The idea that half of all bankruptcies are caused by medical debt has become part of the common folklore.  But where did the idea come from?  What is the evidence for it?  The claim, first made in a 2005 Health Affairs article, is at variance with four decades of economic research, including a finding that even large medical bills have no impact on family living standards.

The paper by David Himmelstein, Elizabeth Warren, Deborah Thorne, and Steffie Woolhandler was published as a Health Affairs web exclusive on February 5, 2005. The authors are strong proponents of government run health care.

The data comes from 1,250 personal bankruptcy cases, assumed to be representative of the almost 1.5 million households that filed for bankruptcy in 2001.  The data on each bankruptcy were abstracted from court records and supplemented with 931 telephone interviews. The paper's conclusions about illnesses in households were based on medical interviews conducted with 391 people. The paper does not specify how those people were selected. It does say that Himmelstein and Woolhandler (H & W), both MDs, coded the diagnoses given by debtors into the categories used for the analysis. 

The classifications used to determine a medical bankruptcy were odd. Only 28.3 percent of the sample cited self-reported illness or injury as a cause of bankruptcy.  However, H & W managed to almost double that figure (to 54.5 percent) by counting the following as "illnesses":

• 1. A birth or addition of a new family member
• 2. A death in a family
• 3. A drug or alcohol addiction
• 4. Uncontrolled gambling
• 5. Loss of at least 2 weeks of work-related income due to illness or injury by anyone in the household
• 6. Out-of-pocket medical bills of $1,000 in the two years before filing by anyone in the household
• 7. Mortgaging a home to pay medical bills.

In a 2005 article in the Northwestern University Law Review, Prof. Todd J. Zywicki called the $1,000 threshold for contributing medical debt "indefensible." That's an understatement.  By H & W criteria, a bankruptcy with $50,000 in student loans and $1,001 in unpaid medical bills would be classified as a "medical bankruptcy."  Moreover, the average U.S. household had out-of-pocket expenses of $2,182 in 2001!  

In a 2006 review (gated) of the H & W study results in Health Affairs, David Dranove and Michael L. Millenson:

• Recalculate the medical bankruptcy rate using the data given in the H & W paper. They conclude that just 17 percent of the H & W sample "had medical expenditure bankruptcies," although it cannot be stated "with any degree of certainty whether medical spending was the most important cause of bankruptcy."
• Explain that "four decades of studies have addressed the bankruptcy-medical spending connection" and that the results from those studies are much closer to their 17 percent estimate than to the 54.5 percent estimates of H & W.
• Cite a 2002 Fay, Hurst, and White American Economic Review study, which found no statistical link between bankruptcies and health problems.
• Cite a 1999 Domowitz and Sartain Journal of Finance study, which found that high medical debt raised the probability of bankruptcy for the tiny proportion of the population that had high medical debt, but that at the margin, credit cards were the largest single contribution to bankruptcy.

Moreover, Helen Levy in an Economic Research Initiative on the Uninsured working paper estimated the effect of being diagnosed with a serious new health condition, (cancer, diabetes, heart attack, chronic lung disease, or stroke) and found that household consumption "remains smooth" in the face of serious health shocks for both insured and uninsured households.