BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 1 of 13 Research RESEARCH Government health insurance for people below poverty line in India: quasi-experimental evaluation of insurance and health outcomes OPEN ACCESS 123 45 Neeraj Sood associate professor , Eran Bendavid assistant professor , Arnab Mukherji associate 6 7 8 professor , Zachary Wagner PhD student , Somil Nagpal senior health specialist , Patrick Mullen 8 senior health specialist 1 Department of Pharmaceutical Economics and Policy, School of Pharmacy, University of Southern California, Los Angeles, CA, USA; 2Leonard D Schaeffer Center for Health Policy and Economics, University of Southern California, Los Angeles, CA, USA; 3National Bureau of Economic Research, Cambridge, MA, USA; 4Division of General Medical Disciplines, Stanford University, Stanford, CA, USA; 5Center for Health Policy and the Center for Primary Care and Outcomes Research, Stanford University, Stanford, CA, USA; 6Center for Public Policy, Indian Institute of Management Bangalore, Bangalore, India; 7School of Public Health, UC Berkeley, Berkeley, CA, USA; 8World Bank, New Delhi, India Abstract from conditions covered by the scheme of 0.56% in eligible villages Objectives To evaluate the effects of a government insurance program compared with 0.55% in ineligible villages (difference of 0.01 percentage covering tertiary care for people below the poverty line in Karnataka, points, −0.03 to 0.03; P=0.95). Eligible households had significantly India, on out-of-pocket expenditures, hospital use, and mortality. reduced out-of-pocket health expenditures for admissions to hospitals with tertiary care facilities likely to be covered by the scheme (64% Design Geographic regression discontinuity study. reduction, 35% to 97%; P<0.001). There was no significant increase in Setting 572 villages in Karnataka, India. use of covered services, although the point estimate of a 44.2% increase Participants 31 476 households (22 796 below poverty line and 8680 approached significance (−5.1% to 90.5%; P=0.059). Both reductions above poverty line) in 300 villages where the scheme was implemented in out-of-pocket expenditures and potential increases in use might have and 28 633 households (21 767 below poverty line and 6866 above contributed to the observed reductions in mortality. poverty line) in 272 neighboring matched villages ineligible for the Conclusions Insuring poor households for efficacious but costly and scheme. underused health services significantly improves population health in Intervention A government insurance program (Vajpayee Arogyashree India. scheme) that provided free tertiary care to households below the poverty line in about half of villages in Karnataka from February 2010 to August Introduction 2012. Tertiary care is often too expensive for people with low incomes. Main outcome measure Out-of-pocket expenditures, hospital use, and As a result, those with conditions requiring tertiary care often mortality. go untreated or are left with devastating hospital bills, both of Results Among households below the poverty line, the mortality rate which exacerbate poverty.1 In addition, the burden of ischemic from conditions potentially responsive to services covered by the scheme heart disease and cancer—diseases that can potentially be dealt (mostly cardiac conditions and cancer) was 0.32% in households eligible with in tertiary care—is rising in many countries with a lot of for the scheme compared with 0.90% among ineligible households just poverty such as China, Bangladesh, and India.2 3 To meet the south of the eligibility border (difference of 0.58 percentage points, 95% need for tertiary care while providing financial security to people confidence interval 0.40 to 0.75; P<0.001). We found no difference in with low incomes, several states in India have rolled out social mortality rates for households above the poverty line (households above insurance programs that provide free tertiary care to households the poverty line were not eligible for the scheme), with a mortality rate below the poverty line. These insurance programs are financed Correspondence to: N Sood, Leonard D Schaeffer Center for Health Policy and Economics, University of Southern California, Los Angeles, CA, USA nsood@healthpolicy.usc.edu Extra material supplied by the author (see http://www.bmj.com/content/349/bmj.g5114?tab=related#datasupp) Appendix: Supplementary material (figures A1-7, tables A1-4). Posted as supplied by author No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 2 of 13 RESEARCH through tax and typically do not require premiums or user fees neighboring villages on either side of the boundary drawn from beneficiaries.4 5 The Vajpayee Arogyashree scheme (VAS) between the communities chosen for early versus late was launched for this purpose in February 2010 in Karnataka, implementation India, a state with over 60 million residents and nominal gross domestic product per capita of about $1400 (£834, €1048) in Methods 2011.6 (Vajpayee is the name of a former Indian prime minister, and Arogyashree means “disease-free.”) The scheme entitled The Vajpayee Arogyashree insurance scheme eligible participants to free care for a targeted range of tertiary Most beneficiaries of the scheme were poor and lived in rural care services, mostly cardiac, oncologic, neurologic, burn, and areas with little or no access to tertiary care. Residents in eligible trauma care. Like other insurance schemes, it aims to improve areas who possessed a “below poverty line” card issued by the use of services for unmet healthcare needs while reducing often state government were automatically enrolled. This enabled catastrophic out-of-pocket expenditures associated with complex beneficiaries to receive free tertiary care at both private and illnesses.7 Unlike India’s national health insurance program for public hospitals empanelled by the scheme as capable of people in poverty at the time of our study (Rashtriya Swasthya providing tertiary care. Beneficiaries paid no premiums or Bima Yojna), however, the Vajpayee Arogyashree scheme co-payments at the point of service. As of June 2013, the scheme covers only tertiary care and requires no enrollment or annual empanelled about 150 hospitals capable of providing tertiary premiums. It also incentivizes providers to seek patients with care, including all major medical centers in the state. Hospitals cardiac and oncologic conditions whose treatment requires costly received a fixed bundled payment based on a reimbursement specialized care. We evaluated the extent to which this scheme schedule for more than 400 tertiary care service packages in led to changes in health outcomes and utilization among its cardiology, oncology, neurology, nephrology, neonatology, intended beneficiaries. burn care, and trauma care (see appendix table A4). As most There is limited evidence on the impact of health insurance on hospitals are in urban centers in southern Karnataka while the health and economic wellbeing of beneficiaries in developing beneficiaries are located in villages as far as several hundred countries.8 Published reviews suggest that while insurance kilometers away, empanelled hospitals were required to organize improves utilization and reduces personal expenditures, the health camps in rural areas to screen patients for tertiary care evidence on health outcomes is mixed.8-11 Evidence from and transport eligible patients to hospitals. Hospitals signed an Colombia suggests that the rapid expansion of health insurance agreement to conduct these health camps during the in the 1990s led to an improvement in neonatal health empanelment process and received a fixed payment per health outcomes.12 Studies in Thailand and the United States also camp conducted. Most rural patients receiving care through the provide evidence for possible health benefits when the insurance scheme were identified through these health camps. scheme is well matched to the health burden and target population.13 14 On the other hand, several studies have found a Experimental design heterogeneous or null effect of insurance on health outcomes. We exploited the phased roll out of the Vajpayee Arogyashree An evaluation of a randomized roll out of Mexico’s Seguro scheme to measure its impact on utilization, financial protection, Popular, which offers extensive insurance coverage for people and mortality. In February 2010, it offered insurance to residents below the poverty line, found it decreased catastrophic health in the northern part of the state of Karnataka; in August 2012 expenditures but had a mixed effect on utilization and health insurance coverage was extended to the entire state. During this improvements.15-19 Similarly, evaluations of insurance and user staggered implementation, we evaluated the program’s outcomes fee reduction schemes in Burkina Faso and Ghana found using a quasi-experimental design that took advantage of the reductions in catastrophic health expenditures without arbitrary boundary in coverage. In particular, we conducted improvements in health outcomes.20-22 A study of the expansion surveys in September 2012 and compared outcomes in of national health insurance in Costa Rica in the 1970s found neighboring villages on either side of the boundary drawn little impact on long term trends in child mortality.23 More recent between the communities chosen for early versus late evaluations of China’s health system reforms, including the implementation. Although surveys were conducted after rural New Cooperative Medical Scheme and the Urban Resident coverage had been announced for the southern part of the state, Basic Medical Insurance, show significant heterogeneity in the implementation in southern districts was slow and spillover was estimated impact on health outcomes and costs, although at least minimal. Of the 4000 admissions to hospital that were covered some subpopulations probably experienced increases in by the scheme in our six study districts before we conducted utilization and financial protection.24 25 the surveys, only 140 were from southern districts. We evaluated outcomes of the scheme, recognizing that its We did not believe that the close geographical proximity within unique coverage features—a discrete bundle of services that one Indian state of the early (which we term “treatment”) and were selected for their established efficaciousness, close match late (“control”) implementation villages would have an effect with high burden epidemiological targets, and on outcomes of interest other than through access to the scheme. underuse—provided a potentially compelling approach for We were thus able to use the geographic discontinuity to look promoting health improvements with insurance. We exploited at outcomes in eligible areas compared with outcomes in the phased roll out of the scheme to measure its impact on adjoining ineligible areas without introducing selection bias. utilization, financial protection, and mortality. Initially, it offered To reinforce similarity between eligible and ineligible insurance to residents below the poverty line in two households, we selected treatment and control villages to study administrative divisions comprising several districts in the by matching on geographic proximity, demographics, and northern part of the state of Karnataka; in August of 2012 the socioeconomic characteristics. In particular, we used the last scheme decided to extend insurance coverage to households available census data (2001) to randomly select 300 control below the poverty line in the entire state. During this staggered villages using probability proportional to size (population) in implementation, we evaluated the program’s outcomes using a the three ineligible districts just south of the eligibility border quasi-experimental design that took advantage of the sharp (in Shimoga, Davangere, and Chitradurga districts) and matched boundary in coverage. In particular, we compared outcomes in these villages (with replacement) to 272 similar treatment No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 3 of 13 RESEARCH villages in the three districts just north of the eligibility border covered services by individuals below the poverty line. First, (Uttara Kannada, Haveri, and Bellary). We sampled 24 villages we estimated utilization as all admissions for potentially covered twice and one village five times. Figure 1⇓ shows the conditions in any tertiary care facility. Second, to better geographical proximity of the sampled villages. The villages discriminate between admissions in tertiary care facilities where were matched by identifying the “nearest neighbor” based on a covered service was actually received and admissions where propensity scores. The census variables used to estimate a covered service was not received (for example, an admission propensity scores included the proportion of the population aged for observation to rule out a myocardial infarction), we created over 6, sex composition of population aged under 6, the a measure of utilization that excluded admissions to the proportions of schedule caste and schedule tribe (historically emergency department only and admissions with a length of disadvantaged communities), female literacy rate, and population stay of four days or less. The rationale is that most services employed. Table 1⇓ shows that treatment and control villages covered by the scheme include planned procedures and lengthy were balanced on all characteristics included in the propensity admissions. The scheme’s administrative data show more than score models. three quarters of covered stays lasted longer than four days. The Because we collected data only after implementation, the above count data for both measures was denominated by the total experimental design will produce an unbiased estimate of the number of households surveyed. effect of the scheme on health outcomes only if households in In addition to use of tertiary care, we also measured differences matched eligible and ineligible areas were similar on all in forgone need for tertiary care. This measure was based on a important observed and unobserved determinants of outcome question that asked respondents if any household member measures. We tested this assumption by comparing differences forwent care for a serious illness, and, if “yes,” to identify the in outcomes across the eligibility border for households above illness for which care was forgone. the poverty line. We expected no difference in outcomes for these households as they were ineligible for the scheme Household survey irrespective of location. All households below the poverty line with an admission for a potentially covered condition and about 10% of households Study population with a condition not covered participated in a detailed household Figure 2 shows our study population⇓. In September 2012, we survey. We thus surveyed 487 and 479 households with enumerated all households in the selected villages (44 571 potentially covered and not covered conditions, respectively, households in the eligible villages and 38 186 households in in eligible villages and 486 and 392 households with potentially the ineligible villages). Respondents were asked for the primary covered and not covered conditions, respectively, in ineligible reason for any admission to hospital during the past year from villages (fig 2⇓). a list of 33 broad conditions; we then conducted an additional The household survey asked respondents to detail out-of-pocket survey in all households with an admission for a potentially health expenditures for all admissions. Total out-of-pocket covered condition and a random sample of households with expenditures were calculated as the sum of spending on hospital admissions for conditions not covered (910 households out of charges, medicines, and diagnostic tests. 10 324 with admissions for condition not covered). Asha survey Data sources In addition to the enumeration and household surveys, the study Enumeration survey team interviewed one community health worker (Asha) in each All households in sampled villages were asked to participate in village (sample size 572). We collected village level information a door to door survey, and 81% of them completed the survey. on demographics, socioeconomic characteristics, and health Surveyors recorded information on whether the household had behaviors. a state issued below poverty line card, anyone in the household was admitted to hospital in the past year, and any members of Census and district level household surveys the household died in the past year. All questions were We used two existing datasets to characterize differences or administered in Kannada, the local language. After excluding similarities between eligible and ineligible areas. We used the 22 648 households that claimed below poverty line status but latest available 2001 census for data on demographic indicators could not produce the card, we analyzed information on 31 476 including proportion of population aged under 6, proportion households in eligible villages (22 796 below poverty line and from historically disadvantaged communities (referred to as 8680 above poverty line) and 28 633 households in ineligible scheduled caste or tribe), female literacy rate, and proportion villages (21 767 below poverty line and 6866 above poverty employed. We used the third round of the district level line) (fig 2⇓). household survey conducted in our study area between Households with a death or admission to hospital in the past December 2007 and March 2008 for data on mortality rates year were asked to identify the cause of death or admission from prior to implementation of the scheme. The survey is an ongoing a list of 33 causes, translated into lay terms. Interviewers were survey commissioned by the government of India that surveys able to verify self reported cause of admission with hospital about 1500 households in each district. We used responses to records available at time of interview from about two thirds of a question that asked respondents about any deaths in the family participants. We used information on cause of death to create since January 2004 to characterize baseline mortality rates in our primary mortality indicator, identifying all households that the study districts. reported a death in the past year from a potentially covered condition (a condition for which a service covered by the scheme Statistical analysis could have been preventive or curative). We also recorded the We first evaluated differences between eligible and ineligible age at death and sex. villages using t tests. We focused on differences in We used the information on admissions to measure utilization demographics, mortality, health related behaviors, and rates. We created two measures to estimate use of potentially No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 4 of 13 RESEARCH socioeconomic or development indicators. Demographic Study outcomes indicators were extracted from the 2001 census and baseline Mortality mortality indicators were extracted from the 2007-08 district level household survey. Indicators for health related behaviors Mortality from conditions covered by the scheme was lower were extracted from the Asha survey and included whether most among eligible households below the poverty line but similar men used tobacco or were heavy drinkers. Development among households above the poverty line (fig 3⇓). Among indicators were also extracted from the Asha survey and households below the poverty line, the mortality rate from included village availability of piped water, electricity, banks, conditions covered by the scheme was 0.32% in eligible an all weather road, government primary health centers, and households compared with 0.90% in ineligible households private clinics. (difference of 0.58 percentage points, 95% confidence interval 0.40 to 0.75; P<0.001; 64% risk reduction). There was no We next used a logit model to compare differences in mortality difference, however, in mortality rates across households above rates in 2012 for covered conditions in households below the the poverty line in treatment and control areas. Among poverty line between eligible areas and ineligible areas. We also households above the poverty line, the mortality rate from used multivariate logit regressions to control for demographics, conditions covered by the scheme was 0.56% in treatment areas health related behaviors, and development indicators described compared with 0.55% in control areas (difference of 0.01 above. Standard errors were clustered at the village level. We percentage points, −0.03 to 0.03). When we included controls then used the same methods to compare mortality among for village characteristics from table 1 and table 2, results were households above the poverty line for the same conditions in similar with a difference of 0.54 percentage points (P<0.001) eligible and ineligible villages. We expected to find no in the mortality rate for conditions covered by the scheme. differences in mortality for households above the poverty line Results were qualitatively similar if we included families who as these households were not eligible for the scheme. Any claimed below poverty line status but did not have a card as differences in mortality for potentially covered conditions in families below the poverty line in the analysis (see appendix households above the poverty line could reflect pre-existing fig A7). differences in mortality. Figure 4⇓ presents the age distribution at death from potentially The secondary outcomes for the study included out-of-pocket covered conditions. We expected the age at death from covered expenditures, utilization of tertiary care for conditions covered conditions among beneficiaries of the scheme to be higher by the scheme, and foregone need for tertiary care. For each compared with non-beneficiaries if the scheme led to increasing secondary outcome we used ordinary least squares to look at population coverage of efficacious health services, which was differences in means for households below the poverty line in supported by the results. For example, among people below the eligible areas compared with similar households in ineligible poverty line, 52% of deaths were in people aged <60 in eligible areas. We also estimated multivariate ordinary least squares households compared with 76% in people ages <60 in ineligible models for these secondary outcomes, clustering standard errors households. The distribution of age at death was significantly at the village level. The multivariate models for out-of-pocket different (Kolmogorov-Smirnov test P<0.001). expenditures controlled for differences in illness or health condition composition by including an indicator variable for each of the seven conditions covered by the scheme. We also Out-of-pocket expenditures estimated these models controlling for demographic Table 3⇓ compares out-of-pocket payments for covered characteristics, health related behaviors, and development conditions between households in treatment and control villages. indicators; as the sample of people admitted to hospital was When we included all types of facilities, the scheme was rather small, addition of too many covariates could lead to associated with a 34% reduction in out-of-pocket health over-fitting of the model and therefore our model of choice expenditures for admission to hospital for covered conditions adjusted only for illness composition, which was the most (95% confidence interval 18% to 51%). This included important confounder. The models for utilization of tertiary care admissions that were less likely to be covered because we controlled for demographics, health related behaviors, and included care provided at any type of facility (individuals were development indicators. Standard errors were clustered at the charged for care of a covered condition that did not result in village level for all analyses because villages served as the tertiary care—for example, seeking care for a cardiac condition primary units in the sample selection; we also repeated all that resulted in a prescription for antihypertensive drugs). The analyses while clustering the standard errors at the district level, difference in out-of-pocket expenditures increased to 58% when without a qualitative change to the significance level. we examine only admissions in tertiary care facilities (95% confidence interval 31% to 84%) and to 64% after we excluded Results short admissions and admissions through the emergency room (35% to 97%). All differences were more pronounced when we Baseline data adjusted for illness composition and when we controlled for the We found no pre-existing differences in mortality rates baseline characteristics from tables 1 and 2 (available on (measured in 2004-08) between treatment districts (north of request). Results were qualitatively similar when we restricted border and eligible for the scheme) and control (south of border the analyses to admissions that could be verified with medical and ineligible) districts (see appendix table A3). Socioeconomic records (see appendix table A2). and health behavior characteristics were also balanced on all but one measure (table 2⇓); a bank was available in a greater Utilization and foregone medical care proportion of control villages (35% (94) compared with 25% We compared rates of hospital admissions between households (75) of treatment villages, P=0.012). in treatment and control villages (table 4⇓). The admission rates in any facility for potentially covered conditions in the year before the survey were similar: 486 (2.1%) in the eligible areas and 485 (2.2%) in the ineligible areas (4.3% difference, 95% confidence interval −17.5% to 8.8%; P=0.52). Rates of No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 5 of 13 RESEARCH admissions in tertiary care facilities (where admissions are more usual activities, pain, and anxiety/depression (results available likely to result in covered services) show a trend towards greater from authors on request). utilization by eligible households: eligible households were Finally, we assessed differences in out-of-pocket costs between 12.3% more likely to use tertiary care for covered conditions eligible and ineligible households for conditions that were not (−20.3% to 44.9%; P=0.46) (table 4⇓). The greatest difference covered by the scheme. As expected, we found that differences in utilization was observed in rates of non-emergency admissions in costs for conditions not covered by the scheme were not for potentially covered conditions to tertiary care facilities that significantly different between eligible and ineligible households led to inpatient stays longer than four days (44.2% difference, (results available on request). −5.1% to 90.5%; P=0.06). Although this result was not significant at the 95% level, the point estimate was large and approached significance, which could suggest of a positive effect Discussion on utilization. Results were qualitatively similar when we Principal findings controlled for the characteristics presented in tables 1 and 2 (last Implementation of a health insurance program (Vajpayee column of table 4) and when we restricted the analyses to Arogyashree scheme) in the northern districts of Karnataka in admissions that could be verified with medical records (see India led to important health benefits among people below the appendix table A1). poverty line, with a reduction of 64% in mortality from In response to questions about forgone care, 52 (0.2%) conditions covered by the scheme. Measurable benefits also households in eligible villages and 76 (0.4%) households in include substantial reductions in out-of-pocket costs among ineligible villages indicated that a household member had beneficiaries. Although we did not have the power to detect an forgone care for a potentially covered illness (35.5% reduction, increase in utilization of services and a decrease in foregone 95% confidence interval −73.5% to 2.5%; P=0.07) (table 4⇓). tertiary care, point estimates were large and approached Forgone care might be complementary to the increased significance, which suggests an increase in utilization. admission rates, and the increase in admission rates for covered conditions was nearly identical to the decrease in forgone care Results in context between the two groups. Prominent studies of health insurance for people below the poverty line in developing countries have commonly failed to Sensitivity analysis measure health improvements. One reason why we found health We conducted a series of sensitivity analyses to examine improvements whereas others did not might be that the scheme potential concerns and sources of bias. The first concern was covered tertiary services, which generate more immediate the possibility that there was measurement error in illnesses benefits, whereas the health effects of primary care interventions reported as reasons for mortality, which could bias our estimates. covered by other programs might not be observed for years. For We dealt with this in several ways. First, we restricted the instance, the evaluation of Mexico’s Seguro Popular did not estimated mortality rates to two common diseases people are find changes on nine self assessed health indicators, despite generally familiar with—cancer and cardiac conditions—which improved financial security. Seguro Popular, however, covers are likely to have less measurement error. When we looked at a broad array of primary healthcare services, and the self mortality rates for these familiar conditions we found results assessed health measures were similarly broad. Similar outcomes similar to overall mortality (see appendix figures A1 and A2). have been observed in Ghana, Burkina Faso, and Costa Rica, Second, we estimated differences in mortality rates from while studies in China and Thailand suggest the possibility of conditions that were not covered (such as diarrhea, diabetes, health improvements after initiation of health insurance asthma, and tuberculosis), where we would expect to find no schemes.14 20 22 23 27 effect. We found that eligible and ineligible households below The implementation features of the Vajpayee Arogyashree the poverty line had similar mortality rates for these conditions scheme can possibly help to explain its measured benefits, (see appendix figure A3). Third, we looked the distribution of including mortality benefits that are rarely measurable in causes for hospital admissions from the scheme’s administrative evaluations of other health insurance schemes. First, the records compared with the distribution of self reported causes requirement for empanelled hospitals to hold health outreach from our household survey. The distributions were similar, camps possibly alleviated the selection created by the provision suggesting there was little measurement error in this self of health insurance, whereby those who are best off are also the reported illness measure (see appendix figure A4). Finally, we ones most likely to take advantage of its benefits. Instead, this examined information on reported cause of death from our feature potentially promoted the provision of services to household survey in ineligible areas compared with a verbal individuals living in regions where tertiary care was rarely used autopsy study conducted in India in 2001-03. The proportions and often foregone. In addition, the automatic enrollment of all of deaths from cancer, cardiac conditions, and chronic below poverty line cardholders with no premiums, user fees, or respiratory infections in people aged over 25 were similar in co-payments removed an important barrier that could limit the the verbal autopsy study and the household survey (see appendix use of health services among individuals in poor health. At the figure A5).26 same time, it also instituted a pre-authorization process to Next, we examined the concern that mortality benefits of the mitigate the risk of overprovision of care. The focus on a scheme might be overstated if people who were treated were discrete, costly, and efficacious set of medical conditions helped left in an unproductive state of high morbidity. To assess this, in identifying the target population that most benefitted from we used a standard health status indicator to estimate health the scheme. Finally, services covered by the scheme were well before and after admission to hospital for people who received matched to reflect conditions with a rising share of disease procedures covered by the scheme. Beneficiaries of the scheme burden in India and whose management is otherwise inaccessible reported improvements in overall health after the admission to for people below the poverty line. These design features behind hospital and were relatively healthy at the time of the survey the scheme could be used by extension to other regions. (see appendix figure A6). We found similar patterns for more specific health domains: ability to walk, self care, ability to do No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 6 of 13 RESEARCH Strengths and limitations quality standards and thus beneficiaries might be receiving care Our study had several methodological strengths, including the in “better” hospitals. Finally, the scheme instituted a matching process of villages across the implementation pre-authorization process that might have limited inappropriate boundaries. While we used propensity score matching using use of tertiary care. census data from several years before the study, we found that Fifth, the evidence we found of reduced mortality associated the villages matched well on multiple observed dimensions not with conditions covered by the scheme could reflect a previously used in the matching process, suggesting that the villages were unmet demand for tertiary care, suggesting that the long term matched on unobserved variables as well. We also focused on effect of the program might be less dramatic. health outcomes proximal to the services covered by the scheme, Finally, people who needed tertiary care and lived in ineligible as its circumscribed set of covered services enabled areas might have migrated north of the eligibility border to gain measurement of changes in cause specific mortality even if access to the scheme. As such people were likely to have been changes in general population health were difficult to measure. sicker, it would cause us to understate the effect of the scheme This study was, however, also limited in several ways. First, it on reducing mortality. Discussion with local experts and was quasi-experimental in that the scheme was not randomly government officials, however, suggests that migration is not assigned to villages. This posed several methodological a big concern as it would require a new below poverty line card challenges but also presented opportunities for using rigorous or address change on an existing card, both of which involve approaches designed to reduce selection bias. The northern arduous and drawn out processes, especially for rural households portion of Karnataka was selected for coverage because the state below the poverty line. government thought that Karnataka’s northern regions were in greater need of tertiary healthcare. The extent to which this is Conclusions and policy implications true is unknown, but for that reason we selected villages on the People below poverty line in India with conditions requiring southern border of the eligibility area and matched them to tertiary care have the choice of trying to access lower cost villages just south of the eligibility boundary. The baseline data government tertiary care services, taking on devastating debt support our assumption that villages just north and just south to pay for care from private sector hospitals, or experiencing of the border were similar on relevant characteristics. The the health consequences of foregoing treatment for their illness. finding of no difference in mortality for households above the India is currently pursuing several strategies to improve health poverty line also supports this important assumption. In future services for its population, including investing in government work we aim to resurvey these households to understand how provided services as well as purchasing services from public expansion of the scheme to the south of the eligibility border and private providers through schemes similar to the Vajpayee affected outcomes. Arogyashree scheme.4 28 While a scheme such as this might Second, the classifications of causes of death by family members provide significant health and economic benefits to people below are important for the findings. If measurement error were a the poverty line, future research will need to assess the cost factor, we would expect to see similar patterns of effectiveness provided compared with alternative social misclassification among households above the poverty line protection and health promotion programs. where none were found. Moreover, as eligible households had more exposure to information about covered illnesses through We presented preliminary findings from this research to World Bank, health camps, we would expect greater reporting of these Government of Karnataka, and academic healthcare researchers. illnesses among eligible households as a potential cause of death, Contributors: NS led all aspects of the study. EB and ZW co-led all which would bias our findings. We also found no evidence of analysis and preparation of the manuscript. AM co-led data collection measurement error affecting results in the various sensitivity and helped with data analysis. SN and PM helped conceive the study, analyses we conducted. Nonetheless, measurement error could coordinated with the government, obtained research funding, and still be a source of bias. provided critical comments on research design and results. NS is Third, we could not directly measure who was covered by the guarantor. scheme, only whether the admission to hospital or death was Funding: This project was funded by the Health Results Innovation Trust related to a condition for which management was potentially Fund at the World Bank. The funding agency played no role in the covered by the scheme. Because of this we probably analyzed conduct of this research and the decision to submit this manuscript for admissions and deaths that were outside of the scope of the publication. scheme together with the truly covered services, thus diluting Competing interests: All authors have completed the ICMJE uniform our estimates of effect size. Indeed, our estimates of disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no out-of-pocket expenditures and utilization increase with greater support from any organization for the submitted work; no financial efforts to identify those admissions that were truly covered by relationships with any organizations that might have an interest in the the scheme. Moreover, these more restrictive measures were a submitted work in the previous three years; no other relationships or closer match to the number of admissions reported in the activities that could appear to have influenced the submitted work. scheme’s administrative records. Ethics approval: The study was reviewed and approved by the Fourth, a related concern is that effects of the scheme on institutional ethics committee at Indian Institute of Management, utilization of tertiary care are imprecisely estimated. It is, Bangalore, India (IRB# IORG0004307). however, important to note that the scheme could affect Data sharing: Technical appendix, statistical code, and data used in mortality even in the absence such a utilization effect. For this study are available from the corresponding author. instance, health camps organized by the scheme and easier access to tertiary care might have increased individuals’ Transparency declaration: The authors affirm that the manuscript is an likelihood of seeking primary healthcare for symptoms (such honest, accurate, and transparent account of the study was reported as chest pain) that might require tertiary care. This could result and no important aspects of the study have been omitted. in earlier detection of disease, which could reduce mortality. 1 Whitehead M, Dahlgren G, Evans T. Equity and health sector reforms: can low-income Similarly, the scheme empanels hospitals that meet certain countries escape the medical poverty trap? Lancet 2001;358:833-6. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 7 of 13 RESEARCH What is already known on this topic Health insurance schemes in developing countries can reduce financial hardship and increase utilization of healthcare Evidence on the impact on health of health insurance for people below the poverty line is mixed What this study adds Insuring poor households for efficacious but costly and underutilized health services, coupled with recruitment of patients who could benefit from these health services, significantly improved population health in India 2 Yusuf S, Reddy S, Ôunpuu S, Anand S. Global burden of cardiovascular diseases. Part 19 Sosa-Rubí SG, Galárraga O, López-Ridaura R. Diabetes treatment and control: the effect II: Variations in cardiovascular disease by specific ethnic groups and geographic regions of public health insurance for the poor in Mexico. Bull World Health Organ 2009;87:512-9. and prevention strategies. Circulation 2001;104:2855-64. 20 Powell-Jackson T, Hanson K, Whitty CJ, Ansah EK. Who benefits from free healthcare? 3 Ferlay J, Shin H-R, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide Evidence from a randomized experiment in Ghana. J Development Economics burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer 2010;127:2893-917. 2013;107:305-19. 4 World Bank. Government-sponsored health insurance in India: are you covered? http:// 21 Ansah EK, Narh-Bana S, Asiamah S, Dzordzordzi V, Biantey K, Dickson K, et al. Effect documents.worldbank.org/curated/en/2012/08/16653451/government-sponsored-health- of removing direct payment for health care on utilisation and health outcomes in Ghanaian insurance-india-covered. children: a randomised controlled trial. PLoS Med 2009;6:e1000007. 5 Fan VY, Karan A, Mahal A. State health insurance and out-of-pocket health expenditures 22 Fink G, Robyn PJ, Sié A, Sauerborn R. Does health insurance improve health? Evidence in Andhra Pradesh, India. Int J Health Care Finance Econ 2012;12:189-215. from a randomized community-based insurance rollout in rural Burkina Faso. J Health 6 List of Indian states by GDP. http://en.wikipedia.org/wiki/List_of_Indian_states_by_GDP. Econ 2013;32:1043-56. 7 Vajapayee Arogyasri home page. http://stg2.kar.nic.in/healthnew/SAST/Home.html. 23 Dow WH, Schmeer KK. Health insurance and child mortality in Costa Rica. Soc Sci Med 8 Escobar M-L, Griffin CC, Shaw RP. The impact of health insurance in low-and 2003;57:975-86. middle-income countries. Brookings Institution Press, 2011. 24 Wagstaff A, Lindelow M, Jun G, Ling X, Juncheng Q. Extending health insurance to the 9 Palmer N, Mueller DH, Gilson L, Mills A, Haines A. Health financing to promote access rural population: an impact evaluation of China’s new cooperative medical scheme. J in low income settings—how much do we know? Lancet 2004;364:1365-70. Health Econ 2009;28:1-19. 10 Ekman B. Community-based health insurance in low-income countries: a systematic 25 Lin W, Liu GG, Chen G. The urban resident basic medical insurance: a landmark reform review of the evidence. Health Policy Plan 2004;19:249-70. towards universal coverage in China. Health Econ 2009;18(S2):S83-96. 11 Moreno-Serra R, Smith PC. Does progress towards universal health coverage improve 26 Office of the Registrar General and Census Commissioner, India. Summary—report on population health? Lancet 2012;380:917-23. causes of death: 2001-03 in India. Government of India. 2011. http://censusindia.gov.in/ 12 Camacho A, Conover E. Effects of subsidized health insurance on newborn health in a . developing country. Econ Dev Cultural Change 2013;61:633-58. 27 Wang H, Yip W, Zhang L, Hsiao WC. The impact of rural mutual health care on health 13 Sommers BD, Long SK, Baicker K. Changes in mortality after Massachusetts health care status: evaluation of a social experiment in rural China. Health Econ 2009;18(S2):S65-82. reform: a quasi-experimental study. Ann Intern Med 2014;160:585-93. 28 High Level Expert Group Report on Universal Health Coverage for India. Planning 14 Gruber J, Hendren N, Townsend R. Demand and reimbursement effects of healthcare Commission of India, 2011. www.uhc-india.org/. reform: health care utilization and infant mortality in Thailand. National Bureau of Economic 29 Rani M, Bonu S, Jha P, Nguyen S, Jamjoum L. Tobacco use in India: prevalence and Research, 2012. predictors of smoking and chewing in a national cross sectional household survey. Tob 15 Bernal MP, Azuara MO, Conti G, Luengas MP. El efecto del Seguro Popular en la salud Control 2003;12:e4-e. infantil. Centro de Investigación y Docencia Económicas AC, 2010. 16 King G, Gakidou E, Imai K, Lakin J, Moore RT, Nall C, et al. Public policy for the poor? Accepted: 24 July 2014 A randomised assessment of the Mexican universal health insurance programme. Lancet 2009;373:1447-54. 17 Gakidou E, Lozano R, González-Pier E, Abbott-Klafter J, Barofsky JT, Bryson-Cahn C, Cite this as: BMJ 2014;349:g5114 et al. Evaluating the impact of the 2001-2006 Mexican Health Reform: an interim report This is an Open Access article distributed in accordance with the Creative Commons card. Lancet 2006;368:1920-35. Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, 18 Sosa-Rubi SG, Galarraga O, Harris JE. Heterogeneous impact of the “Seguro Popular” program on the utilization of obstetrical services in Mexico, 2001-2006: a multinomial remix, adapt, build upon this work non-commercially, and license their derivative works probit model with a discrete endogenous variable. J Health Econ 2009;28:20-34. on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/3.0/. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 8 of 13 RESEARCH Tables Table 1| Village level characteristics used for propensity score matching according to eligibility for government insurance program covering tertiary care for people below poverty line—Vajpayee Arogyashree scheme (VAS) Demographics* VAS eligible (300 villages) VAS ineligible (272 villages) P value† Age <6 14.4% 14.1% 0.14 % Female aged <6 48.5% 48.6% 0.65 Scheduled caste‡ 21.0% 21.3% 0.94 Scheduled tribe‡ 14.9% 12.8% 0.15 Female literacy 43.1% 44.3% 0.29 Population employed 50.6% 49.8% 0.19 *Data from 2001 census. †Estimated from t tests. ‡Historically disadvantaged communities. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 9 of 13 RESEARCH Table 2| Village level development and health related characteristics according to eligibility for government insurance program covering tertiary care for people below poverty line—Vajpayee Arogyashree scheme (VAS) VAS eligible (300 villages) VAS ineligible (272 villages) P value* Development indicators† Piped water 49.7% 48.2% 0.71 Electricity in most households 95.1% 92.6% 0.20 Bank in village 25.0% 34.6% 0.01 Distance to nearest town (km) 13.3 12.2 0.14 All weather road in village 85.3% 87.6% 0.41 Primary health center in village 21.5% 19.8% 0.61 Private clinic in village 44.1% 38.5% 0.18 Health behaviors* Most men heavy drinkers 59.7% 53.8% 0.15 Most use tobacco 67.3% 67.0% 0.91 *Estimated from t tests. †Data from Asha survey (572 villages). Tobacco use measures are consistent with prior work assessing use of tobacco in India. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 10 of 13 RESEARCH Table 3| Out-of-pocket expenditures (in Indian rupees*) for conditions covered by government insurance program covering tertiary care for people below poverty line—Vajpayee Arogyashree scheme (VAS) Mean out-of-pocket expenditures* % Difference in expenditure VAS area Non-VAS area Unadjusted Adjusted† All facilities (n=986) 32 256 49 238 −34% (P<0.001) −39% (P<0.001) Tertiary care facilities (TCFs) 26 725 62 966 −58% (P<0.001) −60% (P<0.001) (n=199) TCFs excluding emergency 24 725 73 134 −66% (P<0.001) −69% (P<0.001) department admissions and stays of ≤4 days (n=139) *1000 rupees = £10 (€12, $16). †Adjusted for illness composition (burns, neonatal conditions, cancers, cardiac conditions, neurological diseases, renal conditions, and poly trauma) using ordinary least squares with standard errors clustered at village level. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 11 of 13 RESEARCH Table 4| Utilization of tertiary care covered by government insurance program covering tertiary care for people below poverty line—Vajpayee Arogyashree scheme (VAS) % Difference No (%) in VAS area (n=22 No (%) in non-VAS area 796) (n=21 767) Unadjusted (P value) Adjusted* (P value) Households using tertiary care facility for potentially covered conditions All facilities 487 (2.1) 486 (2.2) −4.3% (0.52) −5.4% (0.64) All tertiary care facilities (TCFs) 107 (0.5) 91 (0.4) 12.3% (0.46) 19.9% (0.26) Excluding emergency department 77 (0.3) 51 (0.2) 44.2% (0.06) 42.7% (0.08) admissions and stays of 4 ≤days Households reporting forgone need for care for VAS condition Reported forgone need 52 (0.2) 77 (0.4) −35.5% (0.07) −33.4% (0.09) *Adjusted for village level characteristics using ordinary least squares, including whether most households have piped water, whether there is all weather road in village, distance to nearest town, whether there is clinic or hospital in village, whether there is bank in village, average household income, proportion of people who use tobacco, proportion of people who drink alcohol, and average self reported health score. Standard errors clustered at village level. No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 12 of 13 RESEARCH Figures Fig 1 Study region in investigation of government health insurance for people below poverty line. Dots represent sampled villages. Map on left is state of Karnataka; map on right is zoomed out to show southeastern part of India Fig 2 Flow chart of participants in study of government health insurance for people below poverty line in India No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2014;349:g5114 doi: 10.1136/bmj.g5114 (Published 11 September 2014) Page 13 of 13 RESEARCH Fig 3 Proportion of households that reported death during previous year from conditions covered by scheme according to geographic elegibilty. Households above poverty line are not eligible for scheme. VAS=north of border and eligible for scheme. Non-VAS=south of border and not eligible for scheme Fig 4 Mortality by age for conditions covered by scheme No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe