There’s no shortage of empirical debates in health policy, but few have been so fraught as the question of whether health insurance actually delivers health gains (see here, here, and here to start). So, you’re going to read a lot about an Annals of Internal Medicine study released on Monday that found significant mortality reductions following Massachusetts’s health reform. (You should read Austin’s accompanying editorial, too.)
Benjamin Sommers, Sharon Long, and Katherine Baicker estimate that overall mortality in Massachusetts declined 2.9 percent relative to control counties between 2007 and 2010; mortality amenable to health care declined 4.5 percent. This translates to one death prevented for every 830 people who gain insurance, and the effects were larger in counties with low income and low pre-reform insurance rates—the counties we would expect to be most favorably impacted by reform.
Massachusetts isn’t a randomized controlled trial—we will never have a RCT for insurance expansion on that scale—but there are other ways to exploit the state’s health reform as a natural experiment. The authors conducted a difference-in-differences analysis; using propensity scores they were able to identify “control” counties that mirrored Massachusetts demographically and economically prior to reform. In addition to controlling for the usual suspects (age distribution, race/ethnicity, poverty, income, unemployment, and uninsurance) the model matched along baseline mortality rates.
The findings aren’t bulletproof, but they’re close. The study’s penultimate paragraph captures this well:
Most important, our quasi-experimental approach cannot definitively demonstrate a causal relationship underlying the association between the Massachusetts reform and the state’s declining mortality relative to other states. It is possible that the postreform reduction in mortality in Massachusetts was due to other factors that differentially affected Massachusetts, such as the recession. However, our analysis controlled for several distinct time- and county- specific economic measures. We also found no evidence of a similar decline in mortality among elderly adults in Massachusetts that would suggest a secular trend. Although we cannot rule out unmeasured confounders, it is challenging to identify factors other than health care reform that might have produced this pattern of results: a declining mortality rate in Massachusetts since 2007 not present in similar counties elsewhere in the country, primarily for health care–amenable causes of death in adults aged 20 to 64 years (but not elderly adults), concentrated among poor and uninsured areas and not explained by changes in poverty or unemployment rates.
Past investigations of how health insurance impacts health offered conflicting results, but have also faced limits to external validity. Policy changes tend to only impact one payer (Medicare, Medicaid, VA) or population (elderly, children, HIV+ patients) at a time—these are the vagaries of reality. Massachusetts expanded coverage in a much broader sense: the state extended Medicaid eligibility to those under 100% FPL and subsidized the individual market coverage for residents below 300% FPL. There was also saw a small hike in take-up of employer-sponsored insurance.
If you think the study’s primary findings are impressive, consider their implications: “mortality amenable to health care” does not just magically decline. If fewer people are dying, that is almost certainly because diseases are being better treated, managed, or prevented—because of improved health. It’s hard to come by data on objective measures of health at the state level, but the “improved health” story is consistent with other findings in the paper: individuals had better self-reported health, were more likely to have a usual source of care, received more preventive services, and had fewer cost-related delays in care.
This study only looked at outcomes for 2007 through 2010, four years following reform. Assuming that health improved in Massachusetts—the most logical vector for reduced mortality—not all of those gains are going to be captured in a “prevented deaths” metric. That is, reduced mortality implies gains in health, but not all gains in health would manifest as reduced mortality. Austin nails it in his editorial:
Although some health care does not substantially improve health, much of it does. Thus, the conclusion that coverage expansion leads to health benefits by facilitating access is eminently reasonable. What is unreasonable and, in my view, unconscionable is to leverage a selective reading of the evidence on the benefits of health insurance in an argument to deny assistance to Americans who cannot afford to purchase basic coverage.
Sometimes evidence is easy to argue with. This isn’t one of those times.