This fact check has been updated with a rebuttal response
This claim from Sanders — that medical bills drive half a million people into bankruptcy every year — relies partly on research from former Harvard professor and now senator Elizabeth Warren (D-Mass.).
Sanders and Warren are seeking the Democratic presidential nomination, both running on a platform that includes universal health care and lower costs for patients. Interestingly, though, Warren doesn’t appear to make the same claim about 500,000 medical bankruptcies per year.
The Sanders campaign told us he was citing a statistic from a public health journal. Critics say the study he’s citing casts too wide a net because it counts anyone who mentioned medical bills or illness among their reasons for declaring bankruptcy, not just those who said it was the main reason or a big piece.
Bankruptcies typically involve multiple causes, and in some cases, medical bills may be a small piece of the pie. Sanders glosses over those nuances, stating that health-care costs drove people to bankruptcy in all 500,000 cases. The study he’s citing doesn’t establish that.
That study, led by David U. Himmelstein, took a sample of bankruptcy court filings from 2013 to 2016, identified 3,200 bankrupt debtors and mailed them a survey. The response rate was 29.4 percent, with 910 responses and 108 surveys returned as undeliverable.
Debtors were asked whether medical expenses, or loss of work related to illness, contributed to their bankruptcies. Of those who responded, 66.5 percent said at least one of those factors contributed “somewhat” or “very much.”
Sixty-six percent of 750,000 is 500,000, so Sanders’s math adds up at first glance.
“The majority (58.5%) ‘very much’ or ‘somewhat’ agreed that medical expenses contributed, and 44.3% cited illness-related work loss; 66.5% cited at least one of these two medical contributors—equivalent to about 530,000 medical bankruptcies annually,” the AJPH editorial says.
This study includes a range of people for whom medical expenses or illness contributed “somewhat” to bankruptcy. What does “somewhat” mean? It’s broad enough to mean “slightly,” “fairly” or “moderately.” Sanders’s claim works only by erasing this ambiguity and taking “somewhat” to mean “mostly.”
The AJPH editorial did not undergo the same peer-reviewed editing process as a research article.
“In AJPH, many editorials are commissioned by the editor-in-chief from experts in their field(s), as a forum to present their most recent or preliminary findings on specific topics, or to coincide with significant dates or events,” said Morgan Richardson, an AJPH editor. “Lack of peer review does not indicate inaccuracy, but editorials are less likely to be cited in the scientific literature as evidence because the standard of rigor is different due to context.”
However, Himmelstein used a methodology similar to what he, Warren and other researchers used in a 2005 peer-reviewed study that they updated in 2009. Warren was a co-author of those two studies, but not the AJPH editorial published in March.
“Illness or medical bills contributed to 62.1% of all bankruptcies in 2007,” according to the study from 2009. (Again, the term used here is “contributed” and not “caused.”)
It’s interesting to note that the rate rose from 62.1 percent in 2007 to 66.5 percent from 2013 to 2016. The Affordable Care Act, a.k.a. Obamacare, was enacted in between the two studies. This line of research thus suggests medical bankruptcies became more prevalent after the law’s passage.
“We did not ask about the sole or main reason for bankruptcy, because our past experience indicates that this is a meaningless question,” Himmelstein, a professor at CUNY’s Hunter College who supports single-payer health care, wrote in an email. “The vast majority of debtors suffer multiple problems that bring them to file, and cannot identify a single problem among them. Thus, if an illness led to lost work time (and hence income) and medical bills, debtors cannot separate out these different problems; they are of a piece.”
Craig Garthwaite, a health-care policy expert in the Kellogg School of Management at Northwestern University, said the study was flawed. “It’s basically saying that if you go bankrupt and you have medical debt, that’s the cause of your bankruptcy,” he said. “That’s not the way you can do this kind of analysis.”
He added: “Rather than looking at a sample of people who go bankrupt and see how many have medical debt, look at a sample of a bunch of people who have medical debt, and how many of them go bankrupt. And that gives you an idea of causality.”
“Based on our estimate of 4 percent of bankruptcy filings per year and the approximately 800,000 bankruptcy filings per year, our number would be much closer to something on the order of 30,000-50,000 bankruptcies caused by a hospitalization,” one of the co-authors of the NEJM study, economist Raymond Kluender of Harvard Business School, wrote in an email.
“This would lead us to be skeptical of the 500,000 medical bankruptcies statistic, but that very much depends on how one defines a medical bankruptcy. … An enormous share of households have some amount of medical debt, so any survey of individuals will report a high share of them have medical debt but this does not imply that the debt caused them to file for bankruptcy.”
Some people could still face high levels of medical debt without ever going through a hospital, and they wouldn’t be counted in the NEJM study.
A Sanders campaign aide wrote in an email: “Medical debt caused by the greed of pharmaceutical and insurance corporations is crippling millions of Americans, and it’s clear that 500,000 is the bare minimum number of bankruptcy filings caused by medical debt each year.”
Linking to the 2009 article by Warren and her co-authors, the Sanders aide added: “Research attributed 62% of bankruptcies to medical problems in 2007 and found them rising since 2001. And according to reports from the federal judiciary, the total non-business bankruptcy appears to hover around 1 million cases annually since 2006. Over 40% of Americans, or 79 million people, are struggling with medical debt, which is why Bernie knows we must pass Medicare-for-all.”
When we asked Himmelstein whether Sanders was quoting his study accurately, he said yes.
Himmelstein wrote: “37 percent of filers said medical bills ‘very much’ contributed to their bankruptcy. Even if you use that restricted definition, then Sanders’s statement is accurate — or an underestimate. There are about 700,000 bankruptcy filings each year. Many filings are joint husband/wife filings, and based on our past research, we estimate that on average 2.71 persons reside in each debtor’s household. So the total number of persons who undergo bankruptcy is about 1.9 million annually.
“37 percent of 1.9 million is a bit over 700,000. Even if you only count the husband and wife in a filing, the number suffering a bankruptcy to which medical bills ‘very much’ contributed is about 500,000.”
But here, we’ve moved from the statistic Sanders cited from the AJPH editorial to a different number that relies on separate inferences. We will note that the study refers to “cases,” not people.
Garthwaite said of Sanders’s claim: “It’s wrong. It’s just wrong. Just because the number’s big doesn’t make it right, even if you want to agree with the premise. And we should be careful about this. I’m not saying that medical debt and bankruptcy is not a problem, but I think we should have a conversation about the appropriate scale of the problem.”
Update, Sept. 3: The Sanders campaign objected to this fact check, as did the key researchers of the AJPH editorial. Please see the response below by Himmelstein and his colleague Steffie Woolhandler. We stand by our Three-Pinocchio rating.
The Pinocchio Test
This is a classic case of cherry-picking a number from a scientific study and twisting it to make a political point.
Sanders’s statements — “500,000 people go bankrupt every year because they cannot pay their outrageous medical bills” and “500,000 Americans will go bankrupt this year from medical bills” — are unambiguous. He’s saying medical debts caused those 500,000 bankruptcies. However, correlation is not causation, and the study he’s citing doesn’t establish causation for all 500,000 bankruptcy cases.
One of the authors sent us rough estimates showing that Sanders might be on target, but those numbers deserve scientific scrutiny before they can be taken as fact.
In the meantime, the statistic Sanders’s campaign cited includes bankrupt debtors for whom medical expenses may have been a minor or relatively small contributing factor. A different, peer-reviewed study arrived at a much different conclusion, suggesting the medical bankruptcy rate is far lower, although it measured only hospital patients and not all types of medical debt.
The omissions and twists are significant enough to merit Three Pinocchios for Sanders.
Send us facts to check by filling out this form
Response by Himmelstein and Woolhandler
An August 28 Fact Checker article in the Post assigned a “Three Pinocchios” rating to Sen. Bernie Sanders’ statement that 500,000 Americans are bankrupted by medical bills annually. Sanders’ estimate relied on an editorial by David Himmelstein and colleagues in the American Journal of Public Health (AJPH) reporting findings from a Consumer Bankruptcy Project (CBP) survey that asked debtors about causes of their bankruptcy. The editorial (which the Post falsely implied had not undergone peer review) updated previous CBP studies carried out by Himmelstein, along with then-Harvard Law Professor Elizabeth Warren, Steffie Woolhandler and Deborah Thorne that reached similar conclusions and appeared in leading medical and policy journals.
The Post’s denigration of Sanders’ statement rests on an econometric study that found only a modest uptick in bankruptcy filings among persons hospitalized in California between 2003 and 2007. As Himmelstein, Woolhandler and Warren noted in their response to that study in the New England Journal of Medicine, the study excluded most persons with frequent hospitalizations (a group at high risk of medical bankruptcy); assumed that anyone not hospitalized could not suffer medical bankruptcy; that a child or partners’ illness couldn’t lead to bankruptcy; and that potentially bankrupting illnesses always commence at the moment of hospitalization - an assumption contradicted by the study’s own data.
Rather than checking facts, the Post has chosen one side in an ongoing and unsettled scholarly debate, and labeled those on the other side (and public figures who cite their research) “liars”.
Hunter College at City University of New York
Lecturer in Medicine, Harvard Medical School
Steffie Woolhandler MD MPH
Hunter College at City University of New York
Lecturer (formerly Professor) in Medicine, Harvard Medical School