Demonstrators protested in May before a town hall meeting with Rep. Tom MacArthur (R-N.J.) in Willingboro, N.J. (Dominick Reuter/AFP/Getty Images)

There’s no real question that if the Affordable Care Act is repealed the number of Americans without insurance coverage will increase. According to the Congressional Budget Office, the percentage of those under age 65 who can’t afford or opt out of coverage would jump to 18 percent by 2026 under the Republican health-care bill under consideration in the Senate, with the total number of uninsured jumping from about 28 million (under Obamacare) to 49 million — an increase of about 22 million. The size of that increase is subject to debate, but there’s no real question it will be greater than zero.

Where the debate arises is around the effects of that increase in the number of the uninsured. What happens to 22 million people — or 1 million people — when they lose health insurance? How many more of them are likely to die?

This is an important policy point that’s become a rhetorical one. Sen. Bernie Sanders (I-Vt.) said that “thousands of people will die” under the Republican bill; Sen. Orrin G. Hatch (R-Utah) replied with sarcastic nostalgia for the “brief time when we were *not* accusing those we disagree with of murder.” Hillary Clinton called the GOP the “death party,” and evoked a ferocious response.

The available evidence suggests that there will be a human toll from an increase in the number of uninsured. But that number is hard to pin down.

One key reason is obvious: There are serious ethical questions about running an experiment in which people are denied insurance in an effort to determine how much more quickly they might die. We’re left with a number of studies that try to approximate the answer to the question by using inadvertent experiments along those lines (such as when Oregon added a number of people to Medicaid using a lottery) or by comparing pools of people living in places with different insurance rules (like those in states that did or didn’t expand insurance coverage).

This month, the Annals of Internal Medicine published a meta-analysis of recent studies aimed at determining how a lack of insurance coverage affects mortality. Some of the studies, like the one in Oregon, yielded statistically insignificant results. Overall, though, the researchers found that “[t]he evidence strengthens confidence in the Institute of Medicine’s conclusion that health insurance saves lives: The odds of dying among the insured relative to the uninsured is 0.71 to 0.97.”

What does that mean in practical terms? “Roughly between a 3 percent and a 29 percent higher risk of dying for the uninsured than for comparable insured people,” David Himmelstein, one of the study’s co-authors, explained to us on Tuesday. “The majority of the evidence clusters roughly in the middle of that, so something like a 17 percent, 16 percent increase in risk of dying.”

For every 800 people who are uninsured for a year, existing studies suggest that there would be one death, Himmelstein said. For those 22 million the CBO estimates would lose coverage under the Senate bill — 28,600 excess deaths. And that, Himmelstein said, is the middle of the range of estimates.

Himmelstein and his co-author, Steffie Woolhandler, served as unpaid advisers to Sanders’s presidential campaign and are advocates for a single-payer health-care system. When The Washington Post gave Sanders four Pinocchios in January for his claim that 36,000 people would die annually if Obamacare were repealed, the Himmelstein and Woolhandler offered a rebuttal, noting an existing study that found one life saved for every 455 people who gained insurance coverage. The number included in their meta-analysis is more conservative.

Mortality is also one of the negative effects of a loss of insurance addressed in analysis published this month in the New England Journal of Medicine. In that report, Benjamin Sommers, Atul Gawande and Katherine Baicker address the full sweep of those negative effects. They cite previous analysis from Sommers and Baicker (that’s also included in the Woolhandler-Himmelstein analysis) to estimate that one life is saved for every 830 people who gain insurance. They also note that the negative effects of gaps in coverage can be long-term, making it tricky to analyze how loss of insurance affects mortality.

In May, Chandler McClellan, a government health economist, published a study looking at the effects of the Obamacare expansion to people under age 26 (who, when the Affordable Care Act was passed, were able to retain insurance on their parents’ plans). “Reduction in mortality,” McClellan  found, “was primarily in disease-related causes which are amenable to general medical care such as cardiovascular disease, while mortality due to trauma-related causes, which must be treated regardless of insurance status under preexisting laws, was unaffected.” About 357 fewer Americans aged 19 to 25 died every year under the provisions of the ACA, about one a day, he said.

It’s certainly the case that a hard number on the tally of deaths from loss of insurance is tough (and perhaps not even possible) to determine. In broad strokes, Sanders’s assessment that thousands more would die annually appears to be supported by the data: If some 800,000 people lack insurance in a given year, about 1,000 would be expected to die. Under the Senate bill, 15 million more people would be uninsured next year alone, according to the CBO.


“The thing that was most striking to us was the strength of scientific consensus on this issue,” Himmelstein said. “The medical scientific literature is really quite clear at this point and quite in agreement that there is a significant mortality effect. Exactly how big? There’s still considerable uncertainty around that.”

But, he said, “it’s quite clear that there are deaths from people being uninsured — and a substantial number.”