|
PERSPECTIVEDiscounting The Debtors Will Not Make Medical Bankruptcy Disappear
David Dranove and Michael Millenson seem determined to deny that financial fallout from illness pushes middle-class families into bankruptcy. Anxious to erase the headline that three-quarters of U.S. medical bankrupts had health insurance at the onset of their illnesses and the resulting spotlight on inadequate coverage and insurance cancellation practices, they ignore most of our data and misrepresent the rest. They dismiss families’ explanations of their difficulties and blame those ruined by illness for their own problems. However, the data from the bankruptcy courts are undeniable. Bankruptcies affect mainly middle-class, privately insured families, and about half are triggered, at least in part, by illnesses.
IN REVISITING OUR STUDY on medical bankruptcy, David Dranove and Michael Millenson arbitrarily ignore large portions of bankrupt debtors’ own reports of their financial problems, misrepresent the previous literature on medical bankruptcy, and castigate bankrupt families with accusations of profligacy and allusions to biblical retribution.1 In their world, middle-class families have little problem affording care; only a handful are bankrupted by illness; and health insurers (who paid for their study) are doing a bang-up job.
Let’s be clear: Every number, percentage, and variation in how to interpret the data that Dranove and Millenson seize upon in attempting to dismiss our work came from our manuscript. We offered detailed breakdowns of our numbers so that different readers could weigh the data. Unfortunately, these commentators manipulate the data far beyond legitimate reinterpretation. Their central claim is that no more than 17 percent of bankruptcies are medical—far lower than the 54.5 percent figure they attribute to us (in fact, we gave a range of estimates, from 46.2 percent to 54.5 percent, in anticipation of different possible interpretations of the data). In their effort to whittle down the number of medical bankruptcies, they ignore huge chunks of the data we reported. First, they assert that only the 28.3 percent of debtors who gave "illness or injury" as the specific reason for filing for bankruptcy could possibly be classified as "medically bankrupt." Next, they narrow this group to include only the 59.9 percent who also cited bills from medical providers as a major contributor to bankruptcy. (These latter data came from our interviews with a subset of 331 debtors, not, as Dranove and Millenson state, from our questionnaire survey of 1,771 filers.) Thus, their calculation rests on responses to two items from our data, disregarding the hundreds of other pieces of data we collected that give a fuller and more accurate picture of the reasons for filing for bankruptcy. Moreover, they misrepresent both items. They seem unfamiliar with the complexity that is part and parcel of medical bankruptcy. For example, some bankrupt families who took out second mortgages to pay medical bills described their reason for filing as "to save our home." Others cited "birth," "death" or "car accident," even when their other responses revealed unaffordable medical catastrophes such as premature deliveries, birth defects, spinal injuries, or expensive terminal illnesses. Some struggled with lawsuits from hospitals or from credit card and payday lenders to whom they turned to pay for medical care, describing their financial problems as "aggressive collection." Because these families failed to use the specific term "illness or injury" to explain their bankruptcy filings, Dranove and Millenson simply exclude them from the "medical bankruptcy" category, thus greatly undercounting a serious problem. Second, their recalculation requires that debtors cite medical bills from providers (doctors and hospitals) as a major contributor to bankruptcy, to qualify as a medical bankruptcy. In their version, a debtor bankrupted by medication costs (but not saddled with hospital bills) would not qualify as medically bankrupt. As we stated in our paper, medication costs were the predominant out-of-pocket expense for most patients with psychiatric disorders and for virtually all debtors with Medicare coverage. Additionally, in many cases, inability to afford home care led to financial ruin because a breadwinner had to take time off work to care for a sick family member. Dranove and Millenson’s definition misclassifies costs for medications and home care as "non-medical." In sum, we stand by our original estimate that 46.2–54.5 percent of all bankruptcies were caused, at least in part, by illness or medical bills. Dranove and Millenson can reduce this estimate only by ignoring the substantial evidence of medical problems offered by the debtors themselves.
Dranove and Millenson are eager to demonstrate that few medical debtors are middle class, citing the debtors’ modest median income of $25,000 in the year prior to filing. But, as demonstrated in a considerable body of scholarly literature (which they ignore), income in the year prior to filing reflects the setbacks that led to bankruptcy (for example, illness that caused job loss), not socioeconomic status before the onset of financial disaster.2 For this reason, we gave detailed data on debtors’ education (55.8 percent of medical debtors had attended college, compared with 52.6 percent of all U.S. adults); home ownership rates (55.8 percent of medical debtors); and occupational prestige scores (80 percent were above the twentieth percentile). These are clear indicators that prior to the medical catastrophe, medical debtors were squarely in the middle class. If a more marginalized and less politically powerful group were similarly victimized by medical bankruptcies, would America’s Health Insurance Plans have bothered spending premium dollars to discount our work?
Dranove and Millenson also misrepresent the previous literature on medical bankruptcy. The detailed literature review included in the manuscript we originally submitted was deleted because of space limitations. (We were assigned the same word limit for the original study—including a primer on bankruptcy, detailed methods, and results—that Dranove and Millenson took for their reinterpretation.) Exhibit 1
As is evident from the exhibit, older studies found lower rates of bankruptcy—findings that Dranove and Millenson portray as a refutation of our study. But this is exactly our point: Medical bankruptcies have increased sharply. A two-page letter from the Office of the United States Trustee (which Dranove and Millenson incorrectly describe as a study by the Department of Justice) relies solely on what can be gleaned from court records: medical bills identified as medical bills. It incorporates no information from the families themselves about medical debts that wound up as credit card balances, second mortgages, pay-day loans, collection lawsuits, and so on, thus eliminating some of the most important indicators of medical bankruptcies. Most of the other recent studies they cite also relied solely on court records. The only previous work that collected comparable, although far less comprehensive, data also found high rates of medical bankruptcy—a study that Dranove and Millenson ignore.3 They gloss over the fact that 47 percent of filers in the Cincinnati study that they cite had substantial medical debts. Contrary to their assertion, the report includes insufficient detail to allow calculation of medical debt as a percentage of total debt for that 47 percent of filers. Moreover, they ignore the study’s other findings that closely mirror ours, as well as its conclusion that medical debts are often buried in credit card balances and second mortgages. Dranove and Millenson inappropriately compare our figures for debtors’ health spending, which excluded premiums, with a Census Bureau estimate, which apparently included them. Also, they tout a regression analysis by Scott Fay and colleagues, which tells very little about the reasons for filing bankruptcy. The piece analyzed a population-based survey unrelated to bankruptcy in which, as Fay and colleagues forthrightly conceded, about half of the families that filed for bankruptcy in the sample could not be detected because they refused to admit to their bankrupt status. With such gross underreporting, the conclusions about the reasons for filing are unreliable. (Note that the study found no link between job loss and bankruptcy—the other leading indicator of bankruptcy in every other study conducted.) Dranove and Millenson curiously contend that data from a recent Commonwealth Fund survey contradict our findings. According to that survey, 37.2 million adults annually are "contacted by a collection agency about owing money for medical bills," including 21 million who were insured all year. Some 15.2 million reported that a recent medical bill had resulted in "large credit card debt, loan, or debt against home"—further evidence that medical debt often appears as "credit card debt" or "mortgage" in court records. In fact, the Commonwealth data indicate that our findings were conservative. We estimated that 700,000 families ruined by financial fallout from medical problems file for bankruptcy annually. Many more are equally ravaged but are too poor, sick, or demoralized to negotiate the complex legal tangle of formal filing. Ultimately, Dranove and Millenson urge us to dismiss the empirical evidence provided directly from the debtors and to rely instead on multivariate analyses to unearth the "true cause." As clinicians and women who were fooled by the sophisticated multivariate analyses "proving" the benefits of estrogen, we are skeptical that such manipulations invariably reveal truth and trump other kinds of evidence. Indeed, most medical and policy breakthroughs—such as Harvey’s discovery of the circulation of blood, Koch’s that infectious agents cause disease, Semmelweis’s that hand washing prevents puerperal fever, Doll and Hill’s and Wynder and Graham’s that smoking causes cancer, or the Supreme Court’s that educational segregation causes disadvantage—eschewed multivariate analysis. When families melt down financially, they describe a host of experiences, from having debt collectors call every ten minutes all evening long to receiving foreclosure notices on their homes. Multivariate modeling is unlikely to yield a deeper understanding than the methods we used, based on debtors’ detailed descriptions of the role that illnesses and injuries played in their financial ruin.
Dranove and Millenson portray medical debtors as irresponsible deadbeats, their plight a result of bad choices analogous to those made by drunk drivers. But medical catastrophe is not a choice. And, unlike other expenses, most medical debt is not voluntary, any more than breathing is voluntary. Most medical debtors had taken the prescribed steps to protect themselves: Three-quarters had health insurance when they got sick. But for many, private coverage lapsed when they lost work because of illness. In other cases, the insurance they bought in good faith had so many holes that it left them unprotected. To families financially hard-pressed by illness, Dranove and Millenson preach acceptance of the hard message of the prophet Job. Instead of blaming ruined families, or the wrath of Jehovah, they should take a hard look at the defective products sold by their funders in the insurance industry. To Dranove and Millenson, a safety net strong enough to cushion the impact of illness is inconceivable. Yet many other countries have implemented comprehensive health and disability coverage that does the job quite well. In the end, they scold us for promoting "expansive proposals to protect all of us [that] distract from the pressing needs" of the uninsured. Two of us (Himmelstein and Woolhandler) have spent more than a quarter-century caring for the uninsured as primary care doctors in urban public hospitals and as researchers documenting their plight. We have documented the extraordinary bureaucratic costs of private insurance and the fact that national health insurance could capture huge administrative savings, allowing both coverage of the uninsured and upgraded coverage for the middle class. The other two authors have devoted their careers to documenting and alleviating the financial distress of America’s families. In contrast, the insurance company industry group that paid for Dranove and Millenson’s research has spent millions toppling universal coverage proposals. Thus, these authors’ scolding seems misdirected. Our paper garnered public attention because it resonated with the abuse that Americans suffer at the hands of our health finance system—abuse shared by the uninsured and the middle class. Dranove and Millenson hope to hide this truth, but the data are undeniable.
David Himmelstein (Dhimmelstein{at}challiance.org) is an associate professor of medicine at Harvard Medical School and a primary care physician at Cambridge Hospital in Cambridge, Massachusetts. Elizabeth Warren is the Leo Gottlieb Professor of Law at Harvard Law School. Deborah Thorne is an assistant professor in the Department of Sociology and Anthropology at Ohio University in Athens. Steffie Woolhandler is an associate professor of medicine at Harvard, where she co-directs the General Medicine Faculty Development Fellowship Program. She also practices primary care internal medicine at Cambridge Hospital.
| |||||||||||||||||||||||||||||||||
Top
Misrepresenting The Data
Selective Blindness
