The weird federal-state design of Medicaid has created a natural experiment. A few states have already chosen to expand their Medicaid program to low-income childless adults, like the Affordable Care Act will do in 2014. This allowed researchers to compare the impact these Medicaid expansions have relative to neighboring states that didn’t expand it. The results, published in the New England Journal of Medicine, show increasing the number of people on Medicaid did indeed reduce adjusted all-cause mortality, implying that providing people with basic public health insurance does save lives. From the study:
Our study documents that large expansions of Medicaid eligibility in three states were associated with a significant decrease in mortality during a 5-year follow-up period, as compared with neighboring states without Medicaid expansions. Mortality reductions were greatest among adults between the ages of 35 and 64 years, minorities, and residents of poor counties. These findings may influence states’ decisions with respect to Medicaid expansion under the ACA.
Our study shows a mortality reduction associated with state Medicaid expansions to cover adults. Using state-level differences in Medicaid expansion as a natural experiment avoids the confounding between insurance and individual characteristics (e.g., poverty or health status) that plagues cross-sectional observational studies. These results build on previous findings that Medicaid coverage reduces mortality among infants and children3,4 and are consistent with preliminary results of a randomized, controlled trial of Medicaid in Oregon, which showed significant improvement in self-reported health during the first year (although objective measures of health are not yet available and 1-year mortality effects were not significant and were imprecisely estimated).
Expect this study and others like it to be cited in the coming political fight over whether or not states should expand their Medicaid program under the ACA, now that the Supreme Court has made that decision optional.