This is the first nationally representative study in India on health spending associated with NCDs, the way such spending is financed, and the implications this spending has on catastrophic spending and impoverishment. We found that out of pocket expenses for treating NCDs rose sharply over the period from 1995-96 to 2004. The survey data we use suggest that about 40% to 50% of these expenditures are financed by household borrowing and sales of assets. These patterns indicate significant financial vulnerability to NCDs and we find both catastrophic spending and impoverishment more likely for those households that have a member hospitalized with an NCD compared to someone hospitalized with a communicable condition.
A substantial increase in utilization of health services occurred from 1995-96 and 2004 for all NCD categories. The reasons for this rise are beyond the scope of this study. However, possible reasons for this pattern include increases in prevalence, diagnosed disease, awareness and demand among patients for services, awareness and provision of services by providers, access to treatments and sharp rises in incomes. We can expect that this pattern will continue in the future.
Our data also confirm the important role that the private sector currently plays in the provision of health services for both hospital stays and outpatient visits associated with NCDs. Importantly, with its use, the financial risk is higher as out of pocket expenses per hospital stay and per outpatient visit are substantially higher in private than in the public facilities as indicated in Figures and .
We highlight that a substantial proportion of expenditures are for medications, diagnostics and medical appliances. Medications play a critical role in reducing the risk of developing complication and diagnostic investigations are needed to determine the treatment plan and make the best use of medications. If these are foregone, physician consultation and assessment efforts will result in limited benefit.
A higher share of household expenditure is accounted for by out of pocket expenses among the richer subpopulation which seems to counter the idea that NCDs are creating a financial burden on the poor [16
]. However, as noted already, individuals belonging to the lowest expenditure quintiles live much closer to the survival threshold, so allocating even small proportions of their low incomes will increase their likelihood of falling below the poverty line. Other monetary indicators of the financial burden suffered by households with persons with NCDs, such as income losses or premature mortality, which are not part of this study, may also contribute to this pattern.
Another way to examine the extent to which households are financially vulnerable to NCDs is to assess how expensive the costs of hospital stays are for NCDs relative to annual income (or total consumption spending). In 2004, India's income per capita was INR 25,320 while a single hospital stay for cancer or heart disease obtained from private facilities would account for anywhere between 80% and 90% of this income. Even if health care was sought from public facilities, the out of pocket expenses would still have amounted to between 40% and 50% of per capita income. A previous study has shown that the bite out of income per capita taken out by a single hospital stay increased sharply between 1995-96 and 2004 for the poorest individuals [17
There are limitations to our study. Surveys with self-reported diseases and conditions are likely to underestimate the prevalence of different types of health conditions, and there may be a misclassification of diseases as well. We found our data's derived diabetes prevalence rates to be lower than obtained by a large diabetes survey in India with 18,000 participants which measured prevalence directly with laboratory examinations [18
]. Compared to this diabetes survey, our NSSO data had lower diabetes rates in both the urban and rural areas (urban 5.9% versus 2.1%; rural 2.7% versus 0.7%) although urban-rural prevalence ratios are similar. While this may influence estimates of aggregate population-based income losses, if households accurately report all of their health spending, our results would capture the financial implications of specific health conditions at the household level, even if the overall prevalence levels are downwardly biased. Because the survey data we used provided information on outpatient healthcare use conditional only on reporting an ailment, it is possible that both healthcare use and out of pocket expenditures on outpatient care are underreported, although one recent study suggests that the impact of any underreporting is small [15
Modeling the poverty impact for catastrophic and impoverishment also requires the assumption that there are no economies of scale in household spending [13
]. Another key related assumption is that household expenditures would have remained unchanged in absence of health expenditures associated with NCDs. In the absence of additional data, the precise impact of these assumptions on our conclusions is difficult to ascertain. If household consumption were lower in the absence of health spending, for instance if increased health care expense is financed by borrowing or drawing down on savings, we may overestimate the impoverishing effects of ill health. Overestimation of poverty and catastrophic expenditure effects may also result from our reliance on data from household health care use and expenditure surveys that tend to under-estimate overall household spending, thus lowering measures of household ability to pay. On the flip side, our data do not capture the impoverishing impact of frequent expenditures for outpatient care characteristic of the chronic nature of many NCDs given that for individual households we only have information on outpatient care in the 15 days preceding the survey. Thus, our findings relating to the impoverishing impact of NCDs are subject to these appropriate caveats.