By Josh Zeitlin, Editor
Does health insurance save lives?
That question is more pressing than ever as Congress considers new health care reform bills—and it's one that Benjamin Sommers, Atul Gawande, and Katherine Baicker just sought to answer as part of an analysis published Wednesday in the New England Journal of Medicine.
They concluded that the "body of evidence ... indicates that coverage expansions significantly increase patients' access to care and use of preventive care, primary care, chronic illness treatment, medications, and surgery," which "appear to produce significant, multifaceted, and nuanced benefits to health," some of which "will ultimately help tens of thousands of people live longer lives."
They cited three pieces of research in particular on the relationship between insurance coverage and mortality:
- One study found that for every 830 adults who gained coverage, one life was saved per year, based on comparing counties that gained coverage under Massachusetts' 2006 health reform with demographically similar counties nationwide;
- A separate study found that expanding Medicaid coverage to childless adults without disabilities was associated with a 6 percent decrease in mortality over five years of follow-up, based on comparing three states that expanded their programs with neighboring states that did not; and
- A subsequent analysis of that study found that for about every 300 adults who gained Medicaid coverage, one life was saved per year.
It could be tempting to take that data, as the Center for American Progress (CAP), Gawande, and others have done, and apply it to the debate over the Senate GOP health care bill, given that CBO estimates the Better Care Reconciliation Act (BCRA) would increase the number of uninsured Americans by 22 million by 2026. But is it really fair to say the bill would lead to more deaths? And is it fair to say, as CAP has estimated, that it would cost upwards of 10,000 lives a year?
What 10 researchers say
I put those questions to 10 researchers who have written about insurance coverage and mortality. The vast majority said that it would be fair to say the BCRA as currently written would cost lives—although some declined to estimate how many, and others said the evidence is too uncertain to say one way or the other.
Some say it's reasonable to project thousands to tens of thousands more deaths per year
Sommers—an author of the NEJM analysis, an internist, and an associate professor of medicine at Brigham & Women's Hospital and Harvard Medical School—stressed it is "difficult to make a precise prediction," since the strongest studies haven't examined the Affordable Care Act's expansions and because the effects of coverage reductions might not mirror the effects of coverage gains.
But based on the available evidence, Sommers said the BCRA would likely increase deaths over time, and that it's "reasonable" to project the effects of the bill based on past research. If for every 300 to 800 adults who gain coverage, one life is saved per year, that "puts the number of potential deaths in the ballpark" of more than 10,000 a year, he said.
Two other researchers also said the BCRA could increase the number of deaths by thousands per year.
Janet Currie—a Princeton University researcher who has studied the effects of Medicaid coverage on infant and child mortality—told American Health Line that "by undoing the Medicaid expansions, the BCRA would lead to more uninsured children and higher death rates among children, perhaps thousands of unnecessary deaths." She said she's persuaded by studies that showed a reduction in child mortality after coverage expansions between the late 80s and early 2000s.
And Stephanie Woolhandler, a CUNY Hunter College researcher who co-authored a review published Monday in Annals of Internal Medicine of about 50 studies on the subject, said that "acquiring health insurance reduce the odds of dying by between 3 percent and 29 percent." Woolhandler told American Health Line that based on the midpoint of those estimates, the Senate health care bill would result in more than 25,000 additional deaths per year.
Others decline to estimate how much BCRA might increase mortality
However, several other researchers who spoke with American Health Line declined to estimate the degree to which the BCRA might affect mortality.
Laura Wherry of UCLA and Andrew Goodman-Bacon of Vanderbilt both contended that the BCRA would increase morality given its major changes to Medicaid. There are "a large number of rigorous studies [that] have demonstrated that Medicaid decreases mortality for both children and adults," Wherry said, while Goodman-Bacon suggested there is "a lot of evidence" suggesting "that health insurance, and Medicaid specifically, reduces mortality." Goodman-Bacon and Wherry both cited work by Currie and Jonathan Gruber on the effects of Medicaid coverage on pregnant women and children, along with their own respective research.
But Goodman-Bacon and Wherry both cautioned that it's hard to estimate the precise effects of the BCRA given that states may respond to reductions in federal funding in varying ways.
"When you combine uncertainty over 50 states' different policies with the wide variation in Medicaid's effects across sub-populations, it leads to a range of plausible mortality estimates," Goodman-Bacon stressed. "That said, uncertainty about deaths does not mean zero deaths. The argument that 'no one would die' is not empirically supported."
Part of the difficulty in measuring health insurance's effect on reducing mortality is that randomized controlled trials (RCTs)—the gold standard in research, in which individuals are randomly assigned to a control group or an intervention group—on the issue are rare. Longer-term benefits of health insurance also might "take years to manifest" and therefore aren't fully captured in some studies, Harvard researcher John Michael McWilliams told American Health Line.
Those who say health insurance doesn't improve mortality often point to a RCT comparing about 10,000 people who gained Medicaid coverage in Oregon via a lottery with about 10,000 people who did not. The Oregon study didn't find a statistically significant effect on mortality.
Baicker—who co-authored that Oregon study—Sommers, and Gawande addressed the question directly in their NEJM piece, writing that the study's "sample and duration were not well suited to evaluating mortality," since death is a statistically rare outcome among the relatively young individuals who received coverage.
McWilliams said that health insurance "has never been randomized on a large enough scale to settle the debate," so we "can't measure the magnitude" of its effects very well, including on mortality." But he said the lack of consensus evidence shouldn't lead policymakers to "let perfect by the enemy of the good."
"We know that expanding coverage does a lot of good things for the uninsured, so coverage is bound to have a beneficial effect on health outcomes," he told American Health Line. "And it seems very odd to acknowledge that insurance improves access to a lot of the things that keep you alive but then argue that it has no effect on mortality."
Some say the evidence is too uncertain and 'drowns out' other debates
Other researchers have pushed back against the idea that insurance necessarily decreases mortality. Bernard Black, a Northwestern Pritzker School of Law professor, told Reuters on Monday that "we don't know" if there's an effect based on the available evidence.
Oren Cass, a senior fellow at the Manhattan Institute, told American Health Line that "the only fair conclusion is that we have no statistical evidence to support the claim that" the BCRA would increase mortality because of its Medicaid changes.
Cass said his views are most informed by the Oregon Health Insurance Experiment; his understanding that no study has found a consistent positive effect of Medicaid coverage on mortality; a Health Affairs study that found states with higher ratios of social to health spending had better subsequent health outcomes; and research that has found increases in U.S. mortality in recent years, including his findings that mortality has increased at higher rates in Medicaid expansion states.
Separately, Sommers told American Health Line that it's "most likely too early to know whether the ACA is impacting mortality, given the data available."
Meanwhile, Tom Miller, a resident fellow at the American Enterprise Institute, told American Health Line that suggestions health insurance reduces mortality are "pretending to prove more than" the evidence can support. With all the confounding variables and a lack of RCTs, it's difficult to "really prove the point one way or another," he said.
The available research, Miller said, has proven that "you can get more health care services if someone else pays for them. But what those health care services produce is a more complex question" and one that involves tradeoffs. Miller said the research "doesn't say is if you over-invest in that area, what are you not investing in along the way which also might have shaped your helping your overall well-being?"
That debate over tradeoffs, Miller said, "gets all drowned out when you're simply talking about being for against killing people or having people live. And that's the shortcomings of our increasingly heated political discourse."
Charles Courtemanche, a researcher at Georgia State University and co-author of a recent paper on the effects of the ACA on self-assessed health, also told American Health Line that there "is very little convincing evidence that health insurance reduces the risk of mortality for non-elderly adults, which is the relevant population for the ongoing policy debate." However, he cautioned that doesn't mean there's conclusively no effect—just that studies haven't captured one.
In particular, Courtemanche, Miller, and Cass all raised questions about the studies cited in the NEJM analysis on three states that expanded their programs to childless adults without disabilities. They noted that the study authors found the effect on mortality was mostly or purely driven by one state, New York, so any effect may not apply to the entire United States.
Courtemanche said that "there's more convincing evidence that Medicaid saves infants' lives," but that the BCRA wouldn't directly affect that population. "That said, conceivably in the long run switching to block grants could induce individual states to trim coverage for kids," he added.
Insurance is about more than life and death
For now, experts continue to debate the extent to which health insurance may reduce mortality—meaning that lawmakers will have to weigh conflicting views as the Senate health care bill moves toward a possible vote.
But the evidence is clearer on one point: Insurance has effects far beyond preventing deaths. The Oregon Medicaid study, for example, found that individuals who gained insurance were more likely to use outpatient and emergency care—and far less likely to face catastrophic medical expenses.
As Courtemanche said, "Even if there is no effect on mortality, that doesn't necessarily mean that coverage expansions are a waste of money. There are many other motivations, from providing financial protection to vulnerable populations to improving health in more subtle ways that aren't life-saving. The evidence on these outcomes is somewhat stronger."
As lawmakers move forward, they'll have to consider not just whether insurance saves lives but also how to value those subtler, but still potentially life-changing, effects.