Some Indian non-governmental organisations (NGOs) have initiated community health insurance (CHI) schemes to address this problem. Experts define CHI as any not-for-profit insurance scheme aimed primarily at the informal sector, formed on the basis of a collective pooling of health risks, and in which the members participate in its management. The number of CHI schemes in India has grown exponentially in the past decade, partially fuelled by the micro-credit movement.

In theory, CHI is a relevant option for the informal sector with its combination of pre-payment and risk pooling mechanisms. However, there is little evidence that Indian CHI schemes increase access to care and protect against catastrophic health expenditures (CHE). The objective of our research was to investigate whether and under what conditions CHI improves access to hospital care, provides protection against CHE and increases patient satisfaction.

We first undertook a detailed case study of 10 purposively selected CHI schemes to improve our understanding of the variety of CHI schemes in India. Based on these findings, we selected three CHI schemes and studied them in greater detail. Each of the three chosen schemes (ACCORD, KKVS and SEWA)1 represented one of the types of CHI in India.

We conducted household surveys in randomly selected insured and uninsured households at ACCORD and KKVS. Data on socio-economic characteristics, morbidity patterns, health-seeking behaviour, health expenditures and patient satisfaction were collected. In addition, we conducted focus group discussions with insured and uninsured individuals, with hospital patients and providers in order to understand their perceptions of quality of care and of the CHI scheme. In all three schemes, we compiled secondary data on details of hospitalisations from existing registers and records. We measured access and the incidence of CHE.

All 10 CHI schemes studied were initiated by NGOs with the objective of increasing access to health care, preventing indebtedness and empowering communities. These CHI schemes explicitly targeted the poorest and most vulnerable households in Indian society, i.e., scheduled castes and tribes, as well as women. Further, all schemes used existing community organisations to introduce CHI, thereby building on prevailing social capital and trust. Three distinct types of CHI schemes can be distinguished based on the role of the NGO. In the provider type, the NGO was both the insurer and the provider of health care. In the mutual type, the NGO was the insurer and purchased care from providers. Finally, in the linked type, the NGO insured the community with an insurance company and purchased health care from providers.

Most of the schemes were based on voluntary enrolment, with the individual as the unit of enrolment. Membership levels in the schemes ranged from 1,000 to 100,000 individuals. Premiums were community-rated and ranged from US$0.5 to US$5 per person per year. All 10 schemes insured against hospitalisation expenses but only up to a certain amount. While most common diseases were covered, some conditions such as chronic ailments and pre-existing conditions were excluded. Very few of the schemes had proper documentation or monitoring systems. This lack of data meant that none of these schemes had empirical evidence to suggest that they increased access to hospital care or protected families from CHE.

The household survey at ACCORD demonstrated that 57% of insured and 58% of uninsured individuals experienced minor ailments during the period 2004-2005. The proportion of individuals with chronic ailments was 5% among the insured and 2% among the uninsured. The proportions of insured and uninsured individuals with major ailments were 14 and 8%, respectively.

The admission rates among the insured and uninsured were 92 and 42 per 1,000 people per year, respectively. Ninety percent of insured pregnant women delivered in a hospital, while the corresponding figure for the uninsured was 45% (χ2 = 8.6; df = 1). The study also revealed that 65% of insured patients with major ailments were admitted to a hospital, compared to 44% of uninsured patients (OR 2.2; 95% CI 1.31, 3.77). This higher admission rate among insured patients was also found in vulnerable groups, such as children, females, people of lower socioeconomic status and those living far from a hospital. In the lowest income quintile, the probability of admission for insured patients was 3.47 times higher than it was for the uninsured. Insurance status remained a significant determinant of increased utilisation of hospital services after controlling for confounding factors such as age, gender, distance from a hospital and the presence of pre-existing ailments.

This study clearly indicates that the ACCORD CHI was able to increase access to hospital care, even for the poorest and most vulnerable groups in society. Some reasons for this may be its comprehensive benefit package, affordable and subsidised premiums, credible and effective provider and a cashless system.

Our analysis of data extracted from the registers at ACCORD and SEWA showed that, in 2003 and 2004, there were 683 and 3,152 admissions, respectively. All of the patients insured by ACCORD were admitted to a not-for-profit hospital, while those insured by SEWA were mostly admitted (86%) to private-for-profit facilities. The median hospital bill per admission was US$12 for patients at ACCORD and US$46 for patients at SEWA. The median annual household income was US$630 for patients insured by ACCORD and US$545 for those insured by SEWA.

Overall, 74% of patients insured by ACCORD and 38% of patients insured by SEWA did not have to make any payments at the time of hospitalisation. The rest had to make out OOP payments because of co-payments, exclusions or both. Without health insurance, 8% of the families of patients insured by ACCORD and 49% of the families of patients insured by SEWA would have experienced CHE. The CHI scheme managed to reduce the incidence of CHE to 3.5% at ACCORD and 23% at SEWA. Not only was the incidence of CHE halved, but the intensity of the OOP payment also decreased. Without the CHI scheme, families of patients insured by both ACCORD and SEWA would have spent 14% of their annual income on hospital expenses, but the CHI scheme reduced this figure to 9% of annual income in both locations. The chances of experiencing a CHE for families of patients insured by SEWA were increased if the patient was poor, had gone to a private health care provider or needed surgery.

Our study of ACCORD and SEWA showed that in both of these schemes, CHI provided financial protection against OOP payments and CHE. However, this protection was only partial, and some patients enrolled in both of the schemes still experienced CHE. The main reasons for this were the low upper limits in both schemes and the exclusion of some clinical conditions at the SEWA scheme.

The analysis of household survey data from ACCORD indicated that 92 and 87% of insured and uninsured patients, respectively, were satisfied with the care that they received. At KKVS, the corresponding figures were 95 and 79%. While the difference in satisfaction between insured and uninsured patients was not statistically significant at ACCORD, it was at KKVS (χ2 = 7.65; df = 1).

At ACCORD, the main reasons for satisfaction among both the insured and uninsured were the health care infrastructure (84 and 78%, respectively), followed by the interpersonal interaction with the doctors and nurses. However, only about half of the patients, both insured and uninsured, were satisfied with the care process. This was because either they had to wait for a long time to receive care or their relatives were not allowed to visit them. Uninsured patients who sought care in private hospitals were less likely to be treated courteously and to receive medications from the hospital pharmacy.

At KKVS, the reasons for satisfaction were slightly different. Most of the patients, both insured and uninsured, were satisfied with the infrastructure (86 and 98%, respectively) and the doctors’ services (91 and 85%, respectively). However, a sizable number of insured and uninsured patients were dissatisfied with the nursing care they received (47 and 56%, respectively) and the care process (84 and 91%, respectively).

Our research indicates that there was little difference in satisfaction levels between insured and uninsured patients at both ACCORD and KKVS. This finding may be attributed to a lack of strategic purchasing by the respective NGOs. Also, we measured satisfaction levels on a dichotomous scale. Had we used a wider scale, we would probably have obtained a more nuanced response.

This study is one of few that have systematically evaluated the insurance functions of CHI schemes in India. Most of the CHI schemes in Africa and Asia adopt either the provider or the mutual model. However, in India, we observed the emergence of the linked model, where the risk-taker is an insurance company. This model is advantageous in that the risks are pooled more widely, both between the healthy and the ill as well as between the rich and poor.

The evidence from our study demonstrates that Indian CHI schemes can increase access to hospital care and at least partially protect families from CHE. However, to make them more effective, some critical issues must be addressed. Design features, such as minimal exclusions to reduce OOP payments, enrolment of families as a unit to control adverse selection and a referral system to prevent moral hazard, should be addressed. Simultaneously, the community must be empowered so that its members understand the complexity of the CHI and are given the space to make informed decisions. The capacity of the CHI management should be built up so that the organisers can purchase care strategically and monitor the scheme effectively. To increase financial viability, CHI schemes need to consider reinsurance with an insurance company.

However, another route to financial viability is to increase the size of the scheme by federating many CHI schemes into a single body. This provides the added advantage that such a federation will be able to negotiate effectively with both insurance companies and providers. Concomitantly, the government must create a more supportive policy environment for the development of CHI programmes in India. This could include giving legal recognition to these entities and providing the necessary subsidies to permit the poor to enrol.

The government of India recently introduced a fully subsidised national health insurance scheme (RSBY) to protect its poorest citizens from incurring hospital expenses. CHI schemes can complement the RSBY in two ways. CHI schemes can increase the depth of cover by covering ambulatory expenses for RSBY members. The second mechanism is by targeting the near-poor and low-income groups, who are also exposed to the challenges of reduced access and CHE. These changes would help with enhancing health security for a larger section of the Indian population.

To conclude, CHI schemes in India can increase access to hospital care and protect households from CHE, provided that they are properly designed and implemented. Premiums must be affordable, benefit packages must be comprehensive, providers must be regulated, and reimbursements must be cashless and effortless. Such a scheme can play a crucial role in increasing the depth and breadth of social health protection in India.