Effects of Medicare Coverage for the Chronically Ill on Mortality

The effect of health insurance on health remains controversial, despite numerous observational studies, natural experiments, and two randomized trials. Prior work has attempted to detect effects on health status using unselected and small, relative to plausible effect sizes, samples. As a result, there has been a paucity of studies with precise estimates on the effects of health insurance in any direction. An alternative to the previous unselected studies is to consider individuals with specific diseases that are associated with a high probability of mortality and may be affected by insurance coverage.

In this paper, I study individuals with end-stage renal disease who received an exogenous change in health insurance status in 1973 as a result of a 1972 law extending Medicare coverage to individuals with end-stage renal disease (ESRD). ESRD is a costly disease with a high mortality rate, as a result it is plausible for insurance to have large effects on mortality by facilitating access to highly effective, but previously unaffordable treatments. Using data on the number of deaths, by cause, year, and age, I am able to identify the causal effect of the Medicare expansion on mortality related to end-stage renal disease among people under 65, relative to individuals over 65, before and after the Medicare expansion took effect.

I find large and statistically significant reductions in mortality following the implementation of the program. In a triple-difference specification comparing deaths due to kidney diseases to all other causes, before versus after the policy change, and over versus under 65 (the standard Medicare eligibility age) I find that the program reduces mortality by 30%. Using alternative difference-in-difference specifications that restrict to deaths due to kidney disease or deaths among individuals under 65 yield mortality reductions of 8-10%, rather than 30%.

The reductions in mortality are mirrored by increases in Medicare insurance coverage, which was one of the few insurance plans in the 1970s providing coverage for dialysis and kidney transplantation, the two primary treatment modalities for end-stage renal disease. I argue that my results are plausibly related to the insurance expansion because when I allow for heterogeneous effects on mortality and insurance coverage I find that groups with larger mortality reductions are also those with larger increases in Medicare coverage. These results are further reinforced by the fact that in states that implemented Medicaid programs earlier, which also provides insurance coverage for dialysis and kidney transplantation, the effect of the ESRD expansion was substantially smaller than in states with later insurance expansions.

A related set of mechanisms relates to the hypothesis that insurance provides access to expensive treatments. To test this hypothesis, I study the differential effect of the ESRD expansion as a function of the number of dialysis clinics and kidney transplantation programs in the state of residence. Areas with more dialysis clinics, relative to population, were also areas that experienced larger declines in kidney mortality after the policy was implemented, suggesting that access to facilities was an important contributor to the reduction in mortality.