WPS4554 Policy ReseaRch WoRking PaPeR 4554 Measuring Financial Protection in Health Adam Wagstaff The World Bank Development Research Group Human Development and Public Services Team March 2008 Policy ReseaRch WoRking PaPeR 4554 Abstract Health systems are not just about improving health: if it exceeds a certain percentage of the living standards good ones also ensure that people are protected from measure; the second defines spending as impoverishing the financial consequences of receiving medical care. if it makes the difference between a household being Anecdotal evidence suggests health systems often above and below the poverty line. The paper provides perform badly in this respect, apparently with devastating an overview of the methods and issues arising in consequences for households, especially poor ones and each case, and presents empirical work in the area of near-poor ones. Two principal methods have been used financial protection in health, including the impacts to measure financial protection in health. Both relate of government policy. The paper also reviews a recent a household's out-of-pocket spending to a threshold critique of the methods used to measure financial defined in terms of living standards in the absence of protection. the spending: the first defines spending as catastrophic This paper--a product of the Human Development and Public Services Team, Development Research Group--is part of a larger effort in the department to shed light on health financing and delivery issues. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at awagstaff@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Measuring Financial Protection in Health by Adam Wagstaff The World Bank, Washington DC, USA Corresponding author and contact details: Adam Wagstaff, World Bank, 1818 H Street NW, Washington, D.C. 20433, USA. Tel. (202) 473-0566. Fax (202)-522 1153. Email: awagstaff@worldbank.org. Keywords: financial protection; catastrophic spending; impoverishment; health shocks; insurance. Acknowledgements: My thanks to Peter Smith, Elias Mossialos and Sheila Leatherman for asking me to write this paper which was prepared for a book they are editing called Performance Measurement for Health System Improvement: Experiences, Challenges and Prospects. The findings, interpretations and conclusions expressed in this paper are entirely those of the author, and do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. 1 Introduction Health systems are not just about improving health. Good ones also ensure that people are protected from the financial consequences of illness and death, or at least from the financial consequences associated with the use of medical care. Anecdotal evidence suggests health systems often perform badly in this respect, apparently with devastating consequences for households, especially poor ones and near-poor ones. The World Bank's 50-country participatory poverty study known as Voices of the Poor1 found that poor health and illness are universally dreaded as a source of destitution, partly because of the costs of health care but also the income lost due to illness. Voices of the Poor documents the case of a 26 year-old Vietnamese man who, as a result of the large health care costs necessitated by his daughter's severe illness, has moved from being the richest man in his community to being one of the poorest.2 Also recorded was the case of a 30- year-old Indian mother of four who has been forced to sell the family's home and land, and has to walk 10 kilometers a day transporting wood on her head in order to finance the cost of her diabetic husband's medical care.2 How can one measure the success with which a health system protects people against the financial consequences of ill health? What do systems that do well in this regard have in common? And how far do health system reforms improve people's financial protection vis-à-vis health expenses? This paper provides an overview of the methods and issues arising in each case, and presents empirical work in the area of financial protection in health, including the impacts of government policy. The paper also reviews a recent critique of the methods used to measure financial protection. 2 Some preliminaries The measures of financial protection developed to date are based on people's out- of-pocket spending on medical care, and relate out-of-pocket payments to a threshold.3 The idea is that out-of-pocket spending is largely involuntary and does not contribute to household well-being in the way that spending on, say, a new car does, and that a household unfortunate enough to have to spend on medical care is deprived of resources it could have used to purchase other goods and services, including necessities such as food and shelter. One approach is to classify spending as `catastrophic' if it exceeds a certain fraction of household income. Another is to classify it as `impoverishing' if it is sufficiently large to make the difference to the household being above the poverty line and below it, i.e. in the absence of the medical outlays the household's resources would have been sufficient to keep its living standards above the poverty line, while with the outlays its living standards are pushed below the poverty line. Several general issues arise with these approaches. One is that the focus in on the cost of medical care. The income losses associated with illness, injury and death are not captured, even though they may be more important in terms of their impact on household welfare. The justification for this omission is that the measures aim at measuring financial protection vis-à-vis health care expenses, and that protecting households against income losses is not the business of the health financing system but of the social protection system more generally. Second, the assumptions that out-of-pocket spending on health is involuntary and that such spending automatically deprives the household of the resources in question are worth thinking about. They are discussed further below. Third, the focus on what households end up spending is argued by some to miss an 3 important point, namely that people may be deterred from using health services by the high out-of-pocket cost. A country where people pay little out-of-pocket (and which therefore looks good from a financial protection perspective) may be one where people do not use health services. Some argue that this ought to be captured by a measure of financial protection. One the face of it, the suggestion that measures of financial protection should capture forgone utilization caused by a high out-of-pocket cost seems reasonable enough. But on reflection it becomes clear the suggestion is confusing policy objectives and policy instruments. Policymakers have multiple variables they wish to influence (focal variables), including health outcomes and people's expenditure on health (and by implication the resources people have available for other goods and services). In seeking to influence these variables, policymakers have a number of instruments at their disposal, including the share of the cost of health care that people pay out-of-pocket. A change in a given instrument will likely affect several focal variables. So, exempting the poor from user fees at public facilities will likely affect use of services by the poor (non-use and under-utilization by the poor should fall) and the amount that the poor end up paying out- of-pocket. In a health systems assessment, the natural approach is to see how well the system fares in terms of the focal variables, and then to work backwards to see how far the performance can be attributed to the specific set policies that have been adopted. A country might do well on financial protection but poorly on health outcomes and health inequalities; the reason may be that its policies on out-of-pocket payments discourage 4 most people from using health services but that the health services people do use are high quality and appropriate. Another country might do poorly on financial protection and poorly on health outcomes and inequalities; the reason might be that people use services despite the high cost at the point of use but the care is poor quality or inappropriate to people's needs. This example brings home the important point that performance on financial protection depends not just on policies with respect to health financing narrowly defined but also (among other things) on the way providers are paid and regulated. Catastrophic expenditures: The basics A natural starting point--and in many studies the stopping point--is to examine the distribution of `catastrophic' health expenditures, defined as health spending that exceeds some threshold, defined usually in relation to the household's `pre-payment' income. Figure 1 illustrates. The x-axis plots out-of-pocket spending on medical care (M) and the y-axis plots expenditure on other items such as food, housing, transport, etc., labeled non-medical spending (NM). The budget line is a 450 line--each dollar spent on medical care means one dollar less to spend on other things. It is this fact that underpins the concern over financial protection, the view being that medical care outlays are different from spending on other goods and services, being involuntary and the response to a unwanted health shock, and having an entirely negative effect on household welfare by depriving a household of resources that could have been spent on goods and services that do contribute to welfare. In Figure 1 the household has an income equal to x (the intercept on both the x-axis and the y-axis), and spends M0 on medical care and NM0 on other items. One approach is to define out-of-pocket medical spending as catastrophic if 5 it exceeds a certain amount in monetary terms.4 An alternative approach3 is to say spending is catastrophic if it exceeds some specified fraction of pre-payment income, x, defined as the sum of observed medical outlays M0 and observed non-medical spending NM0. Alternatively the threshold could be defined in terms of pre-payment income less a deduction for food and perhaps other necessities too.3,5 The idea is that by subtracting a deduction for basic necessities one gets a better idea of the individual's ability to pay. One could deduct an individual's (or household's) actual food expenditure, labeled F0 in Figure 1. Or one could deduct an amount that represents society's view about the minimum acceptable level of expenditure on food (and perhaps other necessities) as reflected in a poverty line, labeled PL in Figure 1. This latter approach is problematic when a household's pre-payment income falls short of the poverty line: in this case, the household's estimated `ability to pay' is negative and it falls below the catastrophe threshold automatically whatever its medical care outlays.3* *Xu et al.5 use this approach. Their poverty line is just for food expenditures, which is subtracted apparently from non- medical consumption (NM0) rather than pre-payment income (x). Ability to pay is defined as NM0-PL except for households for whom this is negative. In such cases, ability to pay is defined as NM0 less actual food expenditure. This leads to the rather unsatisfactory outcome that a household just below their poverty line could be judged to have the same ability to pay as one just above it. 6 Figure 1: Defining catastrophic health spending Non-medical expenditures (NM) x NM0 Poverty line (PL) F0 450 M0 x Medical expenditures (M) Source: Author. The precise fraction of pre-payment income (with or without some deduction for basic necessities) is, of course, arbitrary, and it makes sense to examine the sensitivity of one's results to the threshold chosen. Figure 2 plots catastrophic spending curves for a variety of years for Vietnam. These curves plot on the y-axis the fraction of households experiencing catastrophic out-of-pocket spending for a given threshold on the x-axis. In this particular instance, the choice of threshold is irrelevant, and the conclusion is that the incidence of catastrophic spending has fallen continuously over the period whatever threshold is chosen. 7 Figure 2: Catastrophic spending curves, Vietnam various years 100% 1993 90% 1998 2004 80% 2006 70% reshold 60% th 50% 40% exceeding % 30% 20% 10% 0% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% threshold (% of nonfood expenditure) Source: Author. One might also want to move beyond counting the number of households who overshoot the threshold to capturing the amount by which they overshoot it, just as in the poverty literature one looks not just at the number of people in poverty but at the poverty gap--the depth below which people fall below the poverty line. The catastrophic payment gap is simply the aggregate or average amount by which out-of-pocket spending exceeds the threshold.3 Figure 3 plots out-of-pocket payments as a share of income on the y-axis against the cumulative share of the population on the x-axis, ranked in decreasing order of out-of-pocket payments as a share of income. By reading off the curve at the threshold one gets the catastrophic payment headcount--the fraction whose payments exceed the threshold. The (aggregate) catastrophic payment gap is the area above the threshold line below the curve--it shows the overall amount by which payments exceed the threshold in the sample. 8 Figure 3: Catastrophic spending gap Out-of-pocket payments as % income Catastrophic payment gap threshold % exceeding threshold Cumul. % pop. ranked in decreasing order of out-of-pocket payments as % income Source: Wagstaff and van Doorslaer3. A final modification is to make some allowance for whether it is well-off households who exceed the threshold or worse-off ones. It seems likely that policymakers would be more concerned if it is the latter rather than the former. One could tabulate the incidence of catastrophic payments and the catastrophic payment gap by pre-payment income quintile, or one could compute a concentration index for each.3 The concentration index for the catastrophic health expenditure `headcount', for example, would be negative if catastrophic expenditures were, on average, more common among the worse off. Of course, it could be that that the fraction of the population experiencing catastrophic spending has increased over time, but has become less concentrated among the poor. A natural summary measure that takes both into account is the catastrophic payment headcount multiplied by the complement of the concentration index.3 This is equivalent to constructing a rank-weighted average of the catastrophic payment indicator 9 (1 if the threshold has been exceeded, zero otherwise), where the weight is decreasing in the person's rank in the income distribution. If N is the sample size, the weight is 2 for the poorest person, and declines by 2/N for each one-person step up through the income distribution, reaching 2/N for the richest person. Catastrophic expenditures: Empirical studies Xu et al.5 report the incidence of catastrophic health spending (using a 40% threshold) in 59 countries and find large differences; their results are shown in Figure 4. Xu et al.6 have recently produced estimates for 89 countries covering 89% of the world's population, again using the 40% threshold. Their estimates range from 0% in the Czech Republic, Slovakia and the United Kingdom to more than 10% in Brazil and Vietnam. Several OECD countries--Portugal, Spain, Switzerland and the United States--all record rates in excess of 0.5%. 10 Figure 4: The incidence of catastrophic out-of-pocket payments in 59 countries 12 ts menyap 10 ocket 8 of-p- out ichporst 6 ta ca gnci 4 eni per 2 ex % 0 man li end ccro caiRa o y h h ai az lag A d ai aibi ut n e ce c ium UKan Spain nFi Israe anl nada et Br ani no aidob ru pt ai Pe US Egy Latv rgalut nama aib eeec augra m iaralg m Kor ne l ituob aino dnaleayw tao ngary m lge ai kra ainevo mn rm So Fra Cz Vi anjiabre iab Za Gr ac uB Jamai Se DjiEst Ic Nor Cr HuSwe Mo st Na B CaRoman De Sl Ge Slovak Az mlooC ntegAr yaugar Leban m Ukraine Ca Pa Po Pa Ni ac ae ne ocix aniua anaGh siutiur ais ankaLi hsedalng dlaniaTh senpipilihP natyzsrgKy anyaGu danerlztiSw Ye Me htiL Ma Rep. Indone Sr Ba Co Source: Xu et al.5. Van Doorslaer et al.7 look at catastrophic spending in 10 Asian territories. They find relatively low rates in Malaysia, Sri Lanka and Thailand, and relatively high rates in China, Vietnam and Bangladesh. This study also looks at the distribution by pre-payment income of those experiencing catastrophic payments. For the most part, they find that catastrophic spending is concentrated among the better off, though this depends to some degree on the threshold chosen. Taiwan (China) is the exception: catastrophic spending is concentrated among the poor whatever the threshold. A different picture emerges in the study by Waters et al.4 of the United States: they find a higher incidence of catastrophic spending among poor families, as well as those with multiple chronic conditions. 11 A number of studies explore how policies and institutions impact on the incidence of catastrophic health spending. Xu et al.5,6 find that rates of catastrophic spending are higher in poorer countries and in countries with limited prepayment systems. In their most recent study6, they find that (controlling for whether prepayment as a share of health spending exceeds 50%) whether a country operates a tax-financed financing system or a social health insurance system makes no difference to the incidence of catastrophic spending. Looking at their cross-country differences, Van Doorslaer et al.7 speculate that the low incidence of catastrophic spending in Sri Lanka, Malaysia and Thailand reflects the low reliance on out-of-pocket spending in financing health care and the limited use of user fees in the public sector. By contrast, the high rate of incidence in Korea is argued to reflect the high co-payments in that country's social insurance system and the partial coverage of inpatient care. Several country-level studies conclude that insurance reduces the risk of catastrophic health spending. Gakidou et al.8 and Knaul et al.9 find that the introduction of the Popular Health Insurance scheme in Mexico from 2001 onwards led to a reduction in the incidence of catastrophic health expenditures. Limwattananon et al.10 find that rates of catastrophic spending in Thailand were lower after the universal health care scheme was introduced in 2001. Habicht et al.11 find that the risk of catastrophic spending in Estonia has increased during the late 1990s and early 2000s, and attribute this in part to rising co-payments (and hence a decrease in the depth of coverage) linked to a decline (in real terms) in government health spending, and in part to a graying of the population and the elderly having shallower coverage, especially for medicines. 12 Other studies point to the limitations of insurance to reduce and eliminate catastrophic spending. Wagstaff and Pradhan12 find that the introduction of a social health insurance scheme in Vietnam in 1993 reduced the incidence of catastrophic expenses, while Wagstaff13 finds that the subsequent extension of the scheme to the poor (financed through general revenues) also did so; however, the percentage reductions were estimated to be small, and high rates of catastrophic spending are observed even among those with insurance. One factor explaining these results is that insurance appears to have increased the utilization of services in Vietnam. Xu et al.14 find that rates of catastrophic out-of-pocket spending among the population as a whole fell in Uganda after the removal of user fees in 2001; however, the rate among the poor increased. They speculate that this was due to the frequent unavailability of drugs at government facilities after the removal of user fees which forced patients to buy drugs from private pharmacies, and that informal payments to health workers increased to offset lost revenues from fees. Devadasan et al.15 look at the effects of two community health insurance schemes in India on the risk of catastrophic out-of-pocket payments, and conclude that the schemes reduced the risk but only by half. They attribute the limited impact to benefit packages having low maximum limits, the exclusion of some conditions from the package, and the use of the private sector for some inpatient admissions. Ekman16 finds that insurance increases the risk of catastrophic spending in Zambia. He suggests that the amount of care per illness episode may have increased, and that quality assurance and the oversight of service providers is important in determining how far insurance reduces the risk of catastrophic spending. Three recent studies from China reinforce these points. Wagstaff and Lindelow17 find that China's urban insurance 13 scheme increases the risk of catastrophic out-of-pocket spending, and attribute the results in part to weak regulation of providers coupled with a fee-for-service payments system and a fee schedule that allows providers to make profits on drugs and high-tech care results in insured patients receiving more complex care and from higher-level (and hence more costly) providers. Wagstaff et al.18 find that China's new rural insurance scheme does not appear to have reduced the incidence of catastrophic health spending; they attribute this to the exclusions, high deductibles, low reimbursement ceilings, and similar supply responses to those seen in the urban setting. By contrast, Wagstaff and Yu19 find that supply-side interventions in rural China (including the introduction of treatment protocols and essential drug lists) did reduce the incidence of catastrophic health spending. Impoverishing expenditures: The basics A difficulty with the "catastrophic" payment approach is that it is blind as to how far `catastrophic' payments actually cause hardship. One household might have spent more than 25% of its pre-payment income on health and yet be nowhere near crossing the poverty line as a result of the expenditure. Another might have spent just 1% of its pre- payment income and yet have crossed the poverty line. An alternative perspective to catastrophic health expenditures is that of impoverishment, the core idea being that no one ought to be pushed into poverty--or further into poverty--because of health care expenses. An obvious way to proceed is to classify a household as impoverished by out-of- pocket payments on medical care if its pre-payment income (x in Figure 1) lies above the 14 poverty line (PL) and its non-medical spending (NM0) lies below the poverty line.3 One could get a sense of how far out-of-pocket payments cause impoverishment by comparing the pre-payment poverty headcount (the fraction of households for whom x>PL) with the post-payment poverty headcount (the fraction of households for whom NM0