 Malawi WHS finds unambiguously that catastrophic payments are found to be concentrated among the poor using both total and nonfood measure of consumption, but the 2004 Malawi Integrated Household Survey rich finds that catastrophic payments are concentrated among the poor when the nonfood measure is used, but among the rich when the total consumption measure is used. Health spending contributes to impoverishment, but the effect is very slight. Out-of-pocket payments are responsible for an increase in the poverty rate equivalent to 0.1 per cent, when using the US$2.00 a day measure, and 0.4 per cent, when using the US$1.25 a day measure, as well as an increase in the depth of poverty (i.e. the poverty gap) of between 1 and 2 per cent (depending on the poverty line). vii 1 Malawi’s health system This section provides a brief overview of Malawi’s health system, focusing on features that are likely to be especially salient for equity and financial protection. 1.1 Equity and financial protection as policy goals Malawi‘s government is committed to improving equity and financial protection in the health sector. Equity is explicitly mentioned as one of the four objectives in the Health Sector Strategic Plan (HSSP). The opening section of the strategic plan illustrates this commitment: “The overall objective of the HSSP is to contribute towards Malawi’s attainment of the health and related Millennium Development Goals. The specific objectives of the HSSP are, therefore, to: • Increase coverage of the high quality Essential Health Package services. • Reduce risk factors to health. • Improve equity and efficiency in the delivery of quality EHP services. • Strengthen the performance of the health system to support delivery of EHP services� (Malawi 2011). 1.2 Health financing system Health expenditure Malawi spends 6.2 per cent (2009) of its gross domestic product (GDP) on health. This is similar to the average spending in other lower income countries in Africa, which have spent an average of 6.5 per cent (2009) of their GDP on health 1. Government spending on health, as a share of total government expenditures, has fallen rapidly in recent years from 20.0 per cent of Malawi’s total government expenditures in 2005 to 12.1 per cent in 2009 (World Health Organization 2009). Despite the decline in the government’s allocation to the health sector, government expenditures still accounted for 58.0 per cent of the total health expenditure in 2009. Private expenditures accounted for 42.0 per cent of the total health expenditures over the same period. Unlike most countries in Sub-Saharan Africa, out-of- pocket payments represented only a small fraction, 11.9 per cent in 2009, of total health expenditures. 1 Non-weighted average of: Benin, Burkina Faso, Burundi, Central African Republic, Chad, Democratic Republic of Congo, Eritrea, Ethiopia, Gambia, Guinea, Guinea-Bissau, Kenya, Liberia, Madagascar, Malawi, Mali, Mozambique, Niger, Rwanda, Sierra Leone, Togo, Uganda, and Tanzania 1 Table 1.1: Health expenditure data, 2009 Indicator Health expenditure as share of GDP 6.2% Government expenditure as share of GDP 30.0% Government expenditure on health as share of total government 12.1% expenditure Government health expenditure, per capita US$11.1 (current), US$29.2 (PPP-adjusted) Government expenditure on health as share of total health 58.0% expenditure Out-of-pocket expenditure on health as share of total health 11.9% expenditure Source: WHO National Health Accounts database (2009) Decentralization and centralization The central Ministry of Health is responsible for the development and enforcement of health policy, regulation of the health sector, creation of standards and norms, allocation and management of resources, provision of technical support, coordination, and monitoring and evaluation. Under the Decentralization Act of 1997, the responsibility of delivering health services at the district and lower levels was devolved to the Ministry of Local Government and Rural Development. Staffing, procurement, contracting are generally still managed from the central level. However central hospitals and district health offices may handle some of their own procurement within the decentralization framework. Revenue-raising/sources of funds Government spending accounted for 58.0 per cent (2009) of total health spending (see Figure 1.1). However, it is important to note that a substantial proportion of those funds came from donors in the form of sector support, and project based funding. In fact, 61.5 per cent of total health expenditures reported in 2004 originated from donors (Ministry of Health and Government of Malawi 2007). The majority of the money channeled through the Malawian Ministry of Health was used to pay for supply side subsidies of curative care at hospitals. According to the 2004/2005 NHA, 66 per cent of government funds were allocated to hospital services, compared to only 21 per cent for health centers, and 6 per cent for preventive services (Ministry of Health and Government of Malawi 2007). As noted earlier, due in part to the large amount of donor and government money for the health sector, out-of-pocket spending only accounted for 11.9 per cent (2009) of total health expenditures. In 2004, out-of-pocket payments were primarily used to access care at private for-profit and not-for-profit hospitals (39 per cent) and government hospitals (24 per cent), but a significant share was also spent on drugs (14 per cent) (see Figure 1.2). 2 Figure 1.1: Health care financing mix, 2005-2009 100% 90% 80% 70% 60% Government Sources 50% Other Private Sources 40% Private Insurance 30% Out-of-Pocket Payments 20% 10% 0% 2005 2006 2007 2008 2009 Source: World Health Organization (2009) Figure 1.2: Composition of out-of-pocket health spending 100% 90% 80% 70% Others 60% Drug Stores and Chemists 50% Private Clinics 40% For-Profit Hospitals 30% Non-Profit Hospitals 20% Government Hospitals 10% 0% 2004 Source: World Health Organization (2009) 3 Risk-pooling Malawi currently has no social insurance specifically for health. Private insurance is the main form of risk pooling found in Malawi, but remains uncommon, accounting for only 6 per cent of total health expenditures in 2009 (World Health Organization 2009). However, private insurance has been growing at a rapid pace: between 2005 and 2009, expenditures by private insurance firms rose 158 per cent (World Health Organization 2009). 1.3 Health care delivery system Provider organization Almost all formal health care services in Malawi are provided by three agencies. The Ministry of health provides about 60 per cent of services, follow by the Christian Health Association of Malawi (CHAM) providing 37 per cent, and lastly the Ministry of Local Government providing 1%. CHAM consists of independent faith-based health facilities. The government works with CHAM by providing an annual grant that covers the salaries of local staff. The CHAM facilities charge user fees with some exemptions for specific interventions and treatments. The government health facilities are organized into three levels: primary, secondary, and tertiary. Primary level facilities consist of rural hospitals, health centers, posts, and clinics. The secondary level includes district hospitals. The tertiary level offers similar services as the secondary level but with the addition of some specialist surgical and other medical interventions. These facilities provide the Essential Health Package in Malawi. Payment mechanisms and provider autonomy Malawi current operates under a “free care� system where primary care at public facilities does not have any cost-sharing. Payroll, procurement, and operating expenses are controlled at the central level for the most part and public providers do not have much autonomy. As the implementation of decentralization progresses, more functions held by the central level will devolve to the district level. However, providers will still be funded and operated through a government agency. Resource availability and utilization Malawi has a relatively good network of health facilities for its size. There were 11 hospital beds per 10,000 persons in 2007 and approximately 85 per cent of the population lives within ten km of a health facility (Ministry of Health and Government of Malawi 2007). However, there is a dearth of physicians in Malawi that raises concerns about access to quality care, especially in rural areas. There were only 0.19 physicians per 10,000 persons in 2008 and 11 hospital beds per 10,000 persons in 2007. 4 2 Inequalities in health Most policymakers regard large inequalities in health outcomes between poor and rich as undesirable. This section reports inequalities in child and adult health outcomes, as well as health behaviors. 2.1 Data availability A Demographic and Health Survey (DHS) was fielded in Malawi in 2004, a Multiple Indicator Cluster Survey (MICS) in 2006, and a World Health Survey (WHS) in 2003. Although the DHS and MICS have rich information for many health outcomes, particularly in relation to child health, the WHS has fuller data availability with respect to many adult health indicators. The DHS and MICS lack consumption and income measures, but one can construct an “asset index� using principal components analysis to rank households from poorest to richest (see Filmer and Pritchett 2001). The WHS contains information on both consumption and assets, but this section uses the asset information for consistency. 2.2 Inequalities in health The tables in this section show how health outcomes vary across asset (wealth) quintiles. The tables show the mean values of the indicator for each quintile, as well as for the sample as a whole. Also shown are the concentration indices which capture the direction and degree of inequality. A negative value indicates that the indicator takes a higher value among the poor, while a positive index indicates that the indicator takes a higher value among the better-off. The larger the index in absolute size, the more inequality there is. Table 2.1 shows that, according to the 2004 DHS, infant and under-five mortality, stunting, underweight, diarrhea, acute respiratory infection (ARI), and fever are worse among the poor. The 2006 MICS also finds stunting, underweight, diarrhea, ARI and fever more prevalent among poorer households. The 2003 WHS data show higher prevalence of malaria among the better-off, but this may be due to under- diagnosis of the health conditions of the poor. All results are statistically significant. Table 2.2 shows that according to the 2003 WHS, tuberculosis, angina, arthritis and difficulty with work and household activities are more common among the poor. In contrast, HIV positive prevalence and obesity among non-pregnant women is more concentrated among the better-off. Some other indicators are suggestive of a disproportionate burden across socioeconomic status, but these are not significant. 5 Table 2.1: Inequalities in child health Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 1 Infant mortality rate 13.0% 11.1% 11.0% 10.8% 7.5% 10.8% -0.078*** Under-five mortality 1 22.3% 19.2% 19.4% 16.8% 12.3% 18.2% rate -0.093*** 1 Stunting 57.4% 56.3% 55.2% 48.9% 38.1% 51.9% -0.067*** 2 Stunting 57.1% 56.1% 53.7% 51.4% 43.6% 52.7% -0.049*** 1 Underweight 25.5% 20.4% 19.3% 16.6% 8.5% 18.5% -0.157*** 2 Underweight 18.6% 15.8% 16.0% 14.5% 11.8% 15.5% -0.082*** 1 Diarrhea 27.0% 24.3% 22.5% 20.0% 18.4% 22.6% -0.072*** 2 Diarrhea 25.7% 25.2% 25.1% 23.5% 20.1% 24.1% -0.041*** Acute respiratory 1 20.1% 20.2% 23.5% 17.8% 11.9% 19.1% infection -0.066*** Acute respiratory 2 28.6% 26.2% 27.9% 27.2% 20.6% 26.3% infection -0.047*** 1 Fever 41.0% 41.8% 38.1% 35.7% 30.3% 37.8% -0.059*** 2 Fever 37.7% 35.5% 36.4% 36.4% 26.4% 34.7% -0.047*** 3 Malaria 62.5% 64.7% 61.8% 76.9% 78.1% 68.9% 0.054*** 1 2 Source: Authors’ estimates using ADePT and data from 2004 Malawi DHS , 2006 Malawi MICS and 2003 Malawi 3 WHS . Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 6 Table 2.2: Inequalities in adult health Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index Tuberculosis 2 14.7% 12.1% 13.6% 9.2% 7.2% 11.4% -0.129*** 8.4% 7.7% 12.3% 13.9% 17.1% 12.0% 0.153*** 1a HIV positive 7.4% 7.9% 9.7% 11.4% 15.5% 10.8% 0.157*** 1b HIV positive Obesity among non- 1a 0.8% 0.6% 1.9% 2.2% 6.3% 2.6% pregnant women 0.447*** Road traffic accident 2 2.2% 3.4% 3.4% 3.5% 4.0% 3.3% 0.081 Non-road traffic 2 9.2% 7.4% 11.4% 8.2% 7.9% 8.8% accident -0.016 Angina 2 17.3% 15.6% 13.3% 14.3% 11.7% 14.5% -0.076*** Arthritis 2 33.6% 34.4% 34.9% 32.1% 26.5% 32.3% -0.045*** Asthma 2 6.0% 5.3% 4.3% 4.8% 6.5% 5.4% 0.002 Depression 2 1.0% 0.9% 1.6% 1.4% 1.7% 1.3% 0.107 Diabetes 2 0.5% 0.0% 0.1% 0.3% 0.2% 0.2% -0.056 Difficulty with work and household 12.7% 12.6% 8.1% 7.3% 6.9% 9.5% 2 activities -0.141*** Poor self-assessed 2 5.0% 7.9% 5.1% 5.7% 5.0% 5.8% health status -0.041 1a 1b Source: Authors’ estimates using ADePT and data from 2004 Malawi DHS , 2010 Malawi DHS , and 2003 Malawi 2 WHS . Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. Table 2.3 shows inequalities in health behaviors that place individuals at risk for developing poor health. The prevalence of smoking among all respondents, smoking among women only, and insufficient physical activity are higher among the poor. The data on concurrent partnerships yield different conclusions. The 2004 DHS reports that the wealthy are more likely to have more than one sexual partner over the course of a year, but the 2006 MICS finds that poor women are more likely to have more than one sexual partner over the course of a year. Both findings are significant at the 1 per cent level. The DHS shows that wealthy women are significantly more likely to practice safe sex (i.e. have higher utilization of condoms) when in concurrent partnerships. Mosquito net use by children and pregnant women is also more common among the better-off. 7 Table 2.3: Inequalities in health behaviors Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 3 Smoking (all) 21.7% 18.4% 14.7% 12.1% 5.4% 14.6% -0.218*** 1 Smoking (women) 3.3% 2.1% 2.3% 1.3% 0.4% 1.8% -0.290*** Insufficient intake of 3 41.9% 40.4% 39.5% 37.4% 38.5% 39.6% fruit and vegetables -0.020 Insufficient physical 3 4.0% 2.6% 1.7% 2.0% 1.1% 2.3% activity -0.228*** 3 Drinking 3.2% 2.6% 3.4% 4.2% 4.2% 3.5% 0.083 Concurrent 1 8.1% 4.3% 5.3% 4.5% 10.2% 6.5% partnerships 0.087*** Concurrent 2 1.3% 1.5% 1.1% 0.5% 0.8% 1.0% partnerships -0.188*** Condom usage (more 1 15.2% 15.3% 23.0% 28.4% 43.5% 28.2% than one partner) 0.245*** Mosquito net use by 1 6.2% 9.1% 12.5% 17.7% 35.0% 15.2% children 0.333*** Mosquito net use by 2 18.9% 23.4% 28.4% 29.3% 47.2% 28.8% children 0.175*** Mosquito net use by 1 6.2% 10.2% 12.9% 17.3% 37.6% 15.4% pregnant women 0.312*** 1 2 Source: Authors’ estimates using ADePT and data from 2004 Malawi DHS , 2006 Malawi MICS and 2003 Malawi 3 WHS . Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. In sum, the tables in this section indicate that, in Malawi, the poor are disproportionately affected by ill health. All child health indicators, with the exception of malaria, are concentrated among the worse off. As previously stated, it is likely that diagnoses of malaria in the data are under-representing the true burden of disease. The majority of the significant indicators of poor adult health status are found to be pro-rich. The wealthy appear more likely to adopt positive health behaviors, such as the use of condoms during concurrent partnerships and mosquito net use by women and children. In contrast, the poor are more likely to exhibit risky health behaviors. 8 3 Inequalities in health care utilization In many countries, for a variety of possible reasons, health care utilization tends to be distributed very unequally across income groups, even after taking into account difference in medical needs. This section reports on inequalities in utilization of health care in Malawi for different types of care, and for different types of health care provider. 3.1 Data availability A Demographic and Health Survey (DHS) was fielded in Malawi in 2004, a Multiple Indicator Cluster Survey (MICS) in 2006, and a World Health Survey (WHS) in 2003. Although the DHS and MICS have rich information for many health outcomes, particularly in relation to child health, the WHS has fuller data availability with respect to many adult health indicators. The DHS and MICS lack consumption and income measures, but one can construct an “asset index� using principal components analysis to rank households from poorest to richest (see Filmer and Pritchett 2001). For variables drawn from the DHS, MICS and WHS, households are ranked by their score on the asset index. 3.2 Inequalities in health care utilization The tables in this section show how health care utilization varies across consumption or asset quintiles. The tables show the mean values of the indicator for each quintile, as well as for the sample as a whole. Also shown are the concentration indices which capture the direction and degree of inequality. A negative value indicates that utilization is higher among the poor, while a positive index indicates higher utilization rates among the better-off. The larger the index in absolute size, the more inequality in utilization there is. Table 3.1 shows coverage of key MCH interventions and treatment of childhood illness using data from the 2004 Malawi DHS and 2006 Malawi MICS. Approximately 65 per cent of children under the age of 5 are fully immunized, while 56 per cent of expectant women received at least 4 skilled antenatal care visits and 58 per cent had their babies delivered by a skilled birth attendant. Rates of all three interventions are higher among the better-off, and the treatment of diarrhea and ARI are pro-rich as well. The data also indicate that the use of a modern method of contraception is higher among wealthy women than among poor women. 9 Table 3.1: Inequalities in maternal and child health interventions Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 1 Full immunization 51.6% 58.5% 65.8% 74.1% 77.7% 64.5% 0.083*** Treatment of 1 55.6% 63.7% 59.1% 64.1% 65.9% 61.2% diarrhea 0.029** Treatment of 2 52.5% 51.1% 54.6% 59.5% 62.3% 55.4% diarrhea 0.044*** Medical treatment of 1 30.0% 34.5% 37.0% 41.6% 45.7% 36.7% ARI 0.081*** Medical treatment of 2 61.7% 58.0% 66.6% 58.9% 73.4% 63.0% ARI 0.026 Skilled antenatal care 1 51.1% 51.0% 52.5% 59.7% 69.7% 56.3% (4+ visits) 0.066*** Skilled birth 1 47.0% 47.7% 52.8% 63.4% 85.3% 57.9% attendance 0.123*** Contraceptive 1 44.6% 47.5% 46.1% 51.5% 53.9% 49.2% prevalence 0.043*** Contraceptive 2 31.1% 34.1% 37.0% 35.2% 35.3% 34.6% prevalence 0.025*** 1 2 Source: Authors’ estimates using ADePT and data from 2004 Malawi DHS and 2006 Malawi MICS . Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. Table 3.2 shows inequalities in preventive care among adults. It shows a high uptake of voluntary counseling and testing (VCT) for HIV among the population as a whole (91.9 per cent), but much lower rates for screening of tuberculosis (2.3 per cent). Breast cancer screening among women is low; less than 2 per cent of the sample has had breast cancer screening. All types of preventive care are more common among the better-off, except breast cancer screening, where data are suggestive of a pro-rich utilization but lack statistical significance. Table 3.2: Inequalities in adult preventive care Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 2 TB screening 1.7% 1.3% 3.5% 2.3% 3.0% 2.3% 0.114* Voluntary counseling 1 90.0% 89.9% 91.2% 91.1% 94.6% 91.9% and testing for HIV 0.011*** Breast cancer 2 0.8% 0.5% 4.4% 0.9% 3.0% 1.9% screening 0.257 1 2 Source: Authors’ estimates using ADePT and data from 2006 Malawi MICS and 2003 Malawi WHS . Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 10 Table 3.3: Inequalities in adult curative care Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index Inpatient or outpatient 49.2% 56.1% 57.8% 59.0% 60.6% 56.4% (12 months) 0.044*** Inpatient (12 months) 8.5% 8.8% 7.5% 10.9% 9.1% 8.9% 0.027 Inpatient (5 years) 19.2% 15.8% 18.4% 20.2% 21.3% 18.9% 0.038* Outpatient (12 months) 44.4% 50.7% 53.6% 53.3% 54.9% 51.2% 0.044*** Source: Authors’ estimates using ADePT and data from 2003 Malawi WHS. Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. Table 3.3 shows the inequalities in adult curative care in Malawi according to the 2003 WHS. Utilization of outpatient and inpatient care and any care is higher among the better-off and statistically significant, excluding inpatient care over the year preceding the survey. In sum, the tables in this section indicate that the worse off in Malawi utilize health interventions and health care at lower rates than the wealthy, putting them at greater risk for ill health. All MCH indicators from Table 3.1 have positive concentration indices, signifying that the utilization of MCH interventions and treatment of illness is higher among the better-off. All significant categories of adult preventive care and adult curative care have higher utilization among the better-off, too. 11 4 Benefit incidence of government spending Policymakers typically take the view that government health expenditure (GHE) ought not to disproportionately benefit the better-off, and if anything ought to favor the poor more than the better- off. Benefit-incidence analysis (BIA) shows whether and how far GHE disproportionately benefits the poor. This section reports BIA results for Malawi, using three different methods for allocating GHE to households, namely the constant unit cost assumption, the constant unit subsidy assumption, and the proportional unit cost assumption. The first is arguably the least plausible of the three, since it implies that higher fees do not translate into more costly care. But it does have the attraction of not needing to be modified if part of (general) GHE goes on demand-side subsidies through, for example, a subsidized health insurance program. Where the results presented below are obtained using the constant-unit- subsidy and proportional-unit-cost assumptions, it is assumed implicitly that supply- and demand-side subsidies have the same distributional impact. 4.1 Data availability The World Health Survey (WHS) that was conducted in Malawi in 2003 records the utilization of inpatient and outpatient care. It allows us to determine whether the individual had at least one inpatient stay and at least one outpatient visit during the year preceding the survey 2. The WHS clearly distinguishes between public and private outpatient care, documents the name of the facility visited, and records the fees paid by the individual during the last inpatient stay or outpatient visit. Household ranking for the WHS in this section is based on consumption. A BIA also needs data on GHE (i.e. subsidies) by level of service. Malawi conducted many National Health Accounts (NHA) exercises during the 1998-2006 period. The 2002-2003 results are used in this section in order to be consistent with the WHS 2003 3 . Government spending on public health centers, dispensaries and maternities (item HP.3.4.9.1 in the NHA report) is measured as the contributions made by the Ministry of Health (HF.1.1.1.1) and local authorities (HF.1.1.1.4) to this provider of ambulatory care. Government spending on public hospitals is computed as the sum of the contributions made by the Ministry of Health to central and district hospitals (items HP1.1.1.1 and HP1.1.1.2). Furthermore we distinguish between inpatient (HC.1.1) and outpatient (HC.1.3) services by using the NHA matrix that puts health providers in relation to the health function they provide. Finally we also take into account that the government allocates subsidies to the Christian Health Association of Malawi (CHAM) (HF2.4.1) which in turn finances various public health providers. 2 Ideally, one would like to observe the actual number of days spent at the hospital and the number of outpatient visits. However, this limitation is offset by the fact that more frequent users are also more likely to have used care during the WHS one-year recall period, thus reducing this potential bias. This approach was also validated by performing a BIA analysis using survey data (from Vietnam) that contained both a binary indicator of utilization and the actual number of inpatient days and doctor visits, and finding that there were not considerable differences between the corresponding BIA results. 3 See http://www.who.int/nha/country/mwi/en/. 12 4.2 Inequalities in benefit incidence The tables in this section show the distribution across consumption quintiles of the utilization of government facilities, the fees paid to these facilities and the estimated subsidies to the health sector. The latter depends on the assumptions made to allocate subsidies to households; results are presented for three sets of assumptions. The tables show the shares of fees or shares of subsidies that go to each quintile. Also shown are the concentration indices which capture the direction and degree of inequality. A negative value indicates that the variable in question is higher among the poor, while a positive index indicates higher values among the better-off. The larger the index in absolute size, the more inequality in the indicator there is. Table 4.1 shows the distribution of utilization of health and maternity centers and dispensaries, and of outpatient and inpatient services in public hospitals. It can be seen that the overall utilization of health and maternity centers and dispensaries is comparable to that of outpatient care in public hospitals (10.9 per cent and 9.9 per cent on average for the whole population). This is about 40 per cent greater than the utilization of inpatient care with 5.9 per cent of the individuals with at least one inpatient stay during the year preceding the survey. As for the distribution of utilization, there is no real evidence of socioeconomic inequality in any of the health services and the concentration indices are not statistically significant. Table 4.1: Inequalities in use of publicly financed facilities Outpatient clinic, health center, dispensary or Outpatient hospital Inpatient hospital maternity center Lowest quintile 9.8% 9.6% 4.6% 2 12.9% 8.2% 6.0% 3 10.4% 10.6% 6.5% 4 10.7% 9.5% 6.3% Highest quintile 11.0% 11.7% 6.0% Total 10.9% 9.9% 5.9% Concentration index 0.003 0.039 0.043 Source: Authors’ estimates using ADePT and data from 2003 Malawi WHS. Note: The utilization data refer to the last year in all cases. Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. The following table (Table 4.2) describes inequalities in health fees for the above utilization services. Inpatient care consumes exactly two thirds (MMWK 59) of all the fees paid by the households for public health services. As for health and maternity centers and dispensaries, this represents about a quarter of total household health outlays on public health services. Finally, at 7.3 per cent, outpatient care in public hospitals has the smallest share of household out-of-pocket expenditure. Table 4.2 also shows that the fees paid increase with income for all types of public health services. This is especially true in the case of inpatient fees in public hospitals, which are markedly more concentrated among the rich. Indeed, the richest quintile alone contributes 94.4 per cent of total fees. The corresponding concentration index (0.788) indicates a strong pro-rich distribution that is strongly statistically significant. 13 Table 4.2: Distribution in fees paid Outpatient clinic, health center, dispensary or Outpatient hospital Inpatient hospital maternity center Lowest quintile 14.9 21.8 0.4 2 18.3 7.8 0.6 3 22.5 21.0 3.8 4 17.8 15.3 0.8 Highest quintile 26.4 34.1 94.4 Total 100.0 100.0 100.0 Concentration index 0.107** 0.167 0.788*** Source: Authors’ estimates using ADePT and data from 2003 Malawi WHS. Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. The following table (Table 4.3) shows the incidence of government spending on health. The first two lines of the table show how aggregate government spending on health varies across the different types of service. The table contains three sets of estimates of the distribution of subsidies across consumption quintiles. The first set is based on the constant unit-cost assumption, i.e. each hospital outpatient visit (for example) is assumed to cost the same, an amount equal to total costs incurred in delivering this type of service (i.e. subsidies plus user fees) divided by the number of units of utilization. This approach can lead to negative imputed subsidies since the amount someone pays in fees could exceed the unit cost. In Table 4.3, as in much of the literature, negative imputed subsidies have been set to zero. The second set of results are based on the assumption that the unit subsidy is constant, equal to total subsidies for the service in question divided by the number of units of utilization of that service. The third set of results assumes that higher fees for a particular type of care indicate a more costly type of care received, i.e. it is assumed that unit costs and fees are proportional to one another. The first two lines of Table 4.3 indicate that more than half (56.6 per cent) of the government subsidies are spent on inpatient care in public hospitals, 11.5 per cent is spent on outpatient care in these hospitals, and 32 per cent on health and maternity centers and dispensaries. The first two sets of results (based on the constant unit cost and subsidy assumptions) are very similar. The concentration indexes indicate that all health services are slightly pro-rich but the statistical evidence is not strong enough to confirm it. On the other hand, the pro-rich bias of all health services dramatically increase when unit costs are assumed to be proportional to the amount spent out-of-pocket. As a result, the subsidies for all types of services but outpatient care in hospitals become strongly statistically significant. With a concentration index of 0.316, the overall effect of government spending is clearly pro-rich and statistically significantly so. 14 The relative difference between the results obtained with the different BIA assumptions stem from the sharp contrast between the distributions of utilization of health services and of fees paid. Indeed, the former is barely related to income whereas the latter is sufficiently pro-rich to drive the results. Taken together, these benefit incidence analyses show that government spending on health is at best independent from income but may very well favor the rich if higher fees should be assumed to reflect higher subsidies. Table 4.3: Inequality in the incidence of government health spending (shares) Outpatient clinic, health center, Outpatient Inpatient hospital Total subsidies dispensary or hospital maternity center Total subsidies (in million 1,026.1 367.7 1,814.0 3,207.8 Malawian Kwacha) Share of total subsidy 32.0% 11.5% 56.6% 100% Constant unit cost assumption Lowest quintile 17.9 19.3 15.9 16.9 2 23.6 16.6 20.6 21.1 3 18.9 21.3 22.4 21.2 4 19.5 19.2 21.8 20.8 Highest quintile 20.1 23.6 19.2 20.0 Total 100.0 100.0 100.0 100.0 Concentration index 0.003 0.038 0.040 0.028 Constant unit subsidy assumption Lowest quintile 17.9 19.3 15.7 16.8 2 23.6 16.5 20.4 20.9 3 18.9 21.3 22.1 21.0 4 19.5 19.2 21.5 20.6 Highest quintile 20.1 23.7 20.3 20.6 Total 100.0 100.0 100.0 100.0 Concentration index 0.003 0.039 0.043 0.030 Proportional cost assumption Lowest quintile 14.9 21.8 0.4 7.5 2 18.3 7.8 0.6 7.1 3 22.5 21.0 3.8 11.8 4 17.8 15.3 0.8 7.9 Highest quintile 26.4 34.1 94.4 65.7 Total 100.0 100.0 100.0 100.0 Concentration index 0.107** 0.167 0.788*** 0.316*** Source: Authors’ calculations using ADePT, 2003 Malawi WHS. Note: With the constant cost assumption imposed, grossed-up survey data for fees have been used rather than NHA data on fees, and negative imputed subsidies have been set to zero. Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 15 5 Financial protection in health Countries finance their health care through a mix of out-of-pocket payments, private and social insurance, general revenues, and international development assistance. All except the latter ultimately come from the pockets of households in the country. Therefore, health systems are not just about improving health but also about ensuring that people are protected from the financial consequences of illness and death, or at least from the financial consequences of having to obtain medical care. This section presents data on two alternative measures of financial protection: one that asks whether out-of- pocket spending is ‘catastrophic’ and the other that asks if it is ‘impoverishing’. Neither captures the income losses associated with illness, and both therefore underestimate the full financial impact of ill health on households. The section also explains the institutional arrangements used in Malawi to provide financial protection in the health sector, and presents data on levels of inequalities in coverage. 5.1 Data availability A World Health Survey (WHS) was fielded in Malawi in 2003. The WHS has information on health expenditure and household consumption. Malawi also fielded an Integrated Household Survey (IHS) in 1997-98 and 2004. The IHS is a comprehensive socio-economic survey of the living standards of households, which also provides information on health expenditure. In order to facilitate international comparisons, the interpretations below focus on data from the WHS. Households are ranked by per capita consumption. 5.2 Catastrophic out-of-pocket payments This subsection provides information on ‘catastrophic’ health payments. Catastrophic payments are defined as health care payments in excess of a predetermined percentage of their total household or nonfood spending. The columns of Table 5.1 give different thresholds above which health payment “budget shares� might be deemed catastrophic. The first line of the table displays the catastrophic payment “headcount�, i.e. the proportion of households with a health payment budget share greater than the given threshold. The second line relates the catastrophic payment headcount to the household consumption distribution, and shows the concentration index of the incidence of catastrophic payments. A positive value of the concentration index indicates a greater tendency for the better-off to have out-of-pocket spending in excess of the payment threshold, whereas a negative value indicates that the worse off are more likely to have out-of-pocket spending exceeding the threshold. The information in Table 5.1 on catastrophic payments is for the 2003 WHS and the 2004 IHS. The table shows that, for the WHS data, when the threshold is raised from 5 to 40 per cent of total household expenditure the estimate of the incidence of catastrophic payments falls from 19.6 to 2.6 per cent. However, using nonfood expenditure, the estimate of the incidence of catastrophic payments falls from 45.2 to 20.8 per cent. Table 5.1 also shows that the concentration index for catastrophic spending. The 2003 Malawi WHS finds unambiguously that catastrophic payments are found to be concentrated among the poor using both total and nonfood measure of consumption, but the 2004 Malawi Integrated Household Survey 16 rich finds catastrophic payments concentrated among the poor for poor when the nonfood measure is used, but among the rich when the total consumption measure is used. Table 5.1: Incidence of catastrophic out-of-pocket spending Threshold share of total household consumption 5% 10% 15% 25% 40% 2003, WHS Headcount 19.6% 11.5% 8.2% 4.5% 2.6% Concentration index 0.031 -0.037 -0.076* -0.222*** -0.391*** 2004, IHS Headcount 17.3% 8.9% 5.6% 2.7% 1.0% Concentration index 0.022*** 0.048*** 0.076*** 0.131*** 0.371*** Threshold share of nonfood consumption 5% 10% 15% 25% 40% 2003, WHS Headcount 45.2% 39.0% 34.3% 27.5% 20.8% Concentration index 0.023** 0.063*** -0.087*** 0.142*** 0.228*** 2004, IHS Headcount 44.5% 30.4% 23.2% 15.0% 8.4% Concentration index -0.042*** -0.023*** -0.009* -0.010 0.017* Source: Authors’ estimates using ADePT, 2003 WHS and 2004 IHS. Note: * CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 5.3 Impoverishing out-of-pocket payments This subsection presents poverty measures corresponding to household consumption gross and net of out-of-pocket health spending. A comparison of the two shows the scale of impoverishment due to health payments. The idea is that a health problem necessitating out-of-pocket medical spending may be serious enough to push a household from being above the poverty line ‘before’ the health problem to being below the poverty line ‘after’ the health problem. Adding out-of-pocket spending to the household’s nonmedical consumption (‘consumption including – or gross of – health payments’) gives us a sense of what its standard of living would have been without the health problem. Its nonmedical spending (‘consumption excluding health payments’) gives us a sense of what its standard of living looks like with the health problem. The assumption here is that out-of-pocket spending is involuntary and caused by health “shocks�; health spending is assumed to be financed by reducing current consumption. The first line of the Table 5.2 shows the poverty “headcount� which represents the proportion of individuals living below the poverty line. Two poverty lines are used: the lower line corresponds to US$1.25 a day at purchasing power parities (PPP); the upper line corresponds to US$2 a day. The poverty gap gives the total shortfall from the poverty line, averaged across the entire population; it is expressed in dollars a day. The mean positive poverty gap is a measure of the intensity of poverty: it 17 indicates the average shortfall from the poverty line among those in poverty; it is also measured in dollars a day. Table 5.2: Impoverishment through out-of-pocket health spending Consumption Consumption Percentage including excluding Change change OOP OOP Poverty line at US$1.25 per capita per day 2003, WHS Percentage in poverty / Poverty headcount 91.7% 92.1% 0.4 pp 0.4% Average shortfall from the poverty line $0.81 $0.82 $0.01 1.7% Average shortfall from the poverty line, $0.88 $0.89 $0.01 1.2% among the poor 2004, IHS Percentage in poverty / Poverty headcount 69.6% 71.5% 1.9 pp 2.7% Average shortfall from the poverty line $0.38 $0.40 $0.02 5.0% Average shortfall from the poverty line, $0.54 $0.55 $0.01 2.2% among the poor Poverty line at US$2.00 per capita per day 2003, WHS Percentage in poverty / Poverty headcount 96.0% 96.1% 0.1 pp 0.1% Average shortfall from the poverty line $1.52 $1.54 $0.02 1.0% Average shortfall from the poverty line, $1.58 $1.59 $0.01 0.9% among the poor 2004, IHS Percentage in poverty / Poverty headcount 87.3% 88.0% 0.7 pp 0.8% Average shortfall from the poverty line $0.98 $1.01 $0.03 2.9% Average shortfall from the poverty line, $1.12 $1.14 $0.02 2.1% among the poor Source: Authors’ estimates using ADePT, 2003 WHS and 2004 IHS. Note: Poverty line is at 2005 purchasing power parities, adjusted to current prices using Malawi’s CPI. Figures are for a 4-week figure and are in Malawian local currency units (LCU). Table 5.2 reports results for the 2003 Malawi WHS and the 2004 Malawi IHS. It can be seen that, according to the 2003 WHS. When out-of-pocket payments are counted as part of a household’s consumption, 91.7% per cent of the population in 2003 (according to the WHS) was poor using a US$1.25 a day poverty line. If we take out-of-pocket payments out from the household’s consumption, recognizing that this expenditure is involuntary and simply enables a household to cope with a health problem, the poverty rate goes up to 92.1 per cent; this is the true poverty rate. Thus, about 0.4 per cent of the population would not have been poor if the resources they were forced to devote to health care had been available to spend on other things. Out-of-pocket spending on health raises the per- capita poverty gap rises by $0.01, equivalent to or a 1.7 per cent increase. The mean positive poverty gap also increases by $0.01, a 1.2 per cent increase. When using a poverty line of $2.00 a day, the 18 increase in the percentage of those impoverished is similar, but the percentage increase in depth of poverty is smaller. The latest data available, from the 2004 MIHS, show that out-of-pocket payments are responsible for an increase in the poverty rate of 2.7 per cent when using the US$1.25 a day measure and of 0.8 per cent when using the US$2.00 a day measure. The poverty gap increased as a result of out-of-pocket payments by $0.02 per capita at the US$1.25 a day poverty line (a 5 per cent increase), and by $0.03 per capita at the US$2.00 a day poverty line (a 2.9 per cent increase). Figure 5.1 shows the effect of out-of-pocket payments on poverty via a “Pen’s parade�. Households are lined up in ascending order of their consumption including out-of-pocket payments. The vertical “paint drips� show the extent to which out-of-pocket payments divert a household’s spending away from items such as food, education, clothing, etc. insofar as health care is used in response to an adverse health event, health spending doesn’t add to the household’s living standards in a way that food spending does. The length of the paint drip, therefore, shows how far health spending compromises a household’s living standards. In this case, we can see that when using a poverty line of US$1.25 a day, the majority of households are already below the poverty line regardless of out-of-pocket spending. In this case, the effects of out-of-pocket health expenditures on the extremely destitute are small, but they grow as the population increases in wealth and some of those approaching the poverty line are brought back down into extreme poverty. The chart also shows that some already-impoverished households are experiencing a deepening of poverty as a result of their health spending. 19 Figure 5.1: The impoverishing effect of out-of-pocket spending 2.5 pre-OOP consumption post-OOP consumption 2 Consumption as multiple of PL 1.5 1 .5 0 0 .2 .4 .6 .8 1 Cumulative proportion of population, ranked from poorest to richest Source: Authors’ estimates using ADePT and 2003 Malawi WHS. Note: Poverty line is US$1.25 a day at 2005 purchasing power parities, adjusted to current prices using Malawi’s CPI. In sum, this section does not find very high levels of catastrophic expenditure; whether looking at data from the 2003 Malawi WHS or the 2004 Malawi IHS, catastrophic payments are relatively modest at most thresholds. When comparing health payments to total household consumption catastrophic payments are found to be concentrated among the poor at higher thresholds using WHS data, and concentrated among the wealthy when using the IHS data. When the nonfood consumption measure is used, catastrophic payments are found to be concentrated among the poor, regardless of data sources used. The data also indicate that health spending increases the absolute number of the impoverished, slightly less for the WHS than for the IHS, but overall by only a small amount in both cases. Indeed, the increase in the poverty rate due to health spending is only 0.1 per cent when using the US$2.00 a day measure and 0.4 per cent when using the US$1.25 a day measure (according to WHS). 20 6 References Filmer, D. and L. Pritchett (2001). “Estimating wealth effects without expenditure data or tears: An application to educational enrollments in states of India.� Demography 38(1): 115- 132. Gwatkin, D., S. Rutstein, K. Johnson, E. Suliman, A. Wagstaff and A. Amouzou. (2007). Socioeconomic differences in Health, Nutrition and Population. Washington, DC: World Bank. O’Donnell, O., E. van Doorslaer, A. Wagstaff and M. Lindelow. (2008). Analyzing health equity using household survey data: a guide to techniques and their implementation. Washington, D.C: World Bank. Malawi, Government of. (2011). Health Sector Strategic Plan 2011-2016. Lilongwe, Malawi. Ministry of Health and Government of Malawi. (2007). Malawi National Health Accounts 2002- 2004 with sub-accounts for HIV and AIDS, Reproductive and Child Health. Bethesda, MD: Partners for Health Reformplus Project, Abt Associates Inc. Wagstaff, A. (2012). “Benefit-incidence analysis: are government health expenditures more pro- rich than we think?� Health Economics 21(4): 351-66. Wagstaff, A., M. Bilger, Z. Sajaia and M. Lokshin. (2011). Health equity and financial protection: streamlined analysis with ADePT software. Washington, D.C.: World Bank. World Health Organization. (2009). “National Health Accounts database.� Retrieved August 1, 2011. 21 7 Annexes 7.1 Measurement of indicators INDICATOR MEASUREMENT DATA CHILD HEALTH Infant mortality rate Number of deaths among children under 12 months of age per 1,000 live births (Note: mortality DHS rate calculated using the true cohort life table approach; the DHS reports use of the synthetic cohort life table approach) Under-five mortality rate Number of deaths among children under 5 years of age per 1,000 live births (Note: mortality rate DHS calculated using the true cohort life table approach; the DHS reports use of the synthetic cohort life table approach) Stunting % of children with a height-for-age z-score <-2 standard deviations from the reference median DHS, MICS (Note: z-score calculated using WHO 2006 Child Growth Standards) Underweight % of children with a weight-for-age z-score <-2 standard deviations from the reference median DHS, MICS (Note: z-score calculated using WHO 2006 Child Growth Standards) Diarrhea % of children with diarrhea (past two weeks) DHS, MICS Diarrhea % of children with diarrhea (past two weeks; youngest child) WHS Acute respiratory % of children with an episode of coughing and rapid breathing (past two weeks) DHS, MICS infection Acute respiratory % of children with an episode of coughing and rapid breathing (past two weeks; youngest child) WHS infection Fever % of children with fever (past two weeks) DHS, MICS Fever % of children with fever (past two weeks; youngest child) WHS Malaria % of children with an episode of malaria (past year; youngest child) WHS ADULT HEALTH Tuberculosis % of adults who reported tuberculosis symptoms (past year) WHS HIV positive % of adults aged 15 to 49 whose blood tests are positive for HIV 1 or HIV 2. DHS Obesity among non- % of women aged 15 to 49 with a BMI above 30 DHS pregnant women Obesity among all % of women aged 18 to 49 with a BMI above 30 WHS women Road traffic accident % of adults involved in a road traffic accident with bodily injury (past year) WHS Non-road traffic accident % of adults who suffered bodily injury that limited everyday activities, due to a fall, burn, WHS poisoning, submersion in water, or by an act of violence (past year) Angina % of adults ever diagnosed with angina or angina pectoris WHS Arthritis % of adults ever diagnosed with arthritis WHS Asthma % of adults ever diagnosed with asthma WHS Depression % of adults ever diagnosed with depression WHS Diabetes % of adults ever diagnosed with diabetes WHS Difficulty with work and % of adults who have severe or extreme difficulties with work or household activities (past 30 WHS household activities days) (Note: This indicator was created from an ordinal variable with five categories) 22 Poor self-assessed health % of adults who rate own health as bad or very bad (Note: This indicator was created from an WHS status ordinal variable with five categories) RISK FACTORS Smoking (all) % of adults who smoke any tobacco products such as cigarettes, cigars or pipes WHS Smoking (women) % of women aged 15 to 49 who smoke cigarettes, pipe or other tobacco DHS Smoking (women) % of women aged 18 to 49 who smoke cigarettes, pipe or other tobacco WHS Insufficient intake of % of adults who have insufficient intake of fruit/vegetables (less than 5 servings) WHS fruit and vegetables Insufficient physical % of adults who spend < 150 minutes on walking/ moderate activity/vigorous activity (past week) WHS activity Drinking % of adults who consume ≥5 standard drinks on at least one day (past week) WHS Concurrent partnerships % of women aged 15 to 49 who had sexual intercourse with more than one partner (past year) DHS, MICS Concurrent partnerships % of women aged 18 to 49 who had sexual intercourse with more than one partner (past year) WHS Condom usage (more % of women aged 15 to 49 who had more than one partner in the past year and used a condom DHS, MICS than one partner) during last sexual intercourse Condom usage (more % of women aged 18 to 49 who had more than one partner in the past year and used a condom WHS than one partner) during last sexual intercourse Mosquito net use by % of children who slept under an (ever) insecticide treated bed net (ITN) (past night) DHS, MICS children Mosquito net use by % of pregnant women aged 15 to 49 who slept under an (ever) insecticide treated bed net (ITN) DHS pregnant women (past night) MATERNAL AND CHILD HEALTH INTERVENTIONS Full immunization % of children aged 12-23 months who received BCG, measles, and three doses of polio and DPT, DHS, MICS either verified by card or by recall of respondent Treatment of diarrhea % of children with diarrhea given oral rehydration salts (ORS) or home-made solution DHS, MICS Medical treatment of ARI % of children with a cough and rapid breathing who sought medical treatment for acute DHS, MICS respiratory infection (past 2 weeks) Skilled antenatal care (4+ % of mothers aged 15 to 49 who received at least 4 antenatal care visits from any skilled DHS visits) personnel (Note: type of skilled personnel varies by country including doctor, nurse, midwife, auxiliary midwife, feldsher, clinical officer, health surveillance attendant, medical assistant) Skilled birth attendance % of mothers aged 15 to 49 that were attended by any skilled personnel at child’s birth (Note: DHS type of skilled personnel varies by country including doctor, nurse, midwife, auxiliary midwife, feldsher, clinical officer, health surveillance attendant, medical assistant) Contraceptive % of women aged 15 to 49 who currently use a modern method of contraception DHS, MICS prevalence ADULT PREVENTIVE CARE TB screening % of adults who were tested for tuberculosis (past year) WHS Voluntary Counseling % of women aged 18 to 49 who were tested for HIV and were told the results of the test WHS, MICS and Testing for HIV Cervical cancer screening % of women aged 18 to 69 who received a pap smear during last pelvic examination (past 3 WHS years) Breast cancer screening % of women aged 40 to 69 who received a mammogram (past 3 years) WHS ADULT CURATIVE CARE Inpatient or outpatient % of adults who used any inpatient or outpatient health care (past year) WHS (12 months) Inpatient (12 months) % of adults who used any inpatient health care (past year) WHS 23 Inpatient (5 years) % of adults who used any inpatient health care (past 5 years) WHS Outpatient (12 months) % of adults who used any outpatient health care (past year; conditional on having not used any WHS inpatient care past 5 years) Unless otherwise noted, all children are under the age of 5 and all adults are aged 18 and older. 24 7.2 Methodological notes Section 2 and 3: Inequalities in health and health care utilization The selection and measurement of health outcome indicators used in sections 2 and 3 on inequalities in health and health care utilization was based on (i) a comparison of indicators used in major health publications and databases, (ii) the advice of World Bank Health Specialists on recommended monitoring and measurement practice in their respective fields, and (iii) how measurable those indicators would be in the available data sources. The following major reports/databases were consulted as a guide to indicator measurement: World Bank Development Indicators, the World Bank’s HNPStats database, WHO’s World Health Survey country reports, and the World Bank’s report series on “Socio-economic differences in health, nutrition and population (Gwatkin et al. 2007). The data sources for this section include the Demographic and Health Surveys (DHS), World Health Surveys (WHS), Multiple Indicator Cluster Surveys (MICS) and multipurpose household surveys (such as the World Bank Living Standard and Measurement Surveys). Where the selected indicators are available in more than one of these surveys, all measures are reported. In all analyses of inequality in this section, i.e. quintile analysis and calculation of concentration indices, households are ranked by an asset index computed using principal components analysis. In order to avoid presenting estimates biased by insufficient power, indicators were removed from the tables if the sample size in any quintile was less than the following thresholds: 250 per quintile for infant and child mortality estimates and 25 per quintile for all other indicators. This follows the practice of Gwatkin et al. (2007). In addition, the statistical significance of all concentration indices is reported. Section 4: Benefit-incidence analysis The section on benefit incidence analysis uses three different methods for allocating government health expenditure to households, invoking three different assumptions that are described in detail in Wagstaff (2011). The first, the constant unit cost assumption, treats the sum of individual fees and government subsidies as constant, and thus any fees paid when using public services results in a reduction in the government subsidy received. The second, the constant unit subsidy assumption, allocates the same subsidy to each unit of service used, irrespective of the fees paid. Finally, the third, the proportional unit cost assumption, makes the cost of care proportional to the fees paid, which implies that the government subsidy received increases as the fees paid increases. In calculating the distribution of fees, service utilization and government subsidies, households are ranked by per capita consumption. The quintile distributions and concentration indices are reported, including measures of statistical significance. The data sources for this section include the WHS and multipurpose household surveys that are used to obtain information on service utilization at difference levels of care and fees paid by patients. Data on government subsidies at each level of service are obtained from National Health Accounts reports, specifically from one or more of the following tables depending on the level of detail provided: financing 25 source by financing agent, financing agent by provider, and provider by function, other detailed country expenditure reviews or directly from budget offices. The limitations of the analysis depend on the data source. One limitation of using the WHS is that we only observe whether or not the individual had an inpatient and outpatient visit, but not the actual number of visits or length of stay. We also observe outpatient visits only for people who did not use inpatient care. The implications of these limitations are being investigated. Section 5: Financial protection Section 5 examines health insurance coverage, catastrophic health care payments and impoverishment due to out-of-pocket expenditures. In this section, households are ranked by consumption. The analysis of catastrophic health care payments follows the popular approach elaborated upon O’Donnell et al. (2008) which defines health spending as “catastrophic� if it exceeds some fraction or threshold of total expenditure, or of total nonfood expenditure, in a given period. As O’Donnell et al. (2008) note, the threshold of 10% for total expenditure and 40% for nonfood expenditure are commonly used in the literature. In addition to measures of incidence, distribution-sensitive measures of catastrophic payments are calculated, specifically the concentration index, and statistical significance is reported. The analysis of impoverishing expenditure uses the poverty lines of US$1.25 and US$2.00 per capita per day at 2005 purchasing power parity (PPP) (with PPP values obtained from the World Development Indicators database) and, in some cases, national poverty lines. Data sources for the analysis of financial protection include the WHS, as well as multipurpose household surveys. Survey data on health insurance coverage is difficult to obtain for most countries. 26