One Big Deadly Bill
A groundbreaking study shows how many American lives the Republican megabill is likely to cost.
WILL TAKING HEALTH INSURANCE away from people cause some of them to die prematurely?
It’s an important question to be asking, especially now that Donald Trump and congressional Republicans are about to enact a law projected to result in nearly 12 million Americans becoming uninsured. And it’s long been the subject of serious debate among researchers, partly because the link between insurance and mortality is inherently difficult to measure.
But a series of recent studies have gone a long way toward proving that link. One such paper, which came out in May, was from Angela Wyse at Dartmouth and Bruce Meyer at the University of Chicago, and got the economics establishment talking because of how it used a clever design—linking up census and tax data—to show that the Affordable Care Act’s Medicaid expansion saved about 27,000 lives between 2010 and 2022.
Had the holdout states like Florida and Texas expanded Medicaid during that time, the two scholars found, they could have saved nearly 13,000 more.
The paper landed just as the debate over the Republican bill was getting serious, which gave it lots of attention and provoked some pushback. Among those critical of the paper—or at least skeptical of its import—were the Paragon Health Institute’s Brian Blase and the Cato Institute’s Michael Cannon, two longtime critics of government health care. Blase also served in Trump’s first administration, and has been influential with Republicans working on their legislation this year.
In making their case, Blase and Cannon leaned heavily on what’s known as the Oregon Health Insurance Experiment, which studied what happened back in 2008—i.e., pre-Obamacare—when the state launched a limited expansion of its Medicaid program. The results of that experiment appeared in 2012 and, like the new paper, got a lot of attention because it did not show expanding Medicaid had saved lives or improved physical health.
It was not the first paper to call into doubt the health effects of insurance. But it was particularly arresting because, as Blase and Cannon noted, the Oregon experiment was a truly randomized trial. The limited slots Oregon had for its expansion forced the state to enroll by lottery, enabling the sort of reliable, apples-to-apples comparisons of results rarely possible in a social science setting.
“The OHIE isn’t the final word, but it is the best word,” Blase and Cannon wrote in a June letter to the Wall Street Journal that got wide circulation among policy wonks.
Blase and Cannon—and other critics of Medicaid—have made this point many times before, including in some exchanges with me. (Entirely civil and respectful exchanges, I hasten to add.) They are correct when they say the Oregon experiment did not prove a link with mortality.
But that observation comes with some key caveats.
One is that the Oregon experiment did show significant improvements in both access to care and financial well-being. In other words, the Oregonians who got Medicaid were more likely to get medical treatment and less likely to fall behind on bills and payments.
That is a big deal. Insulating people from the financial burden of medical care is an important goal—some would say the most important goal—of health insurance.
In addition—and here things are going to get a tad wonky—the Oregon study did not have the statistical power to establish a link to mortality. In layman’s terms, that means the sample size wasn’t big enough to show effects on death, which (happily) is a relatively rare event.
None of this is to detract from the high quality or genuine significance of the Oregon experiment, whose insights into a variety of issues researchers still cite.
But on the specific question of mortality there are now so many more studies, including a 2024 paper that used a novel machine-learning approach to re-examine the Oregon results and found evidence of health improvements the original Oregon analysis did not. One of that paper’s coauthors, notably, was also a coauthor of the earlier Oregon study.
AND THEN THERE’S THE WYSE-MEYER PAPER, whose broad data set allowed it to measure things the Oregon study could not. As Yale economist Zack Cooper told me in an interview, after the Wyse-Meyer paper appeared, “This is what happens when the literature changes. If you were anchoring your policy off of Oregon, you wouldn’t have thought mortality changes would occur. Now, the best research tells us that it does.”
Wyse and Meyer wrote a letter to the Wall Street Journal editorial page, making these and other points—and arguing that GOP Medicaid cuts would lead to 10,000 premature deaths a year.
The Journal declined to run it, perhaps because they’d already run a letter from Wyse and Meyer responding to the staff editorial that started the whole discussion. As a public service, I am posting it below.
I think they make a strong case, but you can decide for yourself by reading their letter alongside the letter from Blase and Cannon that did run in the Journal, where they make their case in full. If you want to go deeper, you can read this paper in which two other Paragon authors argue insurance expansions don’t improve health.
After that, you might want to check out this Brookings Institution essay from University of Michigan economists Tom Buchmueller and Helen Levy, in which they summarize the recent literature and make the case for an impact on mortality.1
One last point: Blase and Cannon in their letter said that observational studies were more prone to ideological or political bias than Oregon-style, randomized experiments. But there’s zero reason to think Wyse or Meyer brought that kind of lens to their work. Among other things, Meyer is a nonresident senior fellow at the American Enterprise Institute, a center-right think tank, and has previously criticized Democratic proposals to expand other anti-poverty initiatives.
That doesn’t prove he and Wyse are correct, of course. And there are plenty of reasons why a serious policy thinker might oppose more spending on Medicaid, or support the Republican bill for other reasons.
But the evidence that insurance reduces mortality seems awfully strong at this point, which means there’s awfully good reason to think taking Medicaid away from millions of Americans will cause a significant number of them to die prematurely.
Here’s the Wyse-Meyer letter:
Medicaid and Mortality Once Again
Brian Blase and Michael F. Cannon argue that the Oregon Health Insurance Experiment showed that Medicaid does not reduce mortality, contradicting our study’s finding of fewer deaths in states that expanded Medicaid under the ACA (Letters, June 9). Their letter misrepresents the evidence.
While the Oregon experiment shed light on many important effects of Medicaid, mortality was not among them. With only about 70 deaths among 9,000 new Medicaid enrollees, the study’s mortality estimate was too imprecise to be informative, and in fact was consistent with both no mortality effect and a reduction several times larger than what we found. By contrast, our study’s 3.5 million new Medicaid enrollees averaged about 15,000 deaths per year over twelve years, giving us the statistical power to pick up a signal through the noise.
All studies have drawbacks, including randomized experiments, which can be inconclusive when studying rare outcomes like mortality and may not generalize beyond their specific settings. While not definitive, our study is the largest to date based on individual data and we used accepted causal methods to study Medicaid expansions as implemented nationwide across diverse populations. Our findings suggest that the restrictions proposed by the House of Representatives would lead to roughly 10,000 additional deaths each year.
Blase and Cannon focus narrowly on the Oregon study’s uninformative mortality result while ignoring its clear evidence of Medicaid’s benefits: reduced out-of-pocket spending and medical debt, greater use of preventative care, better self-reported health, and lower rates of depression. The real lesson from the Oregon experiment is not that Medicaid fails to save lives, but that restricting it would damage the financial wellbeing and mental health of millions of Americans in addition to the premature deaths suggested by our study.
Angela Wyse and Bruce D. Meyer
Dartmouth College and U. Chicago
Hanover, N.H., and Chicago