A total of 305 elderly were included in the study out of which 155 (50.8%) were males and 150 (49.2%) were females. The mean± SD age among the elderly was 68.5+ 9.0 (range 60-101). Elderly were mostly living with children or their extended families (84.5%), were currently married (63%), had primary education and above (74.6%), had no job or living on pension (72.1%) and were living with chronic disease (72.5%).

176 elderly (57.7%) were currently enrolled under CHIS and 129 (42.3%) were not covered under the scheme. Out of the 129 who were not covered, 40 (31%) were from households never-insured under the scheme, 40 (31%) did not renew the scheme for the year 2015-16 and 49 (38%) were uninsured elderly from the insured households.

Elderly who were married (65.1%), living in pucca (61.3%) or pucca/semi-pucca mixed houses (63.3%), who were living with spouses (81.6%), in Below Poverty Line (BPL) list (64.8%) and were working (74.1% ) had statistically a better chance of getting insured compared to their counterparts. Even though not statistically significant males (60.1%) and young-old (60.4%) elderly had more proportion of insured and on the other hand elderly from scheduled caste group had less number of insured (45.2%) as compared to other groups. Table 1 shows the basic characteristics of the insured and uninsured elderly population.

This is one of the few studies from Kerala looking at the population coverage, service coverage and OOP expenses among elderly who are insured under the social protection scheme CHIS. We found that even in the state of Kerala with the highest literacy rate, the coverage of CHIS among elderly was only 57.7%, which is lower than the average for the target population i.e., 80%.24 We found that among currently single elderly from Scheduled Caste who lived alone in kutcha or semi-pucca (poor quality) houses with no employment or pension were more likely to be excluded from CHIS. Thus, the most socially and economically vulnerable elderly were not covered. The existing literature on government sponsored social health insurance schemes for poor also states that the neediest in the target population are excluded in most of the cases.15,16,17 As CHIS is a voluntary scheme and the presence of the person is required in the enrolment center for enrolling in the scheme, the bed ridden and very sick elderly are excluded by the scheme. Those elderly living alone may also lack information about the enrolment dates. The other reason for such a low coverage among elderly is because only 5 members can be enrolled under the scheme from a household, so the head of the household decides who should be enrolled. This also makes the non-working dependent elderly excluded from CHIS.

Next we tried to look at whether the population coverage translated into service coverage i.e., whether hospitalization among insured were covered by CHIS. Elderly being a vulnerable group with highest morbidity among all age groups also had the highest prevalence of hospitalization. Insured elderly had a higher proportion of hospitalizations compared to those who were un-insured. This result is consistent with the results of studies on GSHIS in Vietnam, Mexico, Colombia, Taiwan and India.11, 15, 20, 21. This finding may be due to the fact that more than half of the elderly were insured and most of the insured had chronic diseases. But when we look more into the hospitalizations among elderly we can see that even though CHIS increased hospitalization in the insured only 38.4% of the hospitalization among elderly could make use of CHIS smart card during the study period. So even those who were under the umbrella of social protection scheme, which was meant to cover only hospitalizations were not fully covered by the scheme. In other words this scheme did not cover almost two third of the hospitalizations among insured which it supposed to cover and by default did not cover the out-patient visits which constitutes more than 70% of the total health care utilization among elderly in India.28,29 This also reflects the number and quality of the hospitals empanelled under CHIS.

Lastly we tried to find out whether the elderly who were insured under CHIS incurred out-of-pocket spending, given that CHIS was introduced with the ultimate aim of reducing OOP. Since CHIS did not cover 56.3% episodes of hospitalizations among 61.6% of the insured elderly, the insured did incur both direct and indirect OOP expenses. We also found that instead of reducing the out-of-pocket spending during hospitalization, those who were insured had higher indirect OOP compared to un-insured. Interestingly within the insured group, those who were fully covered under CHIS for all episodes had higher OOP compared to those who were partially and not covered. This finding was contrary to many findings from Vietnam and a systematic review by Ernst Spann et al from Asia and Africa. These studies found that insurance reduced OOP expenses.18, 23 But studies from India have found that either no impact or an increase in OOP expenditures.30 This finding can be attributed to the non-translation of population coverage into service coverage which provides a false assurance of coverage among insured and the lack of coverage for chronic disease related care package in CHIS. The other reason for this is the fact that less than half of the hospitalizations among the insured households were not covered by CHIS. They were either un-insured from the insured households or they could not use the smart card. The main reasons for not making use of the CHIS smart card were refusal from the empanelled hospital to accept the card, lack of knowledge regarding the list of empanelled hospitals/benefit package, referral from an empanelled hospital to an un-empanelled hospital, smart card reading machine not working in the hospital and diagnosed case not covered under CHIS/costs more than 30,000 INR.

The financial burden borne by the households with elderly are immense. The mean OOP per episode of hospitalization is INR 14,569 and the mean drug expenditure per month is INR 1105.09 among insured, which is almost eight times the mean per capita expenditure (MPCE) of an urban household in Kollam (INR 1763.88) and 13 times the MPCE of the lowest quintile in Kerala (INR 1103).31The mean drug expenditure per month among insured is INR 1105.09 which is almost equivalent to that of the total expenditure of the lowest quintile in Kerala.31 So the households are financing their health care for the elderly by distress finance mechanisms and this is not a onetime event as elderly are prone to hospitalizations, and care for chronic diseases is lifelong. This finding calls an immediate attention of the policy makers to develop a service package specially designed for elderly, otherwise more and more households will be pushed below poverty due to the huge OOP spending.