71253 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 – Ghana. 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 Xiao Ye (AFRCE), Suarabh Shome (consultant) and Ellen Van de Poel (consultant) for advice on data, Karima Saleh (AFTHE) for comments on an earlier draft, 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 21, 2012 ii List of Abbreviations and Acronyms ARI Acute respiratory infection BIA Benefit-incidence analysis CAG Controller Accountant General CPI Consumer price index DACF District Assembly Common Fund DHS Demographic and Health Survey GDP Gross domestic product GHC Old Ghanaian Cedi GHE Government health expenditures GHS Ghana Health Services GLSS Ghana Living Standards Survey GNP Gross national product IGF Internally Generated Fund MCH Maternal and child health NHA National Health Accounts NHIL National Health Insurance Levy NHIS National Health Insurance Scheme MoH Ministry of Health MICS Multiple Indicator Cluster Survey PPP Purchasing power parity SHI Social health insurance VAT Value added tax VCT Voluntary counseling and testing WDI World Development Indicators WHO World Health Organization WHS World Health Survey iii HEALTH EQUITY AND FINANCIAL PROTECTION IN GHANA Contents Executive Summary...................................................................................................................................... vi 1 Ghana’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............................................................................................................................. 6 2.1 Data availability............................................................................................................................. 6 2.2 Inequalities in health..................................................................................................................... 6 3 Inequalities in health care utilization .................................................................................................. 10 3.1 Data availability........................................................................................................................... 10 3.2 Inequalities in health care utilization .......................................................................................... 10 4 Benefit incidence of government spending ........................................................................................ 13 4.1 Data availability........................................................................................................................... 13 4.2 Inequalities in benefit incidence ................................................................................................. 14 5 Financial protection in health ............................................................................................................. 18 5.1 Data availability........................................................................................................................... 18 5.2 Catastrophic out-of-pocket payments ........................................................................................ 19 5.3 Impoverishing out-of-pocket payments ..................................................................................... 20 6 Progressivity of health finance............................................................................................................ 23 6.1 Data availability........................................................................................................................... 23 6.2 Progressivity of health care financing ......................................................................................... 23 7 References .......................................................................................................................................... 25 8 Annexes ............................................................................................................................................... 27 8.1 Additional graphs and tables ...................................................................................................... 27 8.2 Measurement of indicators ........................................................................................................ 28 8.3 Methodological notes ................................................................................................................. 30 iv Figures Figure 1.1: Health care financing mix, 2005 – 2009...................................................................................... 3 Figure 5.1: The impoverishing effect of out-of-pocket spending ............................................................... 22 Tables Table 1.1: Health expenditure data, 2009 .................................................................................................... 2 Table 2.1 : Inequalities in child health .......................................................................................................... 7 Table 2.2: Inequalities in adult health .......................................................................................................... 8 Table 2.3: Inequalities in health behaviors ................................................................................................... 9 Table 3.1: Inequalities in maternal and child health interventions ............................................................ 11 Table 3.2: Inequalities in adult preventive care.......................................................................................... 11 Table 3.3: Inequalities in adult curative care .............................................................................................. 12 Table 4.1: Inequalities in use of public and private facilities ...................................................................... 15 Table 4.2: Distribution in fees paid ............................................................................................................. 15 Table 4.3: Inequality in the incidence of government health spending (shares) ....................................... 17 Table 5.1: Incidence of catastrophic out-of-pocket spending .................................................................... 20 Table 5.2: Impoverishment through out-of-pocket health spending ......................................................... 21 Table 6.1: Progressivity of health finance ................................................................................................... 24 v Executive Summary This report analyses equity and financial protection in the health sector of Ghana. In particular, it examines 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 2008 Ghana Demographic and Health Survey, the 2005/06 Ghana Living Standards and Measurement Survey V, the 2003 Ghana World Health Survey and the 2006 Ghana Multiple Indicator Cluster Survey. All analyses are conducted using original data and performed using 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 Ghana. This includes most selected indicators of child health, such as under-five mortality rate, stunting, underweight, diarrhea, acute respiratory infection (ARI) and fever. An exception is malaria, which is more concentrated among the better-off, but this could be because the poor lack access to diagnostic care. The results for measures of adult health are divided down the middle, with half of all statistically significant indicators more prevalent among the wealthy. Tuberculosis, non-road traffic accidents, diabetes and poor self-assessed health status are more concentrated among the poor, while road traffic accidents, diabetes and obesity (among non-pregnant women) are more prevalent among the wealthy. With respect to risky health behaviors, the results suggest that the wealthy exhibit more unhealthy behaviors than the poor. It is the wealthy who are more likely to have insufficient intake of fruits and vegetables and are also more likely to have concurrent partnerships (and thus be at risk for sexually transmitted infections), although they are also more likely to use condoms while having multiple partners. The poor are more likely to smoke. The use of insecticide-treated nets to protect against malaria is also more concentrated among the poor. Do the poor use health services less than the rich? Yes. Health care utilization in Ghana is concentrated among the better-off, suggesting that the less well- off are at greater risk for ill health. All selected MCH interventions, including the take-up of immunization, treatment of diarrhea, medical treatment of ARI, antenatal care take-up, skilled birth attendance and contraceptive prevalence, are more concentrated among the better-off. Among adult preventive services, most results are statistically insignificant, but breast cancer screening is concentrated among the better-off. The use of general health services, both outpatient and inpatient, is concentrated among the wealthy. Is the distribution of government spending on health pro-rich or pro-poor? Mainly, yes. Government spending on outpatient hospital care is found to be pro-rich, regardless of the assumptions made in the benefit-incidence analysis. When it comes to hospital inpatient care government spending is pro-rich using two of three methodological assumptions. Taken together, total subsidies for health are found to be pro-rich, except for when looking at utilization of health centers and vi health posts – but these account for only 16.6 per cent of government spending. Generally, government expenditure on health favors the better-off. What is the effect of out-of-pocket payments on household financial well-being? Substantial. Almost a quarter (23.6 per cent) of households spent more than 10 per cent of total household consumption on out-of-pocket health payments and almost a tenth (8.8 per cent) spend more than 25 per cent. Using the alternative nonfood measure, more than half (50.3 per cent) of households spend more than 10 per cent of nonfood consumption on out-of-pocket payments and more than a fifth (22.8 per cent) spend more than 40 per cent. Health spending is also responsible for an increase in the poverty rate equivalent to 5.5 per cent, when using the US$2.00 a day measure, and 7.6 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 8.6 per cent at the higher poverty line. Is health financing progressive or regressive? Neither. Taxes are found to be slightly progressive and social health insurance contributions mildly progressive. As for voluntary health insurance premiums and out-of-pocket payments, they appear to be progressive and slightly regressive respectively, but the results are not statistically significant. Overall, health care financing in Ghana emerges as being mostly proportional to income. It is important to note that the GLSS data used for this analysis pre-date the recent health insurance reform and it is not known what effect this reform has had on the progressivity of health care financing. vii 1 Ghana’s health system This section provides a brief overview of Ghana’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 Ghana’s government is committed to improving equity and financial protection in the health sector. In 2005, the Government of Ghana amended its Growth and Poverty Reduction Strategy report to include a new target in the country’s development: to reach middle income status by the year 2015 (Republic of Ghana 2005). Ghana’s Minister of Health has called attention to the role that health plays in economic development and has placed equity in both access and delivery of health services as a top priority for reaching middle income status (Ministry of Health 2007a). The mission statement from Ghana’s Ministry of Health (MoH) sums up this undertaking: “The mission is to contribute to socio-economic development and wealth creation by promoting health and vitality, ensuring access to quality health, population and nutrition services for all people living in Ghana and promoting the development of a local health industry.� Minster of Health Accra, Ghana September, 2007 1.2 Health financing system Health expenditure 1 Ghana spends 8.1 per cent (2009) of its gross domestic product (GDP) on health. This is greater than the spending levels in other lower middle-income countries in Africa, which spend an average of 5.8 per cent (2009) of their GDP on health 2. Government spending on health, as a percentage of total government expenditure, has decreased over the past 5 years. In 2009, 12.8 per cent of total government expenditures were spent on health, down from 15.3 per cent in 2005. On a per capita basis, government health expenditure was equivalent to US$28 (current) and US$65 (PPP-adjusted) in 2009. Out-of-pocket expenditure represented a significant proportion (36.8 per cent) of total health expenditures. 1 Data are from 2009 and available from the WHO National Health Accounts database, accessed May 2011. http://www.who.int/nha/country/gha/en/ 2 Non-weighted average of: Angola, Congo, Cote d’Ivoire, Ghana, Nigeria, Senegal, Swaziland, Zambia, and Sudan. 1 Table 1.1: Health expenditure data, 2009 Indicator Health expenditure as share of GDP 8.1% Government expenditure as share of GDP 29.7% Government expenditure on health as share of total government 12.8% expenditure Government health expenditure, per capita US$28 (current), US$65 (PPP-adjusted) Government expenditure on health as share of total health expenditure 53.2% Out-of-pocket expenditure on health as share of total health 36.8% expenditure Source: WHO National Health Accounts database (2009) Decentralization and centralization Decentralization reforms, which began in the late 1980s in Ghana, have faced many challenges. Although constitutional mandates require the transfer of at least 5 per cent of total national revenues to local government, local government has little discretion over the use of the District Assembly Common Fund (DACF) (World Bank 2003). The District Assembly Common Fund is a portion of the national revenue that is set aside and redistributed to the District Assemblies in Ghana. However, the government of Ghana is currently undertaking measures to increase the fiscal autonomy of all levels of government (Government of Ghana 2008). Revenue-raising/sources of funds Government spending accounted for 53.2 per cent (2009) of the total health spending as shown in Figure 1.1. However, it is important to note that a substantial portion of those funds came from donors in the form of direct budget support, a pooled fund, and project-based support. In fact, in 2009 Ghana received 1.5 billion in official development assistance from foreign donors. Of this amount, approximately 15 per cent was allocated to the health sector (OECD 2009). Money that flows through the government is used to pay for supply-side subsidies of curative care at government hospitals and clinics. Government facilities account for roughly 60 per cent of in-patient and out-patient visits (World Bank 2010a). In 2009, out-of-pocket payments financed 36.8 per cent of all health care expenditures. 2 Figure 1.1: Health care financing mix, 2005 – 2009 100% 90% Other Government 80% Sources 70% Social Security Funds 60% 50% Other Private Sources 40% 30% Private Insurance 20% Out-of-Pocket Payments 10% 0% 2005 2006 2007 2008 2009 Source: WHO National Health Accounts database (2009) Risk-pooling Health financing in Ghana has undergone significant changes in recent years, largely owing to the launch of the National Health Insurance Scheme (NHIS) in 2004. Stemming from a presidential promise to eliminate a heavy reliance on out-of-pocket payments and reduce financial barriers of access to care, the NHIS reimburses both public and private providers and is far-reaching in its coverage. In principle, enrolment in the NHIS is mandatory unless the individual can demonstrate private coverage. However, in practice, coverage is optional for workers in the informal sector in which the majority of the population is employed. Only 5 years after its introduction, the NHIS covered approximately 60 per cent of the population (World Bank 2010b). NHIS financing now accounts for two thirds of internally-generated funds at government facilities and over 40 per cent of total health expenditures in Ghana. While household and patient data reveal that the NHIS has increased utilization of care and reduced financial barriers, the scheme faces a number of challenges. Fraud and weak administrative management threaten the sustainability of the scheme. The core symptom of these challenges is delayed reimbursement of providers, which affects both public and private health care sources, but is particularly damaging to private providers (Sealy et al. 2011). Another challenge is sustainability of the scheme as it relies on tax funding for 70-75 per cent of its revenue (Witter and Garshong 2009). Ghana provides free health services for certain vulnerable groups, such as children under five, people over 70, and pregnant women. In addition, immunization and services to combat certain communicable diseases are provided free of charge. 3 1.3 Health care delivery system Provider organization 3 Both the MoH and Ghana Health Services (GHS) manage the health sector. The MoH is the office responsible for budget allocation and policy definition, while the GHS, with branch offices at both the regional and district levels, is mainly responsible for the implementation of the budget and policies. In addition, non-governmental organizations (NGO) are also very active in the health sector. Health facilities in Ghana consist of four levels of care in urban areas and five levels in rural areas. The health post or outreach sites are the first-level of health care in rural areas. The MoH also provides mobile health services, including immunization and family planning, to rural residents. The next levels of care, in ascending order, are health centers or clinics, district hospitals, regional hospitals, and tertiary hospitals. Ghana also has a network of maternal homes, although they are mainly privately owned. Indeed, there are many private and missionary health facilities in Ghana, most of which are hospitals of limited scale and health clinics. In some cases, the Government of Ghana provides subsidies to these facilities. Payment mechanisms and provider autonomy Health facilities are reliant on four main financial sources for their functioning: the Government of Ghana, financial credits, the internally-generated fund (IGF) which is composed of fees collected from patients, and the donors’ pooled health fund. In addition, there are contributions from donor earmarked funds and donor-managed funds (projects). In Ghana, hospitals receive grants from the national government as the most significant source of funding. Providers receive little cash and the government, which borrows from (and so must repay to) international lending organizations, provides most capital (Preker 2005). The Controller Accountant General (CAG) directly to the public health personnel through the banking system distributes salaries and the district health offices to health facilities mostly distribute other recurrent expenditures. Thus, provider autonomy is fairly limited within Ghana’s current system, but the MoH policy objectives include an increase in financial and managerial autonomy in public hospitals in order to create a more efficient service delivery system which adequately addresses concerns at the local level (Ministry of Health 2007a). Resource availability and utilization Ghana had 1.1 physicians and 9.70 nurses per 10,000 persons in 2008, and 9.3 hospital beds per 10,000 persons in 2009. A primary reason for the low physician density is the migration of physicians to developed countries. 61 per cent of physicians who graduated between 1985 and 1994 emigrated to other countries (Dovlo and Nyonator 1999). In addition to the lack of health care workers, their uneven distribution hinders the country’s ability to provide high quality services in the public system. Approximately 69% of physicians in Ghana practice in the two largest cities (Ministry of Health 2009). As a result, the physician to population ratio in the Greater Accra region is 1:5,000, whereas in the rural Northern region it is 1:92,000 (Ministry of Health 2007b). This uneven distribution of health workers 3 This section, and the one that follows, draws on Austrian Red Cross Accord (2009). 4 disproportionately affects the rural poor who do not have access to urban hospitals due to financial or geographical barriers. 5 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 Ghana in 2008, 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 that 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 2008 DHS, under-five mortality rate, diarrhea, acute respiratory infection (ARI) and fever are worse among the poor. The 2006 MICS also shows that diarrhea and fever are worse among the poor. The concentration index for ARI, as measured using MICS data, is not statistically significant at any level. Both DHS and MICS anthropometric data indicate that stunting and underweight are more concentrated among the poor. The infant mortality rate appears marginally worse among the poor, but this inequality is not statistically significant. According to the 2003 WHS, malaria is worse among the better-off, but this could be due to under-diagnosis among the poor. Table 2.2 shows that, according to the 2003 WHS, tuberculosis, non-road traffic accidents and poor self- assessed health status are more concentrated among the poor. However, road traffic accidents and the prevalence of diagnosed diabetes, by contrast, are more concentrated among the better-off. According to the 2008 DHS, obesity rates in non-pregnant women are more concentrated among the wealthy as well. 6 Table 2.1 : Inequalities in child health Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 1 Infant mortality rate 7.2% 4.7% 7.7% 5.6% 5.3% 6.2% -0.029 Under-five mortality 1 rate 10.6% 8.7% 10.3% 7.0% 5.5% 8.7% -0.093*** 1 Stunting 32.8% 33.8% 27.9% 22.3% 16.4% 27.8% -0.119*** 1 Underweight 18.0% 18.4% 12.1% 9.8% 9.5% 14.3% -0.153*** 2 Underweight 19.3% 17.7% 13.1% 10.7% 4.3% 14.1% -0.214*** 1 Diarrhea 25.6% 21.4% 21.5% 16.9% 10.0% 20.1% -0.133*** 2 Diarrhea 20.5% 15.5% 15.0% 12.1% 11.1% 15.4% -0.114*** Acute respiratory 1 infection 11.1% 12.3% 9.5% 13.2% 6.8% 10.8% -0.057* Acute respiratory 2 infection 11.9% 11.7% 13.8% 9.3% 13.9% 12.1% 0.016 1 Fever 19.9% 22.7% 21.8% 20.3% 14.0% 20.1% -0.045** 2 Fever 25.6% 24.6% 20.5% 22.8% 15.2% 22.4% -0.080*** 3 Malaria 57.2% 65.0% 61.8% 61.4% 76.4% 64.2% 0.044*** 1 2 Source: Authors’ estimates using ADePT and data from 2008 Ghana DHS , 2006 Ghana MICS and 2003 Ghana 3 WHS . Note: *CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 7 Table 2.2: Inequalities in adult health Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 2 Tuberculosis 4.3% 6.2% 3.7% 3.2% 3.0% 4.0% -0.087* HIV positive n/a n/a n/a n/a n/a n/a n/a Obesity among non- 1 pregnant women 2.3% 3.3% 5.2% 12.4% 19.5% 9.4% 0.398*** 2 Road traffic accident 2.2% 2.3% 3.4% 2.7% 4.6% 3.0% 0.175*** Non-road traffic 2 accident 7.2% 8.3% 8.3% 4.9% 3.7% 6.4% -0.123*** 2 Angina 5.2% 6.8% 6.2% 4.6% 4.0% 5.3% -0.069 2 Arthritis 7.3% 10.4% 9.5% 7.6% 6.9% 8.3% -0.033 2 Asthma 3.4% 5.2% 4.3% 5.9% 3.4% 4.4% 0.014 2 Depression 2.7% 0.9% 2.0% 0.7% 1.5% 1.6% -0.144 2 Diabetes 0.1% 0.8% 0.9% 1.4% 2.1% 1.1% 0.372*** Difficulty with work and household 2 activities 5.2% 6.0% 5.7% 7.2% 5.1% 5.8% 0.003 Poor self-assessed 2 health status 9.5% 6.6% 6.5% 6.5% 4.6% 6.7% -0.139*** 1 2 Source: Authors’ estimates using ADePT and data from 2008 Ghana DHS and 2003 Ghana 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 (men and women) and smoking among women only is higher among the poor, while the wealthy are more likely to have insufficient intake of fruit and vegetables. The data show that having more than one sexual partner over the course of a year is more common among the wealthy, although condom usage when involved in concurrent partnerships is also more prevalent among the wealthy. Mosquito net use by children and pregnant women is more common among the poor. 8 Table 2.3: Inequalities in health behaviors Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 3 Smoking (all) 10.4% 8.3% 2.7% 4.3% 3.8% 5.8% -0.244*** 1 Smoking (women) 1.0% 0.3% 0.3% 0.2% 0.1% 0.3% -0.394*** Insufficient intake of 3 fruit and vegetables 35.6% 30.3% 36.5% 41.0% 43.2% 37.4% 0.048*** Insufficient physical 3 activity 3.8% 4.2% 4.6% 6.6% 1.3% 4.1% -0.053 3 Drinking 10.9% 8.0% 12.3% 11.2% 12.7% 11.0% 0.050 Concurrent 1 partnerships 7.8% 15.8% 19.5% 18.5% 16.3% 16.0% 0.088*** Concurrent 2 partnerships 0.7% 1.8% 1.9% 2.2% 1.9% 1.7% 0.170** Condom usage (more 1 than one partner) 22.0% 14.8% 20.0% 27.8% 36.5% 25.0% 0.173*** Mosquito net use by 1 children 30.6% 33.9% 24.7% 32.3% 28.1% 30.2% -0.020 Mosquito net use by 2 children 15.4% 13.3% 9.9% 9.0% 4.9% 11.2% -0.164*** Mosquito net use by 1 pregnant women 37.6% 19.4% 22.2% 20.6% 13.8% 22.2% -0.165*** 1 2 Source: Authors’ estimates using ADePT and data from 2008 Ghana DHS , 2006 Ghana MICS and 2003 Ghana 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 generally ill health and risky health behaviors are concentrated among the poor in Ghana. Almost all of the selected indicators of child health that have statistically significant concentration indices, with the exception of malaria, suggest worse outcomes among the poor. As previously stated, it is likely that the prevalence (and concentration) of malaria reported in the data under-represents the true burden of disease because of lack of access to diagnostic care, especially among the poor. Among adults, ill health, by many measures, is more common among the poor. With respect to risky health behaviors, the results are mixed. Although the wealthy are more likely to use condoms during concurrent partnerships, mosquito net use by women and children are more concentrated among the poor. Smoking is more likely among the poor, but the wealthy are more likely to have an insufficient intake of fruit and vegetables. 9 3 Inequalities in health care utilization In many countries, for a variety of possible reasons, the pattern of health care utilization tends to be distributed very unequally across income groups, even after taking into account differences in medical needs. This section reports on inequalities in utilization of health care in Ghana 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 Ghana in 2008, a Multiple Indicator Cluster Survey (MICS) in 2006, a World Health Survey (WHS) in 2003 and a fifth Living Standards Survey (GLSS V) in 2005/06. Although both the DHS and MICS have rich information for maternal and child health (MCH) interventions, the WHS has fuller data with respect to adult preventive care and general utilization. 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 and for variables drawn from the GLSS V, households are ranked by per capita consumption. 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 that 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 2008 Ghana DHS and the 2006 Ghana MICS. Approximately 79 per cent of children under the age of five are fully immunized, while 80 per cent of expectant women receive at least four skilled antenatal care visits and 59 per cent deliver their baby by a skilled attendant (according to the DHS). Rates of all three interventions are higher among the better-off, and the treatment of diarrhea and ARI are pro-rich as well. The 2006 Ghana MICS indicates that the use of a modern method of contraception is higher among wealthy women than among poor women, while the DHS contraception indicator is not statistically significant. 10 Table 3.1: Inequalities in maternal and child health interventions Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 1 Full immunization 74.6% 77.6% 75.2% 85.4% 85.2% 79.0% 0.029** 1 Treatment of diarrhea 46.2% 57.1% 54.5% 55.4% 55.2% 52.6% 0.052** 2 Treatment of diarrhea 27.5% 30.6% 47.3% 46.8% 52.3% 37.3% 0.158*** Medical treatment of 1 ARI 36.9% 40.7% 50.0% 58.9% 78.7% 48.7% 0.129*** Skilled antenatal care 1 (4+ visits) 65.5% 74.3% 77.7% 91.9% 96.8% 79.8% 0.082*** 1 Skilled birth attendance 24.6% 50.3% 65.1% 82.6% 95.4% 58.9% 0.248*** Contraceptive 1 prevalence 32.8% 31.3% 30.7% 32.4% 32.2% 31.9% 0.003 Contraceptive 2 prevalence 6.5% 13.4% 12.7% 16.0% 12.7% 12.5% 0.088*** 1 2 Source: Authors’ estimates using ADePT and data from 2008 Ghana DHS and 2006 Ghana 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 relatively high uptake of voluntary counseling and testing (VCT) for HIV among the population as a whole (70.8 per cent), but much lower rates for tuberculosis screening (1.1 per cent) and breast cancer screening (1.7 per cent). Breast cancer screening is more common among the better-off. Although neither the concentration index for TB screening nor the concentration index for VCT for HIV is significant, the data suggest that TB screening may be pro-poor while VCT for HIV may be pro-rich. Table 3.2: Inequalities in adult preventive care Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index 2 TB screening 1.2% 0.8% 1.3% 1.8% 0.4% 1.1% -0.044 Voluntary counseling 1 57.5% 67.3% 62.5% 72.5% 80.5% 70.8% 0.065*** and testing for HIV 2 Breast cancer screening 0.3% 1.0% 0.6% 3.6% 3.1% 1.7% 0.406*** 1 2 Source: Authors’ estimates using AdePT and data from 2006 Ghana MICS and 2003 Ghana WHS . Note: *CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 11 Table 3.3 shows the inequalities in general health service utilization in Ghana. According to the 2003 WHS, utilization of both outpatient and inpatient care is higher among the better-off, regardless of the period of measurement. All results are statistically significant. The GLSS V confirms these results, finding that outpatient care is significantly concentrated among the better-off. Results for inpatient care are not statistically significant, but this may be due to the fact that the recall period of two weeks used in the GLSS is too short to capture much variation in inpatient care, which is a more infrequent event than the utilization of outpatient care. Table 3.3: Inequalities in adult curative care Lowest Highest Concentration quintile Q2 Q3 Q4 quintile Total index Inpatient or outpatient 1 (12 months) 43.7% 45.4% 50.8% 54.6% 54.5% 49.9% 0.054*** 1 Inpatient (12 months) 5.4% 7.0% 8.0% 10.9% 10.3% 8.3% 0.143*** 1 Inpatient (5 years) 13.1% 12.6% 17.2% 18.5% 20.7% 16.5% 0.112*** 1 Outpatient (12 months) 41.9% 40.0% 45.0% 46.5% 47.2% 44.1% 0.037** 2 Inpatient (2 weeks) 3.3% 4.6% 2.3% 3.1% 2.5% 3.2% -0.070 2 Outpatient (2 weeks) 11.2% 12.9% 13.3% 13.2% 13.5% 12.8% 0.036*** 1 2 Source: Authors’ estimates using AdePT and data from 2003 Ghana WHS and GLSS V . 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 health care utilization in Ghana is concentrated among the better-off, increasing the poor’s risk for ill health. All MCH interventions have positive concentration indices, signifying that utilization of MCH interventions and treatment of illness is higher among the better-off. Among adult preventives services, most results are statistically insignificant, but breast cancer screening is more concentrated among the better-off. The use of general health services, both outpatient and inpatient, is concentrated among the wealthy. 12 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 Ghana, 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. In Ghana, the results show that Government spending on health does not favor the poor as one might have expected. Further, if one judges the assumption that higher fees also reflect higher subsidies to be plausible (i.e. the proportional unit cost assumption), one would even conclude that Government spending actually favors the better-off. The most pro-rich subsidies are those given to private hospitals, while subsidies to public health centers mildly favor the poor. On a per capita basis, government health expenditure was equivalent to US$28 (current) and US$65 (PPP-adjusted) in 2009. Money that flows through the government is used to pay for supply-side subsidies of curative care at government hospitals and clinics. 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 Ghana 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 4. 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 for this section is based on consumption. A BIA also needs data on GHE (i.e. subsidies) by level of service. Ghana undertook a National Health Accounts (NHA) exercise in 2002 and these data are used in this section 5. Government spending on public health centers is obtained by summing up the contributions made by all Government entities, from the Ministry of Health to parastatals (all of which are listed under the item HP.3.4.5.1 in the NHA 4 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. 5 See http://www.who.int/nha/country/gha/en/ 13 report). Government total spending on hospital outpatient care is computed as total public outpatient expenditure minus the subsidies given to health centers and for other ambulatory care (HC.1.3 – HP.3.4.5.1 – HP.3.9.9). As for government total expenditure on inpatient care in hospitals, this is readily available in the NHA (HC.1.1). It is important to note that even though government expenditure on hospital care is primarily spent on public facilities (HP.1.1.1.1 + HP.1.1.1.2 + HP.1.1.1.3), a non-negligible share of about 20 per cent is also used to subsidize private hospitals (HP.1.1.2.2). Finally, we have divided the subsidies made to government-run and private hospitals into outpatient and inpatient subsidies on a pro-rata basis. 4.2 Inequalities in benefit incidence The tables in this section show the distribution across consumption quintiles of utilization for government facilities, fees paid to these facilities, and estimated subsidies to the health sector. The latter depend 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 reveals differences in the distributions of utilization of public health centers and outpatient services in public hospitals. The utilization of the latter continuously increases with income (from an average proportion of users of 8.6 per cent for the poorest quintile to 11.4 per cent for the richest) whereas the utilization of the former decreases with income for the two richest quintiles. This translates into concentration indices for these two health services that have about the same magnitude but opposite signs, indicating a pro-rich concentration of utilization of outpatient care in public hospitals and a pro-poor concentration of utilization of health centers and health posts. As for utilization of inpatient care, it appears to be pro-rich in public facilities but the statistical support is rather weak. Finally, there is not enough statistical evidence to soundly judge the distribution of utilization of the health services that are provided in private hospitals. Table 4.2 shows the utilization of outpatient and inpatient health services. The first three columns relate to government-run facilities and the last two relate to private hospitals. At 11.4 per cent, outpatient care in public hospitals is the type of care that is the most frequently used. Its use is twice as frequent as outpatient care in health centers and health posts (5.7 per cent) and more than four times as frequent as outpatient care in private facilities. It can also be seen that utilization of inpatient care is clearly more prevalent in public facilities (4.8 per cent) compared to private hospitals (2.1 per cent). 14 Table 4.1: Inequalities in use of public and private facilities Outpatient Outpatient Inpatient Outpatient Inpatient public health public public private private center / post hospital hospital hospital hospital Lowest quintile 6.3% 8.6% 3.9% 1.9% 1.7% 2 7.1% 9.6% 4.6% 3.7% 1.9% 3 7.0% 11.2% 4.4% 2.3% 2.3% 4 4.9% 13.6% 4.6% 1.7% 2.3% Highest quintile 3.3% 14.1% 6.3% 2.7% 2.4% Total 5.7% 11.4% 4.8% 2.5% 2.1% Concentration index -0.103*** 0.106*** 0.089* -0.033 0.080 Source: Authors’ estimates using AdePT and data from the 2003 Ghana WHS. Note: The utilization data refer to the last two weeks in all cases. *CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. Table 4.2: Distribution of fees paid Outpatient Outpatient Inpatient Outpatient Inpatient public health public public private private center / post hospital hospital hospital hospital Lowest quintile 13.5 16.4 7.1 6.1 7.3 2 20.1 12.5 16.8 11.1 6.8 3 25.1 18.9 16.3 3.7 19.4 4 16.3 23.5 39.8 4.9 6.9 Highest quintile 25.0 28.8 20.1 74.3 59.7 Total 100.0 100.0 100.0 100.0 100.0 Concentration index 0.103 0.153*** 0.231*** 0.559*** 0.501** Source: Authors’ estimates using AdePT and data from the 2003 Ghana WHS. Note: *CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. Inpatient admissions in public hospitals require the highest fees. On average during the years preceding the survey, Ghanaians spent GHC 13,626 6 on admissions to public hospitals. It is important to bear in mind that this average is computed using the whole population and not only patients (authors estimates using GLSS V). Given that, in general, only a small fraction of the population uses inpatient care, the average fee paid by the patients for this type of care is much greater. The most relevant information in Table 4.24.2 relates to the distribution of the fees paid across income quintiles. Fees paid during visits to health centers and health posts appear to be more concentrated among the rich but the pattern is not very clear and the concentration index is not statistically significant. On the other hand, the fees paid for 6 As of July 2011, US$1 is equivalent to approximately 15,000 old Ghanaians cedis (GHC). 15 all types of hospital care are clearly more concentrated among the rich, and considerably more so in private hospitals relative to public facilities. 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 inpatient and outpatient health services. 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. Under this assumption, each type of care (for example, each hospital outpatient visit) 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, though, 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 government health spending is almost evenly distributed between outpatient and inpatient care and that about 84 per cent of the subsidies finance public facilities. The first set of results (based on the constant unit-cost assumption) shows that the poorest quintile receives on average 17.6 per cent of government health spending while the richest quintile receives 22.4 per cent. The corresponding concentration index (0.036) is slightly positive but not statistically significant, and so we cannot conclude that public spending is related to individual income. This result is mainly due to the effects of public spending on health centers and outpatient services in public facilities, which are respectively pro-poor and pro-rich and, thus, cancel each other out. When unit subsidies (rather than unit costs) are assumed to be constant (the second set of results), the subsidies to all types of services become more pro-rich. This results in a mildly and statistically significant pro-rich overall effect that has a concentration index of 0.052. Finally, when unit costs are assumed to be proportional to the amount spent out-of-pocket, the subsidies for all types of services become further pro-rich, and considerably so. The resulting picture is a strongly pro-rich incidence of public spending that is driven by considerably pro-rich subsidies to hospital services, especially in private hospitals. Taken together, these benefit incidence analyses find no evidence that government spending on health favors the poor. On the contrary, if one judges that the assumption that higher fees also reflect higher subsidies is most plausible, one would conclude that government spending is, in fact, very much pro- rich. When using the constant unit subsidy assumption, one also finds statistically significant evidence that total subsidies favor the better-off. 16 Table 4.3: Inequality in the incidence of government health spending (shares) Outpatient Outpatient Inpatient Outpatient Inpatient Total public health public public private private subsidies center / post hospital hospital hospital hospital Total subsidies 130.7 207.9 325.1 48.1 75.3 787.2 (billions of GHC) Share of total 16.6% 26.4% 41.3% 6.1% 9.6% 100.0% subsidy Constant unit cost assumption Lowest quintile 22.7 15.0 17.9 16.1 16.7 17.6 2 25.2 17.5 19.2 31.2 18.9 20.5 3 24.4 19.6 18.2 20.3 20.9 20.0 4 17.1 24.0 17.2 13.8 24.1 19.4 Highest quintile 10.6 23.9 27.6 18.6 19.5 22.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 Concentration index -0.118*** 0.099*** 0.069 -0.061 0.056 0.036 Constant unit subsidy assumption Lowest quintile 22.0 15.1 16.2 15.8 16.1 16.8 2 24.9 16.8 19.3 30.4 17.7 20.1 3 24.6 19.6 18.6 18.6 22.0 20.2 4 17.1 23.8 19.5 13.4 21.9 20.1 Highest quintile 11.5 24.6 26.5 21.9 22.3 22.8 Total 100.0 100.0 100.0 100.0 100.0 100.0 Concentration index -0.103*** 0.106*** 0.089* -0.033 0.080 0.052** Proportional cost assumption Lowest quintile 13.5 16.4 7.1 6.1 7.3 10.6 2 20.1 12.5 16.8 11.1 6.8 14.9 3 25.1 18.9 16.3 3.7 19.4 17.9 4 16.3 23.5 39.8 4.9 6.9 26.3 Highest quintile 25.0 28.8 20.1 74.3 59.7 30.3 Total 100.0 100.0 100.0 100.0 100.0 100.0 Concentration index 0.103 0.153*** 0.231*** 0.559*** 0.501** 0.235*** Source: Authors’ calculations using ADePT and data from the 2003 Ghana WHS. Note: When the constant cost assumption is 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. GHC= old Ghanaian cedis. *CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 17 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 Vietnam to provide financial protection in the health sector, and presents data on levels of inequalities in coverage. 5.1 Data availability The fifth wave of the Ghana Living Standard Survey (GLSS V), which was fielded in 2005-06, captures health insurance enrollment. Note that when the survey took place, social health insurance had just been implemented in Ghana. It is thus likely that the new system was not fully operational at the time of the survey, and therefore the data may underestimate the true scope of insurance coverage. A World Health Survey (WHS) was fielded in Ghana in 2003. The WHS has information on health expenditure and household consumption, but less detailed information on health insurance coverage. In order to facilitate international comparisons on catastrophic payments and impoverishment, the majority of the tables below use data from the WHS. The GLSS data are used for insurance coverage. For all the following analyses, households are ranked by per capita expenditure. 18 Box 5.1 Health insurance coverage Table 5A shows health insurance coverage by scheme and by consumption quintile. The latter distribution is summarized in the concentration index, a positive value of which indicates higher coverage rates among the rich and a negative value of which indicates higher coverage rates among the poor. The last column of Table A shows that, in 2005, 16.6 per cent of the Ghanaian population had some type of health insurance. This low uptake is explained by the fact that the new health insurance scheme was not fully operational nationwide at the time of the survey but was then being implemented in only a few districts (Ghana Statistical Service 2008). Also, since this figure also includes those who registered to get health insurance but were not yet covered, it also overestimates the extent of financial protection provided by health insurance at the time. The coverage rate clearly increases with income as only about 5 per cent of the lowest quintile is covered and roughly one quarter of the richest has health insurance. The fact that the rich are disproportionally more covered than the poor is reflected into the concentration index, which at 0.24, is positive. When the analysis is broken down by type of scheme, it can be seen that district mutual health insurance is by far the most prevalent. Consequently, the distribution of this scheme drives the distribution of total insurance coverage. Only 0.5 per cent and 0.6 per cent of the population have private health insurance (either mutual of from a private company) or coverage from another type of scheme, respectively. Both of these type of schemes are strongly pro-rich in their distribution. Table 5A: Inequalities in health insurance coverage Lowest Highest Concentratio quintile Q2 Q3 Q4 quintile Total index District mutual 0.055 0.120 0.172 0.192 0.232 0.154 0.234*** Private mutual or 0.000 0.004 0.001 0.005 0.016 0.005 0.527*** company Other scheme 0.001 0.005 0.009 0.010 0.008 0.006 0.228*** Any insurance 0.057 0.129 0.182 0.206 0.256 0.166 0.243*** Source: Authors’ estimates using ADePT and Ghana GLSS V. Note: *CI is significant at 10%, **CI is significant at 5%, ***CI is significant at 1%. 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. 19 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. Table 5.1 shows that 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 38.6 to 4.8 per cent. When the threshold is raised from 5 to 40 per cent of nonfood expenditure, the estimate of the incidence of catastrophic payments falls from 56.2 to 22.8 per cent. Table 5.1 also shows that the concentration index for catastrophic spending is positive for the WHS data, largely regardless of the total household consumption threshold, implying that catastrophic payments are more common among the better-off. When using nonfood expenditure, however, the concentration indices are negative when the spending threshold is equal to or above 15 per cent for nonfood, implying that catastrophic payments are more common among the worse off at higher thresholds. Table 5.1: Incidence of catastrophic out-of-pocket spending Threshold share of total household consumption 5% 10% 15% 25% 40% Headcount 38.6% 23.6% 16.4% 8.8% 4.8% Concentration index 0.025* 0.027 0.072*** 0.148*** 0.249*** Threshold share of nonfood consumption 5% 10% 15% 25% 40% Headcount 56.2% 50.3% 44.6% 34.5% 22.8% Concentration index 0.020** 0.010 -0.003 -0.035** -0.043** Source: Authors’ estimates using ADePT and data from 2003 Ghana WHS. 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. 20 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 indicates the average shortfall from the poverty line among those in poverty; it is also measured in dollars a day. Table 5.2 reports results for the 2003 Ghana WHS. When out-of-pocket payments are counted as part of a household’s consumption, 49.6 per cent of the population in 2003 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 53.3 per cent; this is the true poverty rate. Thus, about 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 by $0.03, equivalent to or a 12 per cent increase. The mean positive poverty gap increases by $0.02, only a 4.3 per cent increase. The rise in the poverty gap is thus mainly due to more households being brought into poverty through out-of-pocket spending on health, and not because of a deepening of the poverty of the already poor. When using a poverty line of US$2.00 a day, the increase in the percentage of those impoverished and the percentage increase in depth of poverty are smaller. Table 5.2: Impoverishment through out-of-pocket health spending Consumption Consumption Percentage Change including OOP excluding OOP change Poverty line at US$1.25 per capita per day Percentage in poverty / poverty headcount 49.6% 53.3% 3.7 pp 7.6% Average shortfall from the poverty line $0.26 $0.29 $0.03 12.1% Average shortfall from the poverty line, among the poor $0.52 $0.54 $0.02 4.3% Poverty line at US$2.00 per capita per day Percentage in poverty / poverty headcount 72.2% 76.2% 4.0 pp 5.5% Average shortfall from the poverty line $0.73 $0.80 $0.06 8.6% Average shortfall from the poverty line, among the poor $1.00 $1.03 $0.03 2.9% Source: Authors’ estimates using ADePT and data from 2003 Ghana WHS. Note: Poverty lines are at 2005 purchasing power parities, adjusted to current prices using Ghana’s CPI. Figures are for a 4-week figure and are in old Ghanaian cedi. 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. 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, out-of-pocket health expenditures among the extremely destitute are small but grow as the population increases in wealth and some of those who would otherwise be just above the poverty line are brought back down into extreme poverty – even some 21 otherwise relatively well-off households are impoverished by healthcare spending. The chart also shows that many already-impoverished households experience a deepening of poverty as a result of their health spending. Figure 5.1: The impoverishing effect of out-of-pocket spending Source: Authors’ estimates using ADePT and 2003 Ghana WHS. Note: Poverty line is US$1.25 a day at 2005 purchasing power parities, adjusted to current prices using Ghana’s CPI. This section shows relatively high levels of catastrophic expenditure across the income distribution, regardless of the measure used. However, at higher thresholds, catastrophic payments are found to be concentrated among the rich when the total expenditure measure is used, but among the poor when the nonfood measure is used. The data also indicate that although health spending does increase the depth of poverty of many already poor households, the increase in the poverty gap is mostly due to the impoverishment of households (which would, were it not for health spending, have been above the poverty line) rather than the deepening of poverty among the already-poor. Indeed, the increase in the poverty rate due to health spending is 5.5 per cent, when using the US$2.00 a day measure, and 7.6 per cent, when using the US$1.25 a day measure. 22 6 Progressivity of health finance There is a general consensus that payments for health care ought to be at least proportional to households’ ability to pay, if not progressive (meaning a poor household contributes a smaller share of its resources than a rich one). The overall progressivity of a health financing system depends on the progressivity of each source of finance, and the share of health spending financed through each source. A system that relies exclusively on out-of-pocket payments is often argued to be likely to be regressive, since out-of-pocket spending often absorbs a larger share of a poor household’s resources than of a rich household’s resources. This is not always the case, however; when it is not, it is likely that the poor are underusing health care, something that can be assessed by the distribution of health utilization. 6.1 Data availability The fifth round of the Ghana Living Standard Survey (GLSS V) captures spending on social and voluntary health insurance premiums and fees in addition to household consumption and out-of-pocket health expenditure,. Note that when the survey took place in 2005-2006, social health insurance was just beginning to be implemented in Ghana. It is thus likely that the new system was not fully operational and not all households were paying contributions at the time. Contributions to social health insurance also include mandatory social security contributions from the formal sector and these were assessed based on the wages earned by formally employed individuals reported in the GLSS V. Like most household surveys, the GLSS V does not record tax payments; however, the survey includes other useful household and individual variables that can be used to fill this gap. Direct taxes were assessed by applying official tax brackets to individual income, and value added tax (VAT) was derived from household consumption using official VAT rates and exemptions. Information on rates, brackets and exemptions were obtained from the Government of Ghana’s official website. The remaining taxes have been allocated according to the assumptions summarized in Annex A. Finally, data on NHA shares are obtained from a mix of WHO National Health Accounts 2002 and GLSS V aggregates. 6.2 Progressivity of health care financing The first five rows of Table 6.1 show each quintile’s average consumption and financing share with households ranked in ascending order of gross consumption (i.e. consumption including health care payments). Health care payments are considered progressive if the poorest quintile’s share in total household consumption exceeds its share in total payments, while the opposite is true of the richest quintile. Payments are regressive if the poorest quintile’s share in total consumption is less than its share in total payments (while again the opposite is true of the richest quintile). This exercise can be done for total health care payments, as well as for each source separately. Table 1.1 also shows the NHA shares—the percentage of total health financing coming from each source. The next line shows the Gini coefficient, which measures the degree of inequality in gross consumption—the higher the number, the more unequal the distribution of consumption. The line below that shows the concentration index, a measure of how unequally distributed health care payments are across consumption quintiles: a positive value indicates that payments are concentrated among the better-off quintiles, while a negative index would indicate a concentration of payments among the poorer quintiles. The next line shows the Kakwani index, defined as the concentration index less the Gini coefficient. A positive value indicates that payments are more concentrated among the better-off than consumption is and is a sign that 23 payments are progressive. A negative Kakwani index indicates that payments are regressive. Finally, the table indicates the size of the “redistributive effect� associated with health care payments. This is the change in consumption inequality brought about by health care payments. A positive number indicates that there was less inequality in consumption after payments than before, which is the case if payments are progressive. The more progressive they are, and the larger the fraction of (gross) consumption accounted for health care payments, the bigger will be the amount of “redistributive effect�. Table 6.1: Progressivity of health finance Out-of- Voluntary Total Consumption Taxes SHI pocket Insurance payments spending Lowest quintile 4.3 3.9 2.7 1.7 3.4 3.4 2 8.8 8.1 7.3 8.3 9.0 8.0 3 13.6 12.9 13.5 4.0 16.2 13.4 4 20.5 20.2 19.4 17.8 21.6 19.9 Highest quintile 52.7 54.8 57.1 68.1 49.8 55.4 NHA shares 53.3 15.6 0.2 30.1 99.2 Gini coefficient 0.476*** Concentration index 0.498*** 0.521*** 0.590*** 0.458*** 0.505*** Kakwani index 0.022*** 0.046*** 0.115 -0.017 0.029** Source: Distribution of consumption, SHI contributions, voluntary insurance premiums, and out-of-pocket payments estimated by authors using ADePT and data from the GLSS V. Table 6.1 shows that health care financing in Ghana in 2005-2006 was mostly proportional, i.e. the better-off spent a similar fraction of their consumption on health care as the poor. The financing sources that contribute to the overall progressivity of health care finance are general taxation, which finances 53 per cent of spending, and SHI contributions, which finance 16 per cent of spending. SHI is more progressive than general taxation because the contributions are paid largely by formal sector workers who are among the better-off, whereas general taxation includes earmarked indirect taxes that greatly attenuate the progressivity of direct taxes. Voluntary insurance is clearly progressive as it is mostly taken up by richer households, but this barely affects the overall health care finance given that its financing share is lower than 1 per cent and the concentration index is not statistically significant. As for out-of-pocket payments, even though their slight regressivity is not statistically significant, it is sufficient to offset the progressivity of the other financing sources because of their considerable financing share (31 per cent). In sum, this section finds that taxes in Ghana are slightly progressive and SHI contributions are mildly progressive. As for voluntary health insurance premiums and out-of-pocket payments, they appear to be progressive and slightly regressive respectively, but the results are not statistically significant. Overall, health care finance in Ghana emerges as being mostly proportional to income. It is important to note that the GLSS data used for this analysis pre-date the recent health insurance reform and it is not known what effect this reform has had on the progressivity of health care financing. Since the 2009 WHO National Health Accounts database (see Table 1.1) reports out-of-pocket expenditure on health of 37 per cent, but this figure was only 30 per cent in the 2005/06 GLSS V, it is possible that health financing might have become more regressive than these findings suggest. 24 7 References Austrian Red Cross Accord. (2009). Health Care in Ghana. Vienna: Austria. Dovlo, D. and F. Nyonator. (1999). “Migration by graduates of the University of Ghana Medical School: a preliminary rapid appraisal.� Hum Res Dev J. 3: 40-51. 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. Ghana Statistical Service. (2008). Ghana Living Standards Survey: Report of the Fifth Round (GLSS V). Accra, Ghana: Ghana Statistical Service. Government of Ghana. (2008). Intergovernmental Fiscal Decentralisation Framework. Ministry of Local Government, Rural Development and Environment. 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. Ministry of Health. (2007a). Creating Wealth through Health. Accra, Ghana. Ministry of Health. (2007b). Facts and figures: human resources for health. Accra, Ghana. Ministry of Health. (2009). Regional distribution of doctors in Ghana. Accra, Ghana. 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. OECD. (2009). “Aid at a glance chart: Ghana.� Retrieved August 20, 2011. http://www.oecd.org/dataoecd/21/40/1881076.gif. Preker, A. S. (2005). Spending wisely: buying health services for the poor. Washington, DC: World Bank. Republic of Ghana. (2005). Growth and Poverty Reduction Strategy (GPRS II) 2006-2009. National Development Planning Commission: Accra, Ghana. 25 Sealy, S. M. Makinen et al. (2011). Private Health Sector Assessment in Ghana. Washington, DC: World Bank. 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. Witter, S. and B. Garshong. (2009). “Something old or something new? Social health insurance in Ghana.� BMC Int Health Human Rights. 9: 20. World Bank. (2003). Decentralization Policies and Practices Case Study Ghana Participants Manual.Washington, D.C.: World Bank. World Bank. (2010a). “Health services utilization and out-of-pocket expenditure at public and private facilities in low-income countries.� World Health Report 2010, Background Paper 20. Washington, DC: World Bank. World Bank. (2010b). Republic of Ghana: Improving the Targeting of Social Programs. Washington, D.C.: World Bank. 26 8 Annexes 8.1 Additional graphs and tables Table A1: Tax progressivity assumptions % revenue share Concentration index Comment PIT 15% 0.2217 CIT 13% 0.2217 Assumed to be distributed as PIT VAT 22% 0.0104 NHIL 5% 0.0104 2.5% additional VAT rate, earmarked for health Other indirect taxes 19% 0.0104 Assumed to be distributed as VAT Trade taxes 18% 0.0104 Assumed to be distributed as VAT Total 0.0220 Source: revenue shares come from Table 1 in the summary of central government operations 2002-2007, Government of Ghana Official Website. 27 8.2 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, (Note: z-score calculated using WHO 2006 Child Growth Standards) MICS Underweight % of children with a weight-for-age z-score <-2 standard deviations from the reference median DHS, (Note: z-score calculated using WHO 2006 Child Growth Standards) MICS 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, infection MICS 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) 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 28 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, than one partner) during last sexual intercourse MICS 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, children MICS 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, either verified by card or by recall of respondent MICS 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, respiratory infection (past 2 weeks) MICS 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, prevalence MICS 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, and Testing for HIV MICS 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 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) Note: Unless otherwise noted, all children are under the age of 5 and all adults are aged 18 and older 29 8.3 Methodological notes Sections 2 and 3: Inequalities in health and health care utilization The selection and measurement of health outcome indicators used in Section 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 3: 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, 30 specifically from one or more of the following tables depending on the level of detail provided: financing 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. Section 6: Progressivity of health care finance This section examines the progressivity of different sources of healthcare financing/payments, including out of pocket payments, health insurance contributions, direct taxation and indirect taxation. The Kakwani index, defined as the concentration index minus the Gini coefficient, indicates whether payments are more/less concentrated among the better-off than consumption is and, thus, is a sign of whether payments are progressive/regressive. The main data source needed for the analysis of progressivity of health care financing is a multipurpose household survey, preferably with a very detailed consumption module. In addition, knowledge of the local context is typically needed to make informed assumptions, such as information on income tax brackets, VAT tax rates and exemptions, excise taxes, and taxes that are earmarked for health. Where the data do not contain information on direct taxes, this value was calculated by applying official tax brackets to individual reported income. However, in low income countries characterized by high degrees of informality and limited tax collection capacity, this approach may overestimate direct payments. Where the data do not contain information on value added tax (VAT), this is derived from household consumption using official VAT rates and exemption categories, obtained from government websites 31 and various literature. Other important assumptions about the distribution of the burden of taxation that are particular to each country are described in Annex A of the corresponding report. 32