Inequalities in Health Care in China : Evidence from the China Health and Nutrition Survey

Transcription

1 Inequalities in Health Care in China : Evidence from the China Health and Nutrition Survey Yifei Yan 1, Jiwei Qian, PhD 2, & Xun Wu, PhD 3,1 * 1 Lee Kuan Yew School of Public Policy, National University of Singapore 2 East Asia Institute, National University of Singapore 3 Division of Social Science and Division of Environment, Hong Kong University of Science and Technology Abstract China s health care system has undergone significant transformation and policy shifts since 1990s. However, its impact on inequalities in health care over time has not been well understood. Our study examines the trend of inequalities in health care from 1991 to 2011, using micro-level panel data from China Health and Nutrition Survey. We categorize the eight waves of respondents according to different socioeconomic status and examined the trend of inequality in terms of insurance coverage, service utilization and out-of-pocket expenditure. Our results reveal that such inequalities were shaped by structural factors like geographical location, profession, employment status, type of work unit, and income level. Although recent reform on strengthening health insurance appear to have achieved a high level of equalities in insurance coverage, inequalities have persisted, or worsened, in aspects like health service utilization and expenditures. Extensive enrolment in health insurance may thus be achieved at the expense of access to health care as well as increasing structural inequalities, undermining the effectiveness of the reform so far. And to fully understand China s health inequalities, one may have to look at structural factors beyond the health sector, such as geographical location, type of profession, nature of work unit, or income level. Key words: health inequality, equity, socioeconomic status, China, CHNS 1

2 INTRODUCTION Although inequalities in health care in China were among the core factors that have prompted the dramatic transformation of the country s health system since the 1990s [1,2,3,4,5,6], some critical aspects of the dynamics of the inequalities have not been well understood. For example: What are the main manifestations of inequalities in health care in China? To what extent have inequalities in health care been shaped by factors outside the health sector? How have inequalities changed over time? What impacts do health policy reforms have on inequalities in health care? The answers to these questions may have important implications for the design of policy interventions to further improve system performance. Efforts at comprehensive analysis of inequalities in health care face several methodological challenges. First, most studies to date have been based on analysis of cross-sectional data collected from a particular point in time, or on comparison of data gathered from a relatively short time period. Both strategies render it difficult if not impossible to compare results from different studies for understanding the overall trend across a longer period of time, given the fast-changing health sector in China since the 1990s. Differences in questionnaire design and sampling methodologies also make it difficult to reconcile differences across various existing studies. And most studies focus on residential status (urban versus rural) and income level in studying inequalities in health care [7,8], whereas other potential dimensions of inequalities, such as employment status, occupation, and education, have not yet been examined sufficiently. The study reported here examined inequalities in health care in China over the two decades , using data from the China Health and Nutrition Survey (CHNS). The dataset is a collaborative project between the Carolina Population Center at the University of North Carolina at Chapel Hill and 2

3 the National Institute of Nutrition and Food Safety at the Chinese Center for Disease Control and Prevention [9,10]. CHNS uses a multi-stage random clustering process to draw a sample of about 4, 400 households in nine provinces. It is well suited for the research for several reasons. Since 1989, nine waves of CHNS have been conducted, spanning more than two decades during which key health policy reforms were introduced. While only 14% of total individual respondents and 7% of households have participated entire nine waves of survey (authors calculation) who are subjected to ageing, drop-out and other issues, it still offers a unique opportunity to study the changes in inequalities over a long period of time. CHNS survey questionnaires and sampling methodologies remained unchanged throughout that period, providing a high level of consistency and compatibility across data collected in different years. CHNS also used the same localities through all its waves of data collection. Perhaps most importantly, the CHNS surveys contain not only questions about health insurance, but also questions about utilization of health services, health expenditures, and out of pocket (OOP) payments. It also supplies information about respondents socioeconomic characteristics, thus providing opportunities to capture changes in inequalities from multiple perspectives. To our knowledge, the present study is the first to employ micro-level panel data systematically to assess inequalities in health care in China over a long period of time. Our results reveal that significant structural inequalities already existed in the early 1990s, when China s health policy reforms began. While the waves of reforms since then appear to have been effective in some aspects, achieving a high level of equalities in health insurance coverage, other measures such as utilization of health services and control of health expenditures suggest that structural inequalities have persisted or even worsened in some circumstances. Our findings lead to several critical insights into the design of health policy reforms aimed at achieving universal health care. 3

4 DATA AND METHODOLOGY The data analyzed in the study, unless otherwise mentioned, come from the eight waves of CHNS beginning in 1989 whose results were reported from 1991 to We excluded children and students from the original CHNS data sets, because their patterns of health-care financing and utilization- aspects that we focus on in this paper- are quite different from that of adults. For example, whether students enroll in an insurance package is often decided by schools rather than students. In this case, socioeconomic conditions for students are not very useful in understanding the health financing conditions for students. Respondents from the three provinces (Beijing, Shanghai, and Chongqing) added to the sample in the most recent wave (2011) were also excluded due to limited observations. Overall, 75.9% of the observations from the original dataset are retained in our reconstructed dataset. We also excluded from the survey data respondents aged 65 years and older, because their proportion to the population rose from 1991 to This trend creates differences across data from succeeding waves of the CHNS surveys as well introducing a potential bias, as the elderly incur significantly higher health care expenditures than other age groups [11]. To examine the influence of key structural factors on inequalities in health, we organized our sample of CHNS survey respondents according to the following six categories and their subcategories: Geographic Location (GL): Consistent with other study using the same dataset[12], here respondents were categorized as urban or rural according to where they were located at the time they participated in a CHNS survey; Geographic Region (GR): Respondents were also given a designation by region of residence: 4

5 coastal provinces (CS): Liaoning, Jiangsu, and Shandong; or inland provinces (IN): Heilongjiang, Henan, Hubei, Hunan, Guangxi, and Guizhou; Major Occupation (OCC): Major occupation of each respondent was categorized as professional (PRO) including advanced professionals such as doctors, professors, etc. as well as midlevel professionals such as nurses, teachers, etc.; and administrators or managers; farmers (FM), including farmers, fishermen, and hunters; and workers and staff (WS), including office staff such as secretaries; skilled and unskilled workers; drivers; and workers in the service sector such as chefs, barbers etc.; Urban Work Units (UWU): Respondents located in urban areas were further categorized as employees of state-related entities, including government, state organs and institutions;, employees of state-owned enterprises (SOEs); and employees of non state entities, including big and small collectives and private enterprises and foreign-invested or -collaborated companies (san zi qi ye). These three groups are also referred to as government employees, SOE employees, and non state employees ; Urban Employment Status (UES): Respondents located in urban areas were further categorized into urban employed, urban informal sectors, and unemployed/ urban retired. Among the three subcategories, those in urban informal sectors included the self-employed, temporary workers, and household workers. We also identified as migrant workers those who answered that they were working in urban areas but had rural hukou. (As the hukou data is only available since 1993, the subcategory of migrant workers is accordingly only relevant since that wave.) Household Income (INC): Respondents were also divided into five groups (the first/bottom, second, third, fourth and fifth/top quintiles respectively) on the basis of per capita household 5

6 income. This piece of data is constructed in the original dataset by the dataset provider. Descriptive statistics for these six categories and their subcategories in the sample can be found in Appendix Table I. The main outcome variables chosen for analysis were (1) health insurance coverage (enrollment), (2) health-seeking behavior, (3) health care expenditure for outpatient treatment, and (4) out-of-pocket (OOP) payments for outpatient treatment. Note that only outpatient treatment was included for the latter two outcomes. While many social insurance plans in China (e.g. NCMS) attempt to protect people from catastrophic health expenditure, which mainly is associated with inpatient health care, we were not able to get sufficient observations for inpatient care while inpatient expenditure also varies greatly. However, as will be shown, even just considering outpatient OOP payment, its proportion in individual income is already quite high especially among low-income groups. Health insurance coverage was measured by the percentage of the respondents who reported being enrolled in health insurance of any kind. Appendix Table II lists different health insurance schemes mentioned in the eight waves of CHNS, Health-seeking behavior was based on respondents answers to a question in the CHNS survey: What did you do when you fell ill? The choices given in the survey were non-treatment, self-treatment, visiting local health workers, and visiting doctors. As the distinction between the last two was not entirely clear-cut in the survey, we collapsed them into one category, visiting health workers/doctors. This gives us a more general depiction of health utilization that minimizes the bias from the situation that the respondents confuse between local health workers and doctors. Note, however, that positive responses to this question are associated with both outpatient and inpatient treatment. 6

7 Overall outpatient expenditure was calculated from health expenditure per episode of outpatient treatment. Respondents to the adult CHNS questionnaire were asked if they had experienced any particular type(s) of illness during the last four weeks [before the survey]. If yes, and if they received outpatient treatment for the illness, they were then asked how much they had spent during this treatment or so far. (In a different question, respondents who had disease/illness during the episode were asked How much money did you spend for the illness or injury? But this could include self-treatment as well, and thus does not fit our scope here.) Out-of-the-pocket (OOP) payment was the amount that each respondent who received medical treatment paid personally for each episode of outpatient treatment, calculated by subtracting the amount of reported reimbursement from health insurance from the overall outpatient expenditure. All figures stated here are inflation-adjusted using CPI data extracted from Statistical Yearbook of China for respective waves of CHNS. They are all reported in 2011 prices. ANALYTICAL RESULTS Health Insurance: Towards Universal Coverage The trend of health insurance coverage is reported in Appendix Table III. Enrollment in health insurance across the survey sample declined gradually from 31.4% of respondents in 1991 to 20.5% in 2000, after which it started to rise, and by 2011, the vast majority of respondents (94.7%) were covered by some form of health insurance. Free medical insurance (FI, including the Government Insurance Scheme and Worker Labor Insurance Scheme) played a key role in the early 1990s, but its role has since declined. In comparison, social health insurance schemes such as the Urban Employee-based Basic Medical Insurance scheme (UEBMI) and New Cooperative Medical Scheme (NCMS) 7

8 contributed greatly to the rise in coverage. TABLE 1 presents health insurance coverage rates for different population subcategories. While the pattern of changes observed for these subcategorizes appears to resemble the overall trend for all respondents in the sample combined, clear manifestations of inequalities along almost all subcategories are evident until Insurance enrollment was consistently higher in urban than in rural areas during the 1990s and through much of the 2000s. A similar gap existed between coastal and inland provinces. The biggest difference was found between professionals (79.7%) and farmers (4.9%) in Within urban work units, government employees had the highest coverage rate (87.8%) among all population subcategories in 1991, followed by SOE employees (86.8%), while only 8.1% of non state employees were enrolled. The subsequent changes since 1991, however, display different patterns. Enrollment of non-state employees consistently rose, while that for both government and SOE employees declined until 2000, with the latter declining faster of the two, which may be attributable to the welfare retrenchment accompanying SOE reforms in China during that period. The gaps among different types of urban employment status were only slightly smaller during the same period. Inequalities in health insurance enrollment were reduced considerably after Even in the group with the lowest enrollment rate, the urban unemployed and those in urban informal sectors, nearly 84% were covered by In addition, farmers, whose health insurance coverage rate was lowest in 1993 (2.2%), had attained the highest rate among all groups by 2011 (98.1%). [TABLE 1 ABOUT HERE] Health insurance participation was found to be highly correlated with income level of respondents for much of the two decades in the study, and the pattern was quite consistent: the higher the income, the higher the coverage rate (Appendix Table IV). The same pattern is observed for both urban and 8

9 rural areas. The gaps among income groups closed rapidly after 2006, consistent with findings of other studies based on different data sets [13], although a sizable difference still existed between the rich and the poor in urban areas, even in the most recent wave of the CHNS survey (2011). Health-seeking Patterns: Declining Service Utilization The impressive progress made in extent of health insurance coverage, however, has not led to proportionate improvement in access to health services (TABLE 2), although in other studies that distinguish between out-patient and in-patient service utilization do find increase in the latter type and/or correlation between insurance and utilization [13,14]. In our sample, the percentage of respondents who saw doctors when falling ill declined after 1991, although slight upward trends can be observed between 2004 and 2011 across almost all segments of population. While the reimbursement rate of a particular insurance type may have influenced health-seeking decisions [15], this declining tendency to visit doctors or local health workers when ill does not necessarily imply an increase in non-treatment. Instead, an increasing number of respondents may have opted for self-treatment. [TABLE 2 ABOUT HERE] Contrary to a common perception that rural residents have more restricted access to health services [2], health care utilization in our sample was higher in rural areas than in urban areas for all waves of the CHNS survey except This might be explained by the increasingly higher OOP payments for urban residents (discussed below). In a similar vein, inland provinces saw higher service utilization rates after 2000 than did coastal provinces, although the latter are widely perceived as more advanced and wealthier areas: In terms of mean individual income of respondents, we calculated from the CHNS data that the figure for coastal provinces is around 1.1 to 1.3 times as that for inland 9

10 provinces, for all waves of the CHNS survey. Professionals showed a dramatic increase in self-treatment (by 35.5 percentage points) accompanied by sharp decline in health care utilization (from above 80% of respondents in 1993 to less than 40% in 2011). As of 2011, professionals health care utilization rate was nearly 30 percentage points lower than that of farmers and more than 20 percentage points lower than that of workers and staff. Government and SOE employees similarly showed a substantial decline in service utilization rate along with rising self-treatment. In recent years, the utilization rate for urban employed fell lower than those for the urban informal sector and urban unemployed, the retired, and even migrant workers. A similar pattern can be seen in the analysis based on income groups (TABLE 3). For each quintile, in more recent years within the study period the service utilization rate was higher for rural respondents than for their urban counterparts in the same quintile. Self-treatment was more common for urban respondents than for rural residents in most of the waves of the CHNS survey used in the study. In sum, the level of inequalities in utilization of health care services appears to have declined from 1991 to The gaps in non-treatment between urban and rural, and coastal and inland, also declined. Ironically, however, this general reduction in inequalities seems to have been achieved via declining utilization of services. In particular, the decline in inequalities in utilization of health services occurred in tandem with a sharper decline in utilization rates among some groups in the population (previously more privileged groups) than others (previously less privileged ones). For example, the percentage of respondents who reported seeing doctors decreased from 82.4% to 53.0% among urban residents over the survey interval, and from 84.3% to 39.6% for professionals, while the decline for responding rural 10

11 residents was notably more moderate, from 75.6 to 66.9%, and for responding farmers from 75.1% to 68.1%, over the same period of time. [TABLE 3 ABOUT HERE] Outpatient Health Expenditure and Out-of-Pocket Payments: A Rising Burden One explanation for the declining utilization of the health services despite wider coverage of health insurance might be the steady escalation of health expenditure, and in particular OOP payments, over the survey years. TABLE 4 shows that all population subcategories in our sample experienced significant cost escalation in health expenditure and OOP payments from 1991 through [TABLE 4 ABOUT HERE] The escalation in outpatient OOP payments was more pronounced for urban and coastal respondents, and for government and SOE employees, than for their rural and inland counterparts or for non state employees. This may in part explain the more steeply declining rates in service utilization among these groups that was noted above. The burden of outpatient OOP payment as percentage of overall outpatient expenditure rose before and declined only slightly afterward, remaining above 80% for most subgroups. Among occupational subcategories, not surprisingly, farmers paid almost entirely OOP in all survey waves until 2009, given their poor insurance participation prior to the introduction of NCMS. And although professionals had lower OOP payments as a percentage of total health service expenditure than did farmers and workers/staff, professionals OOP payments in absolute terms increased substantially between 1991 and Urban unemployed and those in informal sectors persistently incurred a higher burden of OOP 11

12 payment as a percentage of total outpatient expenditure than did the urban employed, although the differential narrowed as the burden of payment increased for the urban employed. [TABLE 5 ABOUT HERE] A high level of inequalities can be seen among different income groups. In comparison to the richest respondents, the poorest residents in both urban and rural areas not only paid a higher share of outpatient expenditure out of pocket but also experienced disproportionately higher levels of financial burden as measured by percentage of OOP payment to income, across all waves of the CHNS survey included in the study. Results for most waves show that the poor had to pay a substantial portion of their annual income for health care when they fell ill, the highest point being nearly 10% in 2004 for both urban and rural areas. Inequality became more entrenched over time, especially in urban areas: while the outpatient OOP burden as a fraction of household income declined for the top two quintiles, it increased for the bottom three quintiles. In addition, whereas the richest consistently spent less than 0.5% of their annual income on OOP payments, for the poorest in rural areas, OOP payments doubled as a percentage of income (from 3.4% to 6.8%) over the two decades. While studies using other datasets have found that inpatient cost falls as respondent income increases [16] or that NCMS insurance decreased patients out-of-pocket spending for higher-cost health services [17], outpatient care remains a significant portion of health expenditure for the poor. DISCUSSION Our findings from research based on CHNS surveys from 1991 through 2011 highlight some key characteristics of inequalities in health care in China. On the one hand, the national government s tremendous effort to strengthen social insurance appears to have been highly effective during the two 12

13 decades studied, as inequality in health insurance coverage was reduced drastically. On the other hand, the increased health insurance enrolment rates were accompanied by an overall decline in the utilization of health care services. Interestingly, in later waves of the CHNS survey sample the decline was more pronounced among the previously more privileged subcategories such as urban respondents, coastal respondents, and professionals and government employees than among the less privileged. Our analysis based on trends in costs of outpatient treatment and out-of-pocket (OOP) payments may help explain the declining utilization rate despite nearly universal health insurance coverage. Compared with insurance coverage and service utilization rates, where the level of inequalities is generally shown to have declined over time, level of inequalities in financial burden has persisted, and even worsened. Our results show that respondents in the inland provinces continued to pay a higher percentage of outpatient expenditure as OOP costs than their counterparts in coastal provinces. Farmers and migrant workers also paid a higher percentage of treatment expenditure out of pocket than did professionals. More critically, the financial burden was higher for the poor not only in the share of outpatient OOP payment in total outpatient expenditure, but also in the share of OOP as a proportion of annual income, and the level of inequalities has worsened. In particular, although only observations for outpatient care have been included, the financial burden is so high for urban and rural bottom 20% households that outpatient health expenditure amounted to over 5% and 6% of their income per capita respectively. CONCLUSION This paper reported the trend of health inequality in China since 1990s using the unique micro-level panel datasets from the China Health and Nutrition Survey. It finds that despite the 13

14 universalisation of insurance coverage, health inequality did not improve proportionately when it comes to service utilization and out-of-pocket payment. In particular, disparities along different socioeconomic status such as geographic locations, occupation type and income level matters. These findings have several policy implications. First, while rising health insurance coverage rates in China represent a critical aspect of the falling level of inequalities in health care, differences across different insurance schemes and the design characteristics of different health insurance schemes may play a key role in determining the level of inequalities in access to the health care throughout the population. Our results point to the possibility that extensive enrolment in health insurance may be achieved at the expense of deterioration in access to health care as well as increasing structural inequalities. In addition, to better understand China s health inequalities, one has to look at structural factors beyond the health sector, such as geographical location, type of profession, nature of work unit, or level of income. For example, contrary to claims made by critics of market-oriented reforms in the 1990s, structural inequalities (e.g., occupational disparity between professionals and farmers) had existed long before the reforms instead of emerging only as a by-product of the reforms. Reduction in inequalities in health care thus may depend on policy instruments beyond the realm of the health sector. Recognizing the limitations of the impacts of health policy reforms may lead to the discovery of effective policy mixes with policy instruments aimed at tackling multiple sources of inequalities. Ultimately, our results call into question the overall effectiveness of health policy reforms in China since the 1990s. From 1991 to 2011, the years covered by our study, health spending per capita surged from 74 yuan to 1,888 yuan, a more than 25-fold increase. While it has been argued that the dramatic increase in health expenditure has been essential in improving access to health care, our analysis, based 14

15 on trends over the two first decades of reforms ( ), suggests that only limited progress has been made on that front despite dramatic increases in health expenditure. 15

16 List of Abbreviations CHNS GL GOV GR HS MW Non-state NT OCC OOP OTE SOE ST UES UIS UWU China Health and Nutrition Survey Geographical Locations Government Employees Geographical Regions (Accessing) health care service (visiting local health workers or doctors) Migrant Workers Non-state Employees Non-treatment (Major) occupation Out-of-pocket (payment) Overall Treatment Expenditure State-owned Enterprise Employees Self-treatment Urban Employment Status Urban Informal Sector and Unemployed Urban Working Units 16

17 References: [1] Baeten S, Van Ourti T, Van Doorslaer E.. Rising inequalities in income and health in China: Who is left behind? Journal of Health Economics 2013; 32(6): [2] Meng Q, Fang H, Liu X, Yuan B, Xu J. Consolidating the social health insurance schemes in China: Towards an equitable and efficient health system. The Lancet 2015; 386(10002), [3] Qian J, Blomqvist Å. Health policy reform in China: A comparative perspective. 2014; Singapore: World Scientific. [4] Ramesh M, Wu X, He AJ. Health governance and health care reforms in China, Health policy and planning 2013; 29: [5] Yip W, Hsiao W. The Chinese health system at a crossroads. Health Affairs 2008; 27: [6] Yip W, Hsiao W. Harnessing the privatisation of China's fragmented health-care delivery. Lancet 2014; 384: [7] Jian W, Chan K, Reidpath, D, Xu, L. China s Rural-Urban Care Gap Shrank For Chronic Disease Patients, But Inequities Persist. Health Affairs. 2010; 29: ; [8] Xu H, Short S. Health Insurance Coverage Rates In 9 Provinces In China Doubled From 1997 To 2006, With A Dramatic Rural Upswing. Health Affairs : [9] Popkin BM, Du Shufa, Zhai, F, Zhang B. Cohort Profile: The China Health and Nutrition Survey- monitoring and understanding socio-economic and health change in China, International Journal of Epidemiology 2009; 1-6. [10] accessed [11] Casey, Bernard, et al. Policies for an ageing society: Recent measures and areas for further reform. 2003; OECD, Paris. [12] Feng J, Yu Y, Lou P. Medical demand and growing medical costs in China-based on the gap between senior citizens medical costs in urban and rural areas. Social Sciences in China 2015; 3: [13] Meng,Q, Xu L, Zhang Y, Qian J, Cai M, Xin Y,... Barber SL. Trends in access to health services and financial protection in China between 2003 and 2011: A cross-sectional study. The Lancet 2012; 379(9818): [14] Liu X, Tang S, Yu B, Phuong NK, Yan F, Thien DD, Tolhurst R. Can rural health insurance improve equity in health care utilization? A comparison between China and Vietnam. Int J Equity Health 2012; 11(1):10. [15] Ma Y, Zhang L, Chen Q. China s New Cooperative Medical Scheme For Rural Residents: Popularity Of Broad Coverage Poses Challenges For Costs. Health Affairs : [16] Tian S, Zhou Q, Pan J. Inequality in social health insurance programmes in China: A theoretical approach. Journal of Asian Public Policy 2015; 8(1): [17] Babiarz K, Miller G, Yi H, Zhang L, Rozelle S. China s New Cooperative Medical Scheme Improved Finances Of Township Health Centers But Not The Number Of Patients Served. Health Affairs. 2012; 31:

18 TABLE 1. Overall Insurance Coverage for Different Population Subcategories GL and GR OCC UWU UES Year Non Informal Migrant Urban Rural CS IN PRO FM WS Gov SOEs -state Emp & Unemp Retired Worker Source: Authors analysis of CHNS data, GL: Geographic Locations GR: Geographic Regions OCC: Major Occupation UWU: Urban working units UES: Urban Employment Status 18

19 TABLE 2. Health-seeking Behavior for Different Population Subcategories Year GL Urban-NT Urban-ST Urban-HS Rural-NT Rural-ST Rural-HS GR Coastal-NT Coastal-ST Coastal-HS Inland-NT Inland-ST Inland-HS OCC Professional-NT Professional-ST Professional-HS Farmer-NT Farmer-ST Farmer-HS W&S-NT W&S-ST W&S-HS UWU GOV-NT GOV-ST GOV-HS SOE-NT SOE-ST SOE-HS Non-state-NT Non-state-ST Non-state-HS UES Employed-NT Employed-ST Employed-HS UIS-NT UIS-ST UIS-HS Retired-NT Retired-ST Retired-HS MW-NT

20 MW-ST MW-HS Source: Authors Analysis of CHNS data, NT: Nontreatment ST: Self-treatment HS: Accessing health care service (visiting local health workers or doctors) W&S: Workers and Staff GOV: Government Employees SOE: SOE Employees Non-state: Non-state Employees UIS: Urban informal sector and unemployed MW: Migrant workers 20

21 TABLE 3. Health-seeking Behavior along Income Percentiles Urban Year 1st Quintile 2nd Quintile 3rd Quintile 4th Quintile 5th Quintile NT ST HS NT ST HS NT ST HS NT ST HS NT ST HS Rural Year 1st Quintile 2nd Quintile 3rd Quintile 4th Quintile 5th Quintile NT ST HS NT ST HS NT ST HS NT ST HS NT ST HS Source: Authors Analysis of CHNS data, NT: Non-treatment ST: Self-treatment HS: Accessing health care service (visiting local health workers or doctors) 21

22 TABLE 4. Mean Overall Treatment Expenditure and Mean Out-of-pocket Payment for Different Population Subcategories Year Urban Rural Coastal Inland Professionals Farmers Workers & Staff OTE OOP OTE OOP OTE OOP OTE OOP OTE OOP OTE OOP OTE OOP (73.1) (78.9) (65.5) (84.0) (31.1) (97.7) (40.1) (76.0) (86.2) (77.3) (85.6) (62.3) (98.3) (62.2) (55.7) (96.0) (70.9) (85.1) (63.5) (98.0) (78.7) (78.5) (97.0) (88.8) (92.5) (69.3) (100.0) (86.0) (84.7) (96.2) (86.3) (92.7) (73.9) (98.2) (88.6) (79.9) (93.4) (87.5) (90.4) (59.3) (95.7) (88.0) (79.6) (77.7) (72.2) (83.2) (80.5) (86.3) (81.4) (89.3) (83.6) (87.1) (85.2) (68.0) (84.0) (88.6) GOV SOE Non state Urban Urban unemp. Urban retired employees employees mployee employed & informal sec OTE (Obs) OOP OTE (Obs) OOP OTE (Obs) OOP OTE (Obs) OOP OTE (Obs) (50.1) (26.8) (98.6) (71.8) (36.0) (100.0) (100.0) (73.4) (42.5) (89.5) (88.1) (61.0) (61.3) (100.0) (88.2) (76.0) (70.0) (57.3) (85.9) (78.6) (26.0) (65.1) (99.9) (67.0) (92.9) (80.7) (78.9) (81.4) (52.0) (90.4) (96.2) (85.3) OOP OTE OOP (Obs) (97.0) (59.8) (98.7) (12.7) (62.9) (38.5) (87.4) (64.0) (95.2) (74.8) (97.9) (66.1) (84.3) (62.6) (95.0) (80.0) Source: Authors Analysis of CHNS data, OTE: (Mean) Overall Treatment Expenditure OOP: (Mean) Out-of-pocket Payment Bracket (): OOP as Percentage of OTE Migrant workers O(Ob OOP s) (100.0) (79.6) (99.3) (95.8) (98.6) (80.5) (89.4) 22

23 Year TABLE 5. Mean Overall Treatment Expenditure and Mean Out-of-pocket Payment for Different Income Percentiles Urban 1st Quintile 2nd Quintile 3rd Quintile 4th Quintile 5th Quintile OTE OOP %inc OTE OOP %inc OTE OOP %inc OTE OOP %inc OTE OOP %inc Year (88.5) (68.8) (75.9) (63.8) (56.9) (100.0) (46.6) (94.5) (70.8) (11.4) (58.2) (54.9) (62.1) (37.0) (71.5) (100.0) (97.0) (98.6) (92.6) (32.0) (98.8) (75.1) (84.1) (68.9) (79.1) (97.9) (98.1) (98.5) (44.1) (67.3) (79.5) (81.2) (88.7) (68.3) (80.4) (90.6) (94.9) (96.9) (80.6) (61.7) Rural 1st Quintile 2nd Quintile 3rd Quintile 4th Quintile 5th Quintile OTE OOP %inc OTE OOP %inc OTE OOP %inc OTE OOP %inc OTE OOP %inc (94.3) (95.8) (95.7) (39.5) (73.0) (99.3) (68.1) (93.1) (78.0) (78.8) (99.6) (92.6) (99.9) (97.6) (91.3) (100.0) (100.0) (97.0) (92.3) (96.4) (100.0) (99.9) (85.8) (94.0) (95.1) (98.2) (93.9) (90.4) (91.1) (93.6) (81.7) (80.1) (67.8) (77.5) (80.8) (96.1) (62.5) (84.3) (91.1) (89.4) 0.4 % inc: OOP as Percentage of Household Income Source: Authors Analysis of CHNS data,