The Congressional Budget Office projects that if the Senate Republicans’ health care bill becomes law, 14 million Americans will lose their health insurance in 2018, and, by 2026, 22 million would lose coverage.
Drawing on that work, we estimate that if the Senate bill becomes law, 22,900 excess deaths will occur in 2020 — and the figure will grow over time. 26,500 extra deaths will take place in 2026. Over the next decade, we estimate that a total of 208,500 unnecessary deaths will occur if the law is passed (see Table 1).
We also calculate anticipated additional deaths, state by state, using state-level coverage losses for the year 2026 (see Table 2). The predicted excess deaths by state range from 30 in North Dakota to 2,992 in California in 2026 alone.
Some commentators have argued that it’s inappropriate — beyond the pale — to suggest that people will die as a result of this legislation. To the contrary, we contend that no debate over a health care policy can ignore evidence that it could have negative effects on health and mortality.
In making these calculations, we draw on the scientific literature demonstrating that expanding health insurance reduces deaths. We specifically apply the results of a particularly robust study of the effects of health care reform in Massachusetts on mortality. Massachusetts’ health care reform — which expanded Medicaid, offered subsidized private insurance, and included an individual mandate — famously served as a model for the ACA. The Massachusetts study looked at county-level mortality data in 2001 to 2005 (pre-reform) and 2007 to 2010 (post-reform), and compared the changes to carefully selected control groups in other states that had not enacted health reform.
For every 830 individuals insured, the authors found, one life was saved. In medical terms, 830 in this context is the “number needed to treat.” To put this into perspective, the colonoscopy number needed to treat is 1250; you need to conduct 1250 colonoscopy screenings to prevent one colorectal cancer death.
Overall, in Massachusetts, insurance coverage expansion was associated with a 3 percent decline in mortality from all causes, and a 4.5 percent decrease in deaths from causes that are especially amenable to being prevented by health care — including heart disease, infection, diabetes, and cancer.
Assuming that one death will occur for every 830 people who lose coverage is the same methodology used by the White House Council of Economic Advisers under the Obama administration to calculate reductions in mortality associated with the ACA.
Deaths due to reductions in health insurance under Senate health bill
|Year||Coverage reduction (in millions of people)||Excess deaths|
|Year||Coverage reduction (in millions of people)||Excess deaths|
This methodology admittedly has limits. There are demographic and health workforce differences between Massachusetts and other states — the Massachusetts population is whiter, older, and more female, and has a higher per-capita physician rate than national averages.
We also assume that losing insurance has the equal and opposite impact of gaining insurance; existing research focuses on the effects of gaining insurance. Further, we look only at mortality effects of insurance losses. It is possible that there are negative mortality effects of losing certain insurance benefits, or being required to pay higher premiums or deductibles, but we did not explore those.
Some conservatives argue that health insurance — Medicaid in particular — has little to no effect on mortality
Some conservatives have pushed back against this methodology — most recently the policy writer Oren Cass, who argued in the National Review and in a paper for the Manhattan Institute that the Affordable Care Act (ACA) has saved “zero” lives — and therefore argued its repeal would not influence health or mortality. The logic underpinning this highly counterintuitive argument is that the ACA insurance gains were primarily due to Medicaid expansion and that Medicaid does not improve health. That’s a claim, however, that can only be made by selectively citing some studies in this area (indeed, the weaker ones).
Skeptics of the effectiveness of the ACA further argue that Massachusetts health reform was primarily driven by private insurance expansion, so experiences from Massachusetts cannot be applied to the ACA. That doesn’t hold up either.
To begin with, Cass’s analysis depends upon drawing a stark distinction between Medicaid and private insurance. But today, nearly 80 percent of Medicaid enrollees receive their care through Managed Medicaid plans, which are run by private insurance companies and function like private plans. This considerably weakens the argument that there is a fundamental difference between Medicaid and private insurance.
And while the simple majority of new coverage gains under the ACA were through Medicaid, it’s not true that Medicaid accounts for all the coverage gains. Of the 20 million newly insured Americans who gained coverage through the ACA, 60 percent became insured through Medicaid and 40 percent purchased new coverage through subsidized private insurance. Recent analysis shows that these gains in subsidized private insurance are primarily due to the ACA, and not to economic recovery, as Cass claims.
The national patterns are not so different from what those were in Massachusetts. There, 47 percent gained insurance through Medicaid expansion, 39 percent did so through subsidized private coverage, and the balance gained insurance through their employer, through individual purchase outside of an exchange, or through a smaller children’s health program.
State by state estimates of excess deaths relative to the ACA, in 2026
|State||Coverage reduction||Excess deaths|
|State||Coverage reduction||Excess deaths|
|District of Columbia||41,200||50|
Debates over the effectiveness of Medicaid are central to the conservative pushback against mortality statistics like the ones we have produced. In exploring this debate, it’s important to recognize that not all of the studies of Medicaid’s effectiveness are equally valid. “Cross-sectional” studies, for example, which basically look at a snapshot of health in different populations at a given moment, are considered the weakest.
That’s because there are many social and economic differences, going well beyond health insurance status, between the uninsured, people on Medicaid, and people with private insurance. For instance, a person can lose their private employer health insurance because they lose their job — and unemployment may affect their health, independent of insurance status. Cross-sectional studies are not as effective as other approaches in accounting for these social and economic differences.
Not all studies are created equal. Public health researchers know this.
Cross-sectional studies of Medicaid, as Vox’s Julia Belluz has noted, have come to very different conclusions. One, which tracked 640,000 respondents to a health survey from 1986 to 2002, found that lacking health insurance did not lead to an increase in mortality. But another study, using a similar approach but different data, found that a lack of insurance led to nearly 45,000 deaths a year.
The author of the first study, Richard Kronick, of UC San Diego, told Belluz: “The primary conclusion I would draw is that it is not possible to have much confidence in the results of the sort of observational analysis used in my study” — or in the other, similar one. (Even Cass acknowledges that Medicaid saves the lives of pregnant women and children, but they were already covered by the program when the ACA was passed.)
Randomized control trials and so-called natural experiments are far more robust than cross-sectional studies.
The best-known randomized control study of Medicaid is the Oregon Health Insurance Experiment (OHIE), made possible by a limited expansion of that state’s Medicaid program, in 2008. It compared low-income Oregonians who were enrolled into Medicaid via a lottery to those, equally eligible, who were not enrolled.
Participants in the study were interviewed and evaluated after two years. ACA critics like to point to the following findings: Medicaid did not produce a statistically significant improvement in three clinical biomarkers associated with health: blood pressure, cholesterol, or glycated hemoglobin (a measure of diabetes progression). On the other hand, gaining Medicaid was associated with improved self-reported mental health, greater treatment and diagnosis of chronic conditions, and decreased depression. (Depression, let’s not forget, is the leading cause of disability globally.)
However, the authors themselves note that the length and size of the study did not allow for accurate examination of mortality effects.
Two especially persuasive studies of Medicaid’s effectiveness
In addition to the Massachusetts study, there have been two other studies that made use of a natural experiment in Medicaid enrollment. Both compared the health of residents in three states that substantially expanded Medicaid eligibility before the ACA — New York, Maine, and Arizona — with health in neighboring states that did not. Both found a 6 percent decline in mortality. Deaths from those “healthcare amenable” diseases — heart disease, infection, diabetes, cancer, and other conditions — dropped even more. The second study found that mortality changes tracked closely with insurance changes at the county level, underscoring that it was health insurance expansion that drove mortality declines.
Whereas the Oregon study lasted only two years, the Medicaid studies and the Massachusetts study looked at four to five years of data. Many public-health researchers believe the effects on mortality would take longer to reveal themselves than two years. And while the OHIE included 10,000 people, the other three studies looked at 150,000 to 500,000.
In sum, three particularly robust studies have found mortality declines when Medicaid is expanded, or when both private and public insurance is expanded. (We further discuss our methodology in calculating excess deaths here.)
The claim that private insurance improves health, but Medicaid does not, is especially tenuous. Indeed, there is a much greater body of literature that demonstrates the positive effects of Medicaid than exists about private insurance. The only two studies we know of that directly compare the effects of Medicaid and private insurance found minimal differences.
And from the patient’s point of view, surveys show the experience of Medicaid and private insurance is very similar.
To be sure, our estimates of additional deaths caused by the passage of the BRCA are not beyond criticism, but they are solid estimates firmly rooted in scientific evidence — unlike the dubious claim that the ACA has saved “zero” lives.
Ann Crawford-Roberts, MPH, is a medical student at the Icahn School of Medicine at Mount Sinai and a graduate of the Harvard T.H. Chan School of Public Health. Nichole Roxas, MPH, is a medical student at the University of Rochester School of Medicine and Dentistry and a graduate of the Harvard T.H. Chan School of Public Health. Ichiro Kawachi, MB.ChB, PhD, is a professor of social epidemiology and the chair of the Department of Social and Behavioral Sciences at Harvard T.H. Chan School of Public Health. Twitter: @acrawrob.
The Big Idea is Vox’s home for smart discussion of the most important issues and ideas in politics, science, and culture — typically by outside contributors. If you have an idea for a piece, pitch us at email@example.com.