The Agenda

Does Not Expanding Medicaid Kill People? Probably Not

Earlier this week, I explored the question of the divergent outcomes between high-quality studies about Medicaid expansion in Oregon and private-insurance expansion in Massachusetts. The former didn’t appear to improve physical health, while the latter seemed to reduce mortality.

With a number of states still considering the weighty question of whether to expand Medicaid under Obamacare, can we expect Medicaid to improve health outcomes?

Some research indicates the answer is a definitive “no.” A 2012 paper from the Manhattan Institute’s Avik Roy gives a nice run down of quite a few clinical studies that indicate Medicaid patients don’t fare, by several measures, any better than uninsured patients. In fact, one study showed that Medicaid patients more likely to die after surgery than uninsured patients.

Meanwhile, though, Democrats are declaring that states’ deciding not to expand Medicaid under the Affordable Care Act means thousands of deaths that could be otherwise avoided. “Thousands Of People Will Die In States That Don’t Expand Medicaid,” blared one headline. This was based on a recent article published in Health Affairs that predicted that 7,115 and 17,104 people will die in opt-out states for lack of Medicaid coverage.

Seventeen thousand per year is a big number — that’s roughly the same number of Americans who will die from homicide this year. Big claims deserve a healthy dose of skepticism, and in this case, even a cursory examination of the one piece of evidence Medicaid advocates cited reveals serious issues with this argument.

The source the Health Affairs authors used to estimate the high-end mortality effects of not expanding is problematic. It’s a study by Harvard economists (one of whom is Ben Sommers, who also authored the Massachusetts study) who concluded that Medicaid expansion results in a significant decrease in mortality based on comparisons between three states that substantially expanded Medicaid and four neighboring states that did not.

The combined data from the three expansion states indicates that Medicaid reduces mortality (by 19.6 deaths per 100,000 adults). However, looking at each state’s data individually yields a different conclusion.

Of the three expansion states (New York, Maine, and Arizona), only New York’s and Arizona’s mortality rates were lower relative to rates in their respective control states. Only New York’s decrease was statistically significant. In Maine, mortality rates actually increased after Medicaid expanded. Avik Roy and Linda Gorman have written about important demographic differences and other confounding variables that complicate the statistically significant positive findings in New York.

The evidence that Medicaid improves certain health outcomes, which Austin Frakt writes about here, is more favorable. The Oregon Health Insurance Experiment, a randomized control trial that produced no evidence of physical-health improvements, found that Medicaid recipients get more care that is linked to better health outcomes and had improvements in depression, cholesterol, and blood pressure.

Confounding variables make the impact of Medicaid and health insurance on health outcomes extremely hard to measure accurately, especially with observational studies — like the Health Affairs study that resulted in such huge differences in mortality rates. It seems quite unlikely, as some observational studies suggested, that Medicaid makes one’s health outcomes worse; instead, people who choose to get health insurance (and who are eligible for Medicaid) are more likely to display various morbidities that hurt their health outcomes. Unfortunately, we also know that the only RCT on the subject, the Oregon study, has lots of problems that make its results of no physical-health improvements inconclusive — still a mark against Medicaid, but not a definitive one.

So where does this leave us? In general, we think health insurance is a good thing. If we didn’t, America wouldn’t be spending over trillions of dollars on it every year. But we also know that it’s unclear how much health insurance, especially Medicaid, will improve outcomes.

The U.S. outspends other wealthy countries on health care by a mile without much better outcomes, which means, much as we value insurance, we ought to be thinking pretty hard about how best to spend any new health-care dollars. It is doubtful that the cost of not expanding Medicaid is thousands of deaths each year, though we can’t know for sure. What isn’t doubtful is that expanding Medicaid will be extremely costly — to federal taxpayers and eventually possibly to each state — as will be expansion of private insurance. Given the known costs and projected benefits, it’s possible private insurance, not Medicaid, is the comparatively worthwhile choice for expanding coverage. It’s also possible, as Chris Conover has suggested in this space, that expanding private insurance is too expensive even if it saves lives, and that our health-care dollars are better spent on other interventions.


The Latest