Articles Progress on catastrophic health spending in 133 countries: a retrospective observational study Adam Wagstaff*, Gabriela Flores*, Justine Hsu, Marc-François Smitz, Kateryna Chepynoga, Leander R Buisman, Kim van Wilgenburg, Patrick Eozenou* Summary Background The goal of universal health coverage (UHC) requires inter alia that families who get needed health Lancet Glob Health 2018; care do not suffer undue financial hardship as a result. This can be measured by the percentage of people in 6: e169–79 households whose out-of-pocket health expenditures are large relative to their income or consumption. We aimed Published Online December 13, 2017 to estimate the global incidence of catastrophic health spending, trends between 2000 and 2010, and associations http://dx.doi.org/10.1016/ between catastrophic health spending and macroeconomic and health system variables at the country level. S2214-109X(17)30429-1 See Comment page e124 Methods We did a retrospective observational study of health spending using data obtained from household *Contributed equally surveys. Of 1566 potentially suitable household surveys, 553 passed quality checks, covering 133 countries between Development Research Group 1984 and 2015. We defined health spending as catastrophic when it exceeded 10% or 25% of household consumption. (A Wagstaff DPhil) and Health, We estimated global incidence by aggregating up from every country, using a survey for the year in question when Nutrition and Population available, and interpolation and model-based estimates otherwise. We used multiple regression to explore the relation Global Practice (M-F Smitz MSc, P Eozenou PhD), World Bank, between a country’s incidence of catastrophic spending and gross domestic product (GDP) per person, the Gini Washington, DC, USA; coefficient for income inequality, and the share of total health expenditure spent by social security funds, other Department of Health Systems government agencies, private insurance schemes, and non-profit institutions. Governance and Financing, World Health Organization, Geneva, Switzerland Findings The global incidence of catastrophic spending at the 10% threshold was estimated as 9·7% in 2000, 11·4% in (G Flores PhD, J Hsu MSc, 2005, and 11·7% in 2010. Globally, 808 million people in 2010 incurred catastrophic health spending. Across K Chepynoga MA, 94 countries with two or more survey datapoints, the population-weighted median annual rate of change of K van Wilgenburg MSc); and catastrophic payment incidence was positive whatever catastrophic payment incidence measure was used. Incidence Institute of Health Policy and Management, Erasmus of catastrophic payments was correlated positively with GDP per person and the share of GDP spent on health, and University, Rotterdam, incidence correlated negatively with the share of total health spending channelled through social security funds and Netherlands (L R Buisman PhD) other government agencies. Correspondence to: Dr Adam Wagstaff, Development Interpretation The proportion of the population that is supposed to be covered by health insurance schemes or by Research Group, World Bank, Washington, DC 20433, USA national or subnational health services is a poor indicator of financial protection. Increasing the share of GDP spent awagstaff@worldbank.org on health is not sufficient to reduce catastrophic payment incidence; rather, what is required is increasing the share of total health expenditure that is prepaid, particularly through taxes and mandatory contributions. Funding Rockefeller Foundation, Ministry of Health of Japan, UK Department for International Development (DFID). Copyright © 2017 The World Bank and World Health Organization. Published by Elsevier. This is an Open Access Article published under the CC BY 3.0 IGO license which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In any use of this Article, there should be no suggestion that The World Bank or WHO endorse any specific organisation, products or services. The use of The World Bank or the WHO logo is not permitted. This notice should be preserved along with the Article’s original URL. Introduction especially large relative to a family’s total income or Although, globally, the share of health spending by patients consumption) might reflect people getting needed care but themselves at the point of care (so-called out-of-pocket being protected from out-of-pocket costs. However, a low payments) has been falling, out-of-pocket spending as a incidence of catastrophic payments could also mean share of income has not been declining. This fact has people not getting (and not paying for) needed care. The prompted concerns about the two aspects of universal two dimensions of UHC need to be examined together. health coverage (UHC): first, that everyone—poor and rich The second dimension of UHC (financial protection) alike—should receive needed health care (referred to as can be captured through two indicators.2,3 In this Article, service coverage);1 and second, that families who do get we aimed to present global estimates for one of needed care do not suffer undue financial hardship as a these indicators—namely, catastrophic out-of-pocket result (referred to as financial protection).2 Strong spending. This measure is the official indicator for performance on one UHC dimension does not guarantee monitoring of UHC financial protection among the strong performance on the other. A low incidence of Sustainable Development Goals (SDGs; indicator 3.8.2), catastrophic payments (ie, out-of-pocket payments that are with large expenditure suggested to be defined as 10% www.thelancet.com/lancetgh Vol 6 February 2018 e169 Articles Research in context Evidence before this study rather than 116, and explored how catastrophic payments vary In a global study of catastrophic spending from 2007, which with the share of total health spending channelled through was based on data from 116 health surveys covering different types of publicly and privately financed prepayment 89 countries and with a median survey year of 1997, arrangements. We also investigated the degree to which catastrophic spending was defined as spending that absorbs catastrophic payment incidence was associated with the fraction more than 40% of total consumption, net of an allowance for of the population covered by a health insurance scheme or by a food expenditures. This threshold was set equal to average national or subnational health service, an indicator suggested as food spending among households in which the food spending a possible measure of universal health coverage (UHC). share (as a percentage of total consumption) was in the Implications of the available evidence 45th to 55th percentile range, the assumption being that, In roughly half of countries, the incidence of catastrophic at least in low-income and middle-income countries, the daily spending has been rising, at both the 10% and 25% thresholds, food intake of this group averages 2000 kcal. The study whereas in around 40% of countries, catastrophic spending reported mean and median catastrophic spending incidence incidence has been increasing using the non-food measure. of 2·3% and 1·5%, respectively, and estimated that 150 million However, for all measures, the population-weighted median people globally incur catastrophic spending annually. annual rate of change of catastrophic payment incidence has Catastrophic spending was (partly) correlated with the share been positive. The incidence of catastrophic spending varies of prepayment in total health spending (negative) and the considerably across countries at any given point in time. Gini coefficient for income (positive), and in low-income and This variation does not reflect differences in the share of the middle-income countries with the share of gross domestic population covered by a health insurance scheme or by a product (GDP) devoted to health (positive). national or subnational health service: variations exist among Added value of this study countries officially covering the entire population, and incidence We not only used the official Sustainable Development Goal changes over time during periods when health coverage (SDG) indicator for financial protection but also compared our arrangements and rates have not changed. What coverage rates results with findings obtained when catastrophic spending was miss, and catastrophic payment incidence captures, is the extent defined as occurring if out-of-pocket spending exceeded 40% of of de jure and, more importantly, de facto coverage of different non-food consumption—a definition that is close to the one services. Just increasing the share of GDP spent on health does used in two previous global studies. Our data are more recent not seem to be sufficient to provide financial protection. We find than those used in two previous studies from 2003 and 2007, that the incidence of catastrophic payments decreases with both extend country coverage from 89 to 133, report trend data for the share of health spending that is channelled through social 94 countries, and estimate catastrophic spending incidence security funds and the share channelled through other globally for 3 years—2000, 2005, and 2010. As in the government financial protection arrangements; evidence two previous studies, we analysed country-level correlates of suggests that the negative association is stronger for catastrophic spending incidence, but did so using 553 datapoints government financial protection arrangements. and 25% of total household expenditure. A companion associated with coverage by a health insurance scheme paper4 presents results for the second widely used or by a national or sub­ national health service,8 an indicator of financial protection—namely, medical indicator proposed by some but rejected by others as a impoverishment.3,5 Impov­ ish­ er­ ment is not an official possible measure of UHC.9 SDG indicator but supplements the catastrophic payment indicator by trying to highlight the poverty Methods implications of out-of-pocket spending. Catastrophic payments as a measure of financial hardship Our study updates and extends two previous global We focused on one measure of financial hardship studies undertaken in 20036 and 2007.7 We use the that has been used widely in previous studies,3,6,7,10–16 official SDG definitions for catastrophic payments and typically referred to as catastrophic health expenditure. include data for 133 countries (median year 2010). We Catastrophic spending can be measured in different See Online for appendix estimate annual average changes in incidence of ways (appendix). The idea is, in effect, to measure the catastrophic spending for 94 countries and report global incidence of financial hardship caused by health and regional estimates for 2000, 2005, and 2010. We payments—ie, the number of households with health also use multiple regression methods to search for spending that is large relative to their ability to pay. macroeconomic and health system variables that are There is no right or wrong way to measure ability to associated with the incidence of catastrophic spending pay. One key question is whether it is reasonable to at the national level. We also aimed to investigate the expect households to borrow or use savings to finance degree to which catastrophic payment incidence is their health spending, as many do.17,18 If the answer e170 www.thelancet.com/lancetgh Vol 6 February 2018 Articles is no, ability to pay should be measured using Estimating catastrophic spending aggregates current income. If the answer is yes, then the The household surveys we use are nationally represen­ household’s health expenditure—even if financed out tative, so our analysis of a household survey leads directly of borrowing—represents resources available to the to a national estimate of the incidence of catastrophic household, and ability to pay should be measured by spending for that country in that year. We also estimated total consumption gross of health spending. The the regional and global incidence of catastrophic answer could depend on what type of care is being spending, using UN regions and three reference years: purchased: a government might be reluctant for 2000, 2005, and 2010. The process entailed estimating households to have to borrow or use savings for incidence at the country level, then aggregating up. acute medical care but be comfortable about them We used surveys from up to 5 years before and up to contributing from their savings towards the cost of 5 years after to estimate incidence in each of the long-term elder care. The choice of yardstick matters three reference years, using a mix of survey datapoints, less for the overall incidence of catastrophic spending imputation, extrapolation, and modelling as needed than for the measurement of inequality in catastrophic (appendix). Table 1 provides a breakdown of the types of spending, with a consumption-based measure of ability country datapoints used to estimate the global and to pay leading to more pro-rich inequality than an regional incidence of catastrophic payments. For income-based measure. In any event, because in low- example, for the reference year 2010, we used actual income and middle-income countries it is difficult to survey-based datapoints for 101 countries, for which at measure income with any accuracy, we have little choice least one point was available between 2005 and 2015. but to use consumption, except in a few countries for Together, these countries represent 86·1% of the world’s which consumption is not available; we, therefore, do population. For 54 of these 101 countries, the survey was not report results on inequality in the incidence of undertaken in 2010, so we relied on the actual survey- catastrophic spending in this Article. based estimate of the incidence of catastrophic payments. However ability to pay is measured, the question arises For the other 47 countries, we aligned survey estimates as to whether there should be some adjustment for to the reference year by projecting the incidence of essential items of spending. In some studies, researchers catastrophic payments, using the elasticity of catastrophic have subtracted from consumption food spending3 or payments with respect to the aggregate share of an allowance for food spending6,7 to capture the fact out-of-pocket spending over total consumption based that poorer households have fewer resources to devote on national accounts data. For a remaining set of to non-nutritional needs. Both approaches overlook, 110 countries (accounting for 13·9% of the world’s however, other non-discretionary spending—eg, related population), we did not have a datapoint between 2005 to clothing, shelter, and heating—that in some countries and 2015. For 23 of these 110 countries, we used the is even more important (relatively speaking) than aggregate share of out-of-pocket spending over total food expenditure, including for poor populations.19 consumption to estimate the value of catastrophic Researchers on two studies tried to address this problem, payments in the reference year. Finally, we imputed the but neither study is universally applicable, with one19 incidence of catastrophic payments using the median being better suited to high-income countries and the regional value for the other 87 countries (9·8% of the other20 being better suited to low-income and middle- world’s population). The country estimates for the income countries. In view of these difficulties, we reference year were then aggregated up to the regional used total consumption as our ability-to-pay measure, and global levels to get the number of people experiencing consistent with the official SDG indicator. We do, catastrophic out-of-pocket expenditures. We then calcu­ however, compare our results with data obtained with lated the global and regional rates by expressing these non-food consumption in the denominator (appendix)— numbers as a share of the relevant population, equivalent a definition that gets close to that used in the previous two global studies. 1995–2005 (reference 2000–10 (reference 2005–15 (reference The thresholds we used are the two proposed SDG year 2000) year 2005) year 2010) thresholds: 10% and 25% of total consumption. The Countries Proportion Countries Proportion Countries Proportion 10% threshold is the more common of the two in (n) of global (n) of global (n) of global empirical work to date, used in 41% of studies; only 6% population population population used the 25% threshold (appendix). The 10% threshold is (%) (%) (%) somewhat higher than the typical threshold used in Reference year point 27 38·4% 36 19·9% 54 31·4% national tax systems to ascertain whether out-of-pocket Two points within band 19 6·6% 29 54·0% 13 21·8% medical expenses are large enough to be tax deductible: One point within band 61 38·0% 48 15·5% 34 32·9% in the USA the threshold is 10%, but in Greece and Fitted 15 6·9% 11 0·8% 23 4·1% Switzerland the threshold is 5%, whereas in Canada and Regional median 89 10·1% 87 9·8% 87 9·8% Korea the threshold is just 3%; in some countries, Table 1: Categories of datapoints used for aggregation including Brazil and Colombia, there is no threshold. www.thelancet.com/lancetgh Vol 6 February 2018 e171 Articles to taking a population-weighted average of the relevant our final dataset accounted for 93% of the world’s country rates. population in 2015, with variation across UN regions: Africa (88%), Asia (95%), Europe (89%), Latin America Aggregate correlates of catastrophic spending and the Caribbean (89%), North America (100%), and We used multiple regression to investigate not only Oceania (63%). the partial relation between a country’s incidence of catastrophic out-of-pocket health expenditures but also Ability to pay defined as total consumption or income various macroeconomic indicators and health system In low-income countries it is hard to measure income, characteristics. We included two macroeconomic indica­ in part because many families produce and consume tors: gross domestic product (GDP) per person and the some of their food on a family plot and this does Gini coefficient for income inequality. We also included not show as income.22 Consumption is, therefore, used total health expenditure (THE) as a share of GDP. To more widely; we have used consumption in this capture the overall share of THE that is prepaid, and the Article except for a few middle-income and high-income mix across different prepayment programmes, we countries, for which we have used income in the included the shares of THE spent by social security absence of data for consumption. Ideally, a consumption funds, general government agencies excluding social aggregate should capture consumption across a broad security funds (referred to hereafter as other government range of categories, such as that proposed by the agencies), private insurance schemes, and non-profit Classification of Individual Consumption according to institutions serving households.21 We postulated that, Purpose (COICOP), published by the UN Statistics among public sources, what is likely to matter for Division, including the use value of durables and the catastrophic payment incidence is not the source of value of the flow of services that the household receives finance (ie, taxes, non-tax revenues, social insurance from occupying its dwelling.22,23 We did not attempt to contributions, etc) but rather which agent spends the reconstruct a consumption aggregate for our datapoints, funds and how it operates—eg, the financial incentives which would be a massive undertaking, but rather we it faces, who it covers, the agency’s pool size, the services relied on datasets for which an aggregate already exists. the agency covers, the generosity of its coverage, and the contractual and payment arrangements it has Out-of-pocket spending with providers. Out-of-pocket spending includes not only payments made by the user at the point of use but also cost- Household datasets sharing and informal payments, both in kind and in To measure a country’s incidence of catastrophic cash, but it excludes payments by a third-party payer.21 spending, we required microdata (ie, unit record data) Many household expenditure surveys include questions from nationally representative household surveys on health spending, but, being general surveys, most containing information on out-of-pocket health spending have some shortcomings in terms of identifying out-of- and on total household consumption. We set out to pocket health spending. First, it is sometimes not clear assemble as large a dataset as possible of such surveys. whether the spending reported is gross or net of any We derived the dataset from household surveys available reimbursement by third parties (eg, private insurance to us as of March, 2017. We undertook inventories of the company or government agency), in which case out-of- microdata catalogues of the International Household pocket spending could be over­ estimated. We excluded Survey Network and the World Bank, and of several countries and surveys for which this uncertainty is a house­ hold survey collections. We also searched for house­ problem (eg, France), in case we overestimated the hold surveys online, and obtained microdata from extent to which health spending is a source of financial household surveys used by other researchers.6,7 Through hardship. Second, recall periods are sometimes this process, we identified 1566 potentially suitable inappropriate, particularly in general expend­ iture house­­hold survey datasets, from 155 countries. Of these, surveys, in which the last 3 months and the last 171 were inaccessible and 424 lacked key variables. The 12 months are used frequently, periods that are too long remaining 971 datasets were subject to a quality assurance for items such as outpatient care and medicines. process that entailed comparing consumption per person Multipurpose surveys are better in that spending and the health budget share with World Bank and WHO data are gathered via a health module that varies data, then checking every datapoint and every country’s recall period by type of service.24 Third, variations time series manually (appendix). At the end of this in comprehensiveness probably exist across surveys. confirmation process, we were left with 553 datapoints A review of 100 survey questionnaires found that, in from 133 countries spanning the period 1984–2015. 80% of surveys, questions were asked about spending These data­ points break down across countries (figure 1) on pharmaceutical products, hospital services, medical and collections (appendix). Only one datapoint was services, and paramedical services.24 Nonetheless, it available for 37 of 133 countries; the remaining is difficult to be sure the surveys are equally 96 countries had multiple surveys. The 133 countries in comprehensive. e172 www.thelancet.com/lancetgh Vol 6 February 2018 Articles Both 1996–2005 and 2006–15 Only 2006–15 Only 1996–2005 Only pre-1996 Dataset(s) analysed but discarded Dataset(s) inadequate No datasets identified Figure 1: Data availability for catastrophic health expenditure, by country Data for macroeconomic and health system indicators 10% threshold compared with other countries (which We obtained GDP and THE from the World Bank’s Open could be interpreted as good performance) was mirrored Databases and the Gini coefficient for income from by low incidence at the 25% threshold compared with Milanovic’s All the Ginis (ALG) dataset.25 We obtained other countries, but exceptions were noted. Using non- proportions of THE channelled through social security food consumption in the denominator and setting the schemes, other government agencies, private insurance, threshold at 40% gave a population-unweighted mean and non-profit institutions from WHO’s Global Health catastrophic incidence of 2·1% (SD 2·7) and resulted in Expenditure Database (GHED). We filled gaps in the ALG catastrophic payments being more concentrated in the and GHED datasets by carrying forward the most recent world’s poorest regions—Africa and Asia (appendix). datapoint and carrying backward the oldest datapoint; for This alternative measure correlates less strongly with the countries with data missing completely for the share official SDG measures than they do with each other (rank of THE channelled through social security, private correlations are 0·554 and 0·709). insurance, and non-profit institutions, we assumed they Aggregating across countries, estimates showed that, did not use the financing agency with missing data. in 2010, 808·4 million people incurred catastrophic Further details of data sources are in the appendix. spending at the 10% threshold, equivalent to 11·7% of the world’s population (table 2). At the 25% threshold, Role of the funding source these figures were 179·3 million people and 2·6% of The funders had no role in study design, data collection, the world’s population, and using 40% of non-food con­ data analysis, data interpretation, or writing of the report. sump­tion as the threshold, the figures were 208·2 million The corresponding author had full access to all data in people and 3·0% of the world’s population. Estimates the study and had final responsibility for the decision to for 2010 revealed variations across UN regions, with submit for publication. Latin America and the Caribbean having the highest incidence at the 10% threshold (14·8%), and Oceania Results having the lowest (3·9%). The incidence of catastrophic out-of-pocket payments in Figure 3 shows the average annual change in the inci­ the most recent surveys varied strikingly across countries. dence of catastrophic out-of-pocket payments at the 10% At the 10% threshold, incidence ranged from 0·3% in and 25% thresholds across all available surveys, for Zambia in 2010 to 44·9% in Lebanon in 1999 (figure 2A). 94 countries for which surveys were available for 2 years Mean incidence across countries was 9·2% (SD 7·6) and or more. At the 10% threshold, the average annual median was 7·1% (IQR 3·4–13·4). Incidence was change ranged from –2·7% per year in Congo inevitably lower at the 25% threshold (figure 2B), with (Brazzaville [2005–11]) to 3·3% per year in Armenia mean and median incidences of 1·8% (SD 2·1) and 1·0% (2010–13). In 48 of 94 countries, the incidence of (IQR 0·34–2·5), respectively. The rank correlation catastrophic out-of-pocket spending increased over time. between the two catastrophic payment measures At the 25% threshold, catastrophic payment incidence was 0·877, so for the most part low incidence at the rose in 54% of countries. The population-unweighted www.thelancet.com/lancetgh Vol 6 February 2018 e173 Articles A 7 6 5 6 3 9 5 3 11 10 6 2 4 11 3 1 14 2 4 7 20 43 7 16 2 3 9 9 12 2 9 9 5 13 29 17 16 8 4 5 18 6 15 3 11 14 16 45 18 6 17 5 22 75 16 1 26 27 7 14 3 11 1 17 10 17 3 3 4 10 1 3 6 17 3 28 3 4 11 6 7 1 10 1 11 25 10 15 3 11 1 8 11 3 17 12 1 6 6 15 6 2 5 15 10 4 8 26 3 12 2 0 8 3 1 1 10 9 4 13 33 1 2 15–45% 17 14 10–14% 6–9% 3–5% 0–2% No data B 0.9 1. 0 0.5 0. 7 0.5 1. 2 0.6 0.5 1. 1. 68 0.7 0.5 0.2 1. 4 0.1 1. 6 0.1 0. 1 0.4 1. 1 6.7 0.7 0.3 2. 3 3. 6 0.1 0.3 0.3 0.5 1. 30.7 1. 1 1. 0 4.90.60.8 9.0 0.8 0.8 3. 3 1. 2 4.91. 1 4.0 1. 8 0.3 2. 7 1. 5 4.8 2. 0 2. 4 10. 0 0.1 2. 7 0.9 0.9 3. 8 0.0 3. 9 3. 3 1. 9 4.8 0.3 1. 8 0.1 3. 9 2. 9 4.4 0.3 0. 1 0.4 2. 1 0.0 0.4 0.2 2. 4 0.7 8.9 0.2 0.6 3. 6 1. 4 1. 3 0.0 1. 8 0. 2 0.9 8.9 0.6 3. 60.50. 0 0.2 1. 6 3. 0 0.1 2. 8 2. 6 1. 5 0.0 0.20. 4 0.8 3. 3 0.7 4.2 2. 5 0.4 1. 2 3. 5 0.0 4.5 0.1 0.0 3. 2 0.2 0.3 0.0 2. 0 1. 8 0.5 2. 0 11. 5 0.2 0.1 3·2–11·5% 4.7 1. 9 1·5–3·1% 0·7–1·4% 0·2–0·6% 0·0–0·1% No data Figure 2: Incidence of catastrophic health spending at the 10% (A) and 25% (B) thresholds, latest year median change in catastrophic out-of-pocket payment Counter examples exist, however (figure 3); Tanzania incidence was 0·03% per year (IQR –0·18 to 0·41) for the and Uganda, for example, have achieved quite large 10% threshold and 0·01% per year (–0·05 to 0·07) for the reductions in catastrophic spending at the 10% threshold 25% threshold, whereas the population-weighted figures but not at the 25% level, whereas at the other end of the were 0·45% per year (–0·13 to 1·02) and 0·22% per year chart Bulgaria and Moldova have seen catastrophic (0·00 to 0·31), respectively. The discrepancy in these payment incidence rising at the 10% threshold but not at values indicates that catastrophic payment incidence has the 25% threshold. been falling more slowly or rising more quickly in more The trend in annual average change was more populous countries. The rank correlation between the encouraging if the sample was restricted to 2005 and annual average changes in the two catastrophic payment onwards. The population-unweighted median annual measures was 0·880; thus, for the most part, relative changes in incidence of catastrophic out-of-pocket improvements at the 10% threshold were mirrored payments were –0·07% per year (IQR –0·36 to 0·27) for by relative improvements at the 25% threshold. the 10% threshold and –0·00% per year (–0·08 to 0·08) for e174 www.thelancet.com/lancetgh Vol 6 February 2018 Articles 2000 2005 2010 Proportion of Number of Proportion of Number of Proportion of Number of population (%) people (million) population (%) people (million) population (%) people (million) 10% threshold (total consumption) Global 9·7% 588·5 11·4% 741·3 11·7% 808·4 Africa 8·7% 70·7 10·3% 94·1 11·4% 118·7 Asia 10·4% 381·6 12·2% 479·2 12·8% 531·1 Europe 6·5% 47·4 7·0% 51·2 7·2% 53·2 Latin America and the Caribbean 13·4% 70·5 17·5% 98·3 14·8% 88·3 North America 5·5% 17·2 5·3% 17·4 4·6% 15·6 Oceania 3·5% 1·1 3·4% 1·1 3·9% 1·4 25% threshold (total consumption) Global 1·9% 112·8 2·4% 154·9 2·6% 179·3 Africa 1·5% 12·3 1·9% 17·7 2·5% 25·6 Asia 2·1% 77·1 2·8% 108·7 3·1% 128·7 Europe 0·9% 6·5 1·0% 7·3 1·0% 7·2 Latin America and the Caribbean 2·6% 13·6 3·2% 18·0 2·5% 14·9 North America 1·0% 3·1 0·9% 3·0 0·8% 2·6 Oceania 0·5% 0·1 0·4% 0·1 0·5% 0·2 Table 2: Global estimates of catastrophic spending the 25% threshold; however, the population-weighted Azerbaijan, Canada, and the UK all officially cover 100% figures remained positive at 0·27% per year (–0·08 to 2·01) of their populations automatically with national or and 0·10% per year (–0·00 to 0·57), respectively. The trend regional health services;8,26 yet the incidence of in annual average change was also more encouraging with catastrophic payments was considerably higher in the non-food version of the catastrophic payment indicator: Armenia and Azerbaijan (16% and 8%, respectively, at catastrophic payment incidence increased in only 38% of the 10% threshold) than it was in Canada and the UK countries, whereas the population-unweighted median (3% and 2%, respectively, at the same threshold). The rate of change was –0·05% per year (IQR –0·15 to 0·04); incidence of catastrophic payments also varied between the population-weighted median, however, remained Hungary (7%), South Korea (13%), Montenegro (9%), positive at 0·04% per year (–0·00 to 0·45). and Romania (12%), despite the fact that—in all four Globally, the number of people incurring catastrophic countries—100% of the population is officially covered payments increased between 2000 and 2010, whichever by a national health insurance scheme.8,26 Moreover, threshold was used and whether or not total con­ even though—in these eight countries—arrangements sumption or non-food consumption was used in the and insurance coverage rates have stayed the same in denominator (table 2). At the 10% threshold, the number recent years, incidence of catastrophic payments has of people incurring a catastrophic payment increased not always remained unchanged; indeed, in some cases, from 588·5 million (9·7% of the world’s population) in a clear upward trend was evident (figure 3). The USA is 2000 to 741·3 million (11·4%) in 2005, rising to a counter example: insurance coverage rates stayed 808·4 million (11·7%) in 2010. A similar pattern was largely unchanged over the period 1995–2013;27 yet, the evident in the numbers for the 25% threshold and incidence of catastrophic payments fell. Figure 3 also indicator. The global trend estimates were based on shows that when additional population groups acquired estimates for all countries, including those that had coverage in formal insurance schemes, the incidence of limited trend data or no data. The pattern of a global catastrophic payments did not always change in the increasing incidence of catastrophic payments was, expected direction: in Mexico, Thailand, and Vietnam, however, consistent with the population-weighted catastrophic payment incidence has indeed fallen as the estimates for countries with at least 2 years of data for fraction of the population with insurance coverage catastrophic payments. The incidence of catastrophic has expanded, but this effect has not happened in payments has evolved differently across the various China, Indonesia, or the Philippines.13,14,28–30 In short, UN regions: the global rise in catastrophic payment catastrophic payment incidence cannot be inferred incidence has been driven by increases in Africa and Asia; from the fraction of the population covered by health North and South America—and for some indicators, insurance schemes or public health services. This other regions too—have seen reductions. conclusion is not sensitive to the definition of Incidence of catastrophic spending can vary across catastrophic expenditures used; the same conclusion is countries with similar types of health system. Armenia, reached when using the non-food definition. www.thelancet.com/lancetgh Vol 6 February 2018 e175 Articles Congo (Brazzaville) 2005–11 10% threshold Guinea 2002–12 Pakistan 1991–2010 25% threshold Bolivia 1999–2002 Côte d’Ivoire 1998–2008 Albania 2002–12 Panama 1997–2008 Guatemala 2000–14 Burkina Faso 1998–2009 Cameroon 1996–2014 Paraguay 1996–2001 Tajikistan 1999–2007 Uganda 1996–2002 Zambia 2004–10 Ukraine 2002–13 Bosnia and Herzegovina 2001–11 Rwanda 2000–10 Vietnam 1992–2014 South Africa 1995–2010 Turkey 2002–12 Colombia 1997–2010 Tanzania 2008–12 Thailand 1994–2010 Mexico 1984–2012 Croatia 1998–2010 Peru 2000–15 Mozambique 2002–08 Slovakia 2004–10 USA 1995–2013 Ethiopia 1999–2004 Sri Lanka 1995–2009 Cape Verde 2001–07 Bangladesh 2000–10 Ghana 1991–2005 Finland 1998–2010 Malaysia 1993–2004 Kyrgyzstan 2005–11 Kazakhstan 1996–2013 Madagascar 2001–05 Greece 1998–2010 Mali 2001–06 Slovenia 1999–2012 Laos 2002–07 Kenya 1997–2005 UK 1995–2013 Malawi 1997–2010 Israel 1997–2012 Jamaica 1991–2004 Denmark 1997–2010 Italy 2001–10 Serbia 2003–10 Niger 2005–11 Mongolia 2002–12 Belarus 1998–2012 Norway 1996–98 Russia 1997–2014 Spain 1985–2010 Indonesia 2001–15 Czech Republic 1999–2010 Montenegro 2005–14 Luxembourg 1998–2010 Poland 1998–2012 Philippines 1997–2015 Switzerland 2000–04 Hungary 1998–2010 Lithuania 1998–2010 South Korea 1999–2008 Tunisia 1995–2010 Kosovo 2003–11 Ireland 1999–2010 Azerbaijan 2002–05 Portugal 1990–2010 Costa Rica 1992–2012 India 2004–11 Belgium 1997–2010 Romania 1998–2012 Iran 2005–13 Latvia 2002–06 Macedonia 1996–2008 Estonia 1995–2010 Nicaragua 1993–2014 Moldova 1999–2013 Bulgaria 1997–2010 Argentina 1996–2004 Jordan 2002–06 China 1995–2007 Egypt 1997–2012 Nepal 1995–2010 Georgia 1997–2013 Yemen 1998–2005 Nigeria 2003–09 Morocco 1998–2006 Chile 1997–2006 Armenia 2010–13 –4 –3 –2 –1 0 1 2 3 4 Average annual percentage point change Figure 3: Annual percentage point change in incidence of catastrophic health spending Data were calculated using all surveys available for the country in question by regressing catastrophic expenditure rate on year of survey; the number shown is the coefficient from this regression. Surveys span the period 1984–2015, with a median year of 2005 (IQR 2001–2009). The median first year was 1998 and the median last year was 2010. e176 www.thelancet.com/lancetgh Vol 6 February 2018 Articles 10% threshold 25% threshold 25th percentile Median 75th percentile 25th percentile Median 75th percentile GDP per person, 2011 (intl$) 0·205 (p=0·04) 0·216 (p=0·02) 0·230 (p=0·01) 0·060 (p=0·05) 0·064 (p=0·03) 0·069 (p=0·02) Gini index of income inequality 0·088 (p=0·04) 0·112 (p=0·01) 0·147 (p=0·02) 0·028 (p=0·01) 0·034 (p=0·0006) 0·042 (p=0·01) THE (% of GDP) 0·728 (p=0·00) 0·682 (p=0·01) 0·616 (p=0·01) 0·165 (p=0·08) 0·161 (p=0·08) 0·154 (p=0·10) Social security (% of THE) –0·061 (p=0·07) –0·120 (p=0·0004) –0·206 (p=0·0002) –0·017 (p=0·04) –0·035 (p=0·0010) –0·062 (p=0·0018) Other government agencies (% of THE) –0·092 (p=0·01) –0·158 (p<0·0001) –0·252 (p<0·0001) –0·016 (p=0·04) –0·037 (p=0·0005) –0·066 (p=0·0009) Private insurance (% of THE) 0·202 (p=0·35) 0·077 (p=0·66) –0·103 (p=0·45) 0·054 (p=0·41) 0·019 (p=0·71) –0·031 (p=0·42) Non-profit institutions (% of THE) –0·106 (p=0·11) –0·149 (p=0·10) –0·209 (p=0·20) –0·028 (p=0·11) –0·030 (p=0·24) –0·034 (p=0·50) Observations (n) 508 508 508 508 508 508 Social security=other government 0·249 0·084 0·006 0·814 0·784 0·225 agencies (probability) GDP=gross domestic product. THE=total health expenditure. Table 3: Multiple regressions showing marginal effects of macroeconomic and health systems characteristics on catastrophic spending incidence at different income levels per person By contrast, incidence of catastrophic health spending at insurance coverage and provider incentives and, hence, both the 10% and 25% thresholds was significantly and the possibility that acquisition of coverage could leave positively associated with GDP per person (table 3). people vulnerable to providers taking the opportunity to Income inequality also had a positive partial association generate more income by delivering and charging for with catastrophic spending at all income levels, which additional services, not all of which might be medically became stronger at higher income levels. A positive partial necessary. association was noted between catastrophic spending and Our regression results are—by their nature— the share of GDP spent on health, but this association associations and do not necessarily reflect causation. The became weaker at higher income levels. A negative partial positive partial relation between catastrophic spending association was recorded between catastrophic spending and the share of GDP spent on health could reflect, as incidence and the share of THE channelled through social previously postulated,7 greater service availability, more security funds and other government agencies. These use of expensive technology, and higher prices, all effects were stronger at higher income levels per person. of which are likely to be correlated positively with The results suggest that an increase in the share of THE catastrophic payment incidence; the relation also suggests channelled through social security schemes might offer that simply spending more on health is not sufficient to somewhat less financial protection than an increase in the provide financial protection. The negative correlation share of THE channelled through other government between catastrophic payment incidence and the share of agencies. By contrast with our results for govern­ health expenditure channelled through social security ment agencies, no evidence was found to suggest that funds and other government agencies, but the absence of health spending channelled through private insurance such an association in the case of private insurance and non-profit institutions provides financial protection. and non-profit institutions, suggests an important role for public financial protection arrangements (funded Discussion by taxes and mandatory insurance contributions) and Our data show substantial variation across countries in a questionable role for private ones (funded through the incidence of catastrophic spending in the most recent voluntary premiums and contributions). The finding that household survey, ranging from less than 1% to more catastrophic payment incidence is associated less strongly than 40%. In the latest surveys (median year 2010), the with spending through social security funds than with median incidence of catastrophic spending was 7% at the spending through other government agencies could 10% threshold, 1% at the 25% threshold, and 2% with reflect shallower coverage in social insurance schemes the 40% threshold of non-food. We found no evidence and higher inpatient admission rates and costs.8,31 of any link between catastrophic payment incidence and Our findings on trends are mixed. The proportion of the share of the population that is supposed to have countries with rising incidence of catastrophic payments health coverage, and we noted changes over time in is 50% using the two SDG indicators and less than 50% catastrophic payment incidence in countries, even during with the version using non-food consumption. However, periods when health coverage rates had not changed. adjusting for population size produces a different picture. What coverage rates miss—and catastrophic payment The population-weighted median annual rate of change incidence captures—are the extent of de jure and (more of catastrophic payment incidence is positive whatever importantly) de facto coverage for different services. indicator is chosen. At the global level, we estimate that Coverage rates also miss the interaction between the number of people with catastrophic spending at the www.thelancet.com/lancetgh Vol 6 February 2018 e177 Articles 10% threshold increased from 589 million (10% of the catastrophic spending might simply reflect a situation world’s population) in 2000 to 741 million (11%) in 2005, in which only a few people get the health care they and continued to increase, albeit at a slower rate, need because facilities are few or inadequate; data for to 808 million (12%) in 2010. Such an increase is both sides of the UHC coin need to be examined also noted when using other thresholds (ie, 25%), but simultaneously. Finally, even though we have more at a lower rate, and between 2000 and 2005 with than four and a half times as many datapoints as the other definitions of ability to pay (non-food with a previous global study, there are still gaps—some 40% threshold), but not between 2005 and 2010. countries are absent, some have only one datapoint, and Our study has several limitations. First, our data come some are quite old. As such, our global estimates are from various surveys. We have tried to minimise the produced using a combination of survey-based data­ risks associated with heterogeneity by focusing when points, interpolated and extrapolated datapoints based on possible on one collection for a given country, making econometric modelling, and imputation using regional use of ex post-harmonised datasets, and cross-checking medians. Therefore, we did not attempt to conduct basic summary statistics from our surveys with other inference around our global estimates. Uncertainty sources. However, despite our efforts to minimise the around our estimates comes from both sampling error heterogeneity of surveys within countries, differences in around the survey-based datapoints and non-sampling definition probably remain, including in the way ability error associated with the modelled estimates used to to pay is measured: mostly, ability to pay is measured align the incidence of catastrophic payments to a specific using consumption, but in some surveys it is measured reference year. using income. This discrepancy is one reason we have In conclusion, while catastrophic payment incidence not reported results on inequality in catastrophic has been falling in around half of countries using the spending, which is highly sensitive to the choice. Second, SDG indicators, and in more than half of countries using we have not looked at persistency of the large out-of- the non-food version of the catastrophic payment pocket expenditures over time within households. Panel indicator, the population-weighted median annual rate of data provide the best opportunity to assess this issue,32 change of catastrophic payment incidence has been rising but availability of such data is very limited. With cross- whatever indicator is chosen. At the global level, we sectional data, richer information on health status of estimate that 808 million people (12% of the world’s household members, and some assumptions about the population) incurred catastrophic health spending at the variability of health expenditures over time faced by every 10% threshold. This figure is higher than it was in 2000 household, measuring exposure to medical expenditure (599 million [10%]) and in 2005 (741 million [11%]). The risk is possible,33 but such information is typically not incidence of catastrophic payments varies considerably available in the datasets we are using. Third, we do not across countries. This variation does not reflect the include the indirect costs associated with care-seeking share of the population that is supposed to be covered (eg, transportation costs) when estimating financial by health insurance or national or subnational health hardship, which can represent a substantial burden. services, making catastrophic payments an un­informative Surveys without a particular focus on health-seeking indicator of financial protection and pointing to the need behaviour (most of our surveys are household budget to look beyond it when designing health system reforms surveys or household income and expenditure surveys), aimed at accelerating progress towards UHC. Greater use do not have information on cost of transportation related of prepayment, particularly through social security funds to utilisation of health services. Fourth, care-seeking also and other government agencies, is likely to be key—not has an opportunity cost beyond any monetary price—eg, merely covering more people but covering a larger share income losses, assets depletion, and indebtedness. Our of total health spending. datasets do not allow us to capture these costs, so we are Contributors not able—as would be possible with a richer dataset—to AW, GF, JH, M-FS, KC, KvW, and PE contributed to the literature review. adjust measures of financial protection to disentangle PE, GF and AW screened datapoints, analysed the dataset, and wrote the first draft of the manuscript. All authors contributed to assembly of the the short-term and long-term outcomes of coping with dataset and writing of the manuscript. health-care cost. Findings of some studies with such Declaration of interests richer datasets34 suggest that people might be able to We declare no competing interests. cope with the cost of care but not with income losses, Acknowledgments but evidence is scarce. Fifth, we do not have the space We acknowledge financial support for this work from the Rockefeller to discuss in detail the conceptual underpinning, Foundation and the Ministry of Health of Japan (to KvW, GF, and JH at advantages, and disadvantages of the different definitions WHO). We also acknowledge funding from the UK Department for of ability to pay used in this Article.19,20 Nonetheless, International Development (DFID) under the programme for improving countries’ health financing systems to accelerate progress towards we find a regular pattern at the global level for all universal health coverage (to KvW, GF, JH at WHO). KC was supported by three measures—ie, an increase in the incidence of a grant from the Swiss School of Public Health. We thank Tessa Edejer, catastrophic expenditures. Sixth, our analysis shows Annie Chu, Camilo Cid, Dan Hogan, Grace Kabaniha, Awad Mataria, merely one dimension of UHC. A low incidence of Claudia Pescetto, Maria Pena, Lluis Vinals Torres, Hui Wang, Ke Xu, e178 www.thelancet.com/lancetgh Vol 6 February 2018 Articles and Agnes Soucat for advice and for facilitating the country consultation 14 Limwattananon S, Neelsen S, O’Donnell O, et al. Universal coverage on universal health coverage indicators undertaken by WHO, and we with supply-side reform: the impact on medical expenditure risk and thank the nominated focal points in countries who responded to the utilization in Thailand. J Public Econ 2015; 121: 79–94. WHO consultation. We also thank Caryn Bredenkamp, Tania 15 van Doorslaer E, O’Donnell O, Rannan-Eliya RP, et al. Catastrophic Dmytraczenko, Olivier Dupriez, Marco Ranzani, Aparnaa Somanathan, payments for health care in Asia. Health Econ 2007; 16: 1159–84. Ajay Tandon, and Joao Pedro Wagner de Azevedo for assistance with 16 Flores G, Krishnakumar J, O’Donnell O, van Doorslaer E. microdata; Qinghua Zhao for help with PovcalNet; Caryn Bredenkamp, Coping with health-care costs: implications for the measurement Tania Dmytraczenko, Timothy G Evans, and Owen Smith for advice; of catastrophic expenditures and poverty. Health Econ 2008; Joe Kutzin, Aparnaa Somanathan, and Ajay Tandon for comments on an 17: 1393–412. earlier draft; and participants at a meeting on universal health coverage 17 Wagstaff A, Lindelow M. Are health shocks different? Evidence held at the Rockefeller Foundation’s Bellagio Center in November, 2015. from a multishock survey in Laos. Health Econ 2014; 23: 706–18. The findings, interpretations, and conclusions expressed in this Article 18 Demirguc-Kunt A, Klapper L, Singer D, Van Oudheusden P. are entirely those of the authors and do not necessarily represent the The Global Findex Database 2014: measuring financial inclusion around the world—policy research working paper 7255. April, 2015. views of the World Bank, its Executive Directors, the governments of the http://documents.worldbank.org/curated/en/187761468179367706/ countries they represent, or of WHO. The designations employed and the pdf/WPS7255.pdf (accessed Oct 23, 2017). presentation of the material in this publication do not imply the 19 Thomson S, Evetovits T, Cylus J, Jakab M. Monitoring financial expression of any opinion whatsoever on the part of the World Bank or protection to assess progress towards universal health coverage in the WHO concerning the legal status of any country, territory, city, or area Europe. September, 2016. http://www.euro.who.int/__data/assets/ or of its authorities, or concerning the delimitation of its frontiers or pdf_file/0005/317543/10-Research-article-monitoring-financial- boundaries. Dotted and dashed lines on maps represent approximate protection-assess-progress-universal-health-coverage-Europe. border lines for which there may not yet be full agreement. Our estimates pdf?ua=1 (accessed Oct 23, 2017). are not based necessarily on the latest available data to the country, and 20 Wagstaff A, Eozenou PH-V. CATA meets IMPOV: a unified approach might also differ from estimates produced by the country itself. to measuring financial protection in health—policy research working paper 6861. May, 2014. http://documents.worldbank.org/curated/ References en/114091468149393397/pdf/WPS6861.pdf (accessed Oct 23, 2017). 1 Hogan DR, Stevens GA, Hosseinpoor AR, Boerma T. Monitoring 21 Organisation for Economic Co-operation Development. A system of universal health coverage within the Sustainable Development health accounts 2011. Paris: Organisation for Economic Goals: development and baseline data for an index of essential Co-operation and Development, 2011. health services. Lancet Glob Health 2017; published online Dec 13. http://dx.doi.org/10.1016/S2214-109X(17)30472-2. 22 Deaton A, Zaidi S. Guidelines for constructing consumption aggregates for welfare analysis: living standards measurement 2 Boerma T, Eozenou P, Evans D, Evans T, Kieny M-P, Wagstaff A. study working paper, no 135. Washington: World Bank, 2002. Monitoring progress towards universal health coverage at country and global levels. PLoS Med 2014; 11: e1001731. 23 Balcazar CF, Ceriani L, Olivieri S, Ranzani M. Rent imputation for welfare measurement: a review of methodologies and empirical 3 Wagstaff A, van Doorslaer E. Catastrophe and impoverishment in findings—policy research working paper 7103. November, 2014. paying for health care: with applications to Vietnam 1993–1998. http://documents.worldbank.org/curated/en/313651468152726645/ Health Econ 2003; 12: 921–34. pdf/WPS7103.pdf (accessed Oct 23, 2017). 4 Wagstaff A, Flores G, Smitz M-F, Hsu J, Chepynoga K, Eozenou P. 24 INDEPTH Network. Household out-of-pocket health expenditures Progress on impoverishing health spending in 122 countries: tracking for disease specific health account and universal health a retrospective observational study. Lancet Glob Health 2017; coverage measure: developing household OOP measurement published online Dec 13. http://dx.doi.org/10.1016/S2214- methodology. Accra: INDEPTH Resource & Training Centre, 2017. 109X(17)30486-2. 25 Milanovic B. Description of All the Ginis dataset. New York: 5 van Doorslaer E, O’Donnell O, Rannan-Eliya RP, et al. Effect of Luxembourg Income Study, 2016. payments for health care on poverty estimates in 11 countries in Asia: an analysis of household survey data. Lancet 2006; 26 Smith O, Nguyen SN. Getting better: improving health system 368: 1357–64. outcomes in Europe and Central Asia—Europe and Central Asia Reports. Washington: World Bank, 2013. 6 Xu K, Evans DB, Kawabata K, Zeramdini R, Klavus J, Murray CJL. Household catastrophic health expenditure: a multicountry 27 Obama B. United states health care reform: progress to date and analysis. Lancet 2003; 362: 111–17. next steps. JAMA 2016; 316: 525–32. 7 Xu K, Evans DB, Carrin G, Aguilar-Rivera AM, Musgrove P, 28 Wagstaff A, Lindelow M, Jun G, Ling X, Juncheng Q. Evans T. Protecting households from catastrophic health spending. Extending health insurance to the rural population: an impact Health Aff (Millwood) 2007; 26: 972–83. evaluation of China’s new cooperative medical scheme. J Health Econ 2009; 28: 1–19. 8 Paris V, Devaux M, Wei L. Health systems institutional characteristics. Paris: OECD Publishing, 2010. 29 Wagstaff A. Estimating health insurance impacts under unobserved heterogeneity: the case of Vietnam’s health care fund for the poor. 9 McIntyre D, McKee M, Balabanova D, Atim C, Reddy KS, Health Econ 2010; 19: 189–208. Patcharanarumol W. Open letter on the SDGs: a robust measure for universal health coverage is essential. Lancet 2016; 388: 2871–72. 30 Sparrow R, Suryahadi A, Widyanti W. Social health insurance for the poor: targeting and impact of Indonesia’s Askeskin programme. 10 Berki S. A look at catastrophic medical expenses and the poor. Soc Sci Med 2013; 96: 264–71. Health Aff (Millwood) 1986; 5: 138–45. 31 Wagstaff A, Moreno-Serra R. Europe and central Asia’s great 11 Ranson M. Reduction of catastrophic health care expenditures by a post-communist social health insurance experiment: aggregate community-based health insurance scheme in Gujarat, India: current impacts on health sector outcomes. J Health Econ 2009; 28: 322–40. experiences and challenges. Bull World Health Organ 2002; 80: 613–21. 32 Feenberg D, Skinner J. The risk and duration of catastrophic health 12 Ekman B. Catastrophic health payments and health insurance: care expenditures. Rev Econ Stat 1994; 76: 633–47. some counterintuitive evidence from one low-income country. Health Policy 2007; 83: 304–13. 33 Flores G, O’Donnell O. Catastrophic medical expenditure risk. J Health Econ 2016; 46: 1–15. 13 King G, Gakidou E, Imai K, et al. Public policy for the poor? A randomised assessment of the Mexican universal health 34 Gertler P, Gruber J. Insuring consumption against illness. insurance programme. Lancet 2009; 373: 1447–54. Am Econ Rev 2002; 92: 51–76. www.thelancet.com/lancetgh Vol 6 February 2018 e179