MA Health Reform and Medical Debt – Getting the Facts Straight

by Rebecca Haffajee 

Earlier this week, the Boston Globe reported that medical debt is still a problem in Massachusetts, with scant change since the implementation of health reform legislation in 2006. Specifically, the article reports that of approximately 3,000 adults surveyed in 2010, 17.5% had trouble paying medical bills in the past year and 20% were carrying medical debt and paying it over time, statistically insignificant changes since 2006. The source of this finding is the Massachusetts Health Reform Survey (MHRS) funded by Blue Cross Blue Shield of MA Foundation, whose latest results published in January 2012 track annual trends from 2006 – 2010. The Globe story seems to suggest that in the absence of reductions in medical debt, health reform is failing to achieve one of its goals. The survey findings, however, don’t present a story of causal inference; they (at best) identify a loose association.

Just to recap some basics of MA health reform: the law required most residents to obtain insurance. It established Commonwealth Care through the Health Connector – an exchange of sorts – so that low income residents not eligible for Medicaid could qualify for a subsidized plan.  The Connector also offers Commonwealth Choice non-subsidized plans for individuals and employers.  Since passage of the law, insurance coverage among MA residents has increased from 94% to 98%.

The MHRS study design consists of 1 “pre” measurement, or the survey fielded in 2006 just before reform implementation, and 4 “post” measurements (2007-2010).  This design fails to provide a reliable counterfactual that reveals what would have happened in the absence of the health reform “treatment”.  A slightly better design would have administered survey questions for many years before health reform implementation. But even this design would be considered somewhat weak for causal inference given the presence of other factors that could have happened concurrently with the policy change that could explain outcomes. For instance, the recession could dramatically impact how much medical debt is incurred or not paid off, even with health insurance — especially with the proliferation of high deductible health plans in recent years.

In fact, medical debt could have increased in statistically significant ways absent health reform.  We won’t know if that is the case unless we have a comparable control population that did not undergo health reform but is identical to the MA population in all other ways.  Such a study is feasible to conduct, but must be carefully designed and analyzed to avoid endogeneity concerns that there are features about MA that make it unique from other populations and that are related to its passage of health reform.  A major endogeneity issue here is the uniquely high level of insurance in MA pre-reform.

In short, the MHRS doesn’t allow us to make a claim that health reform is failing to reduce medical debt.  And the aforementioned points about research design for causal inference don’t even touch upon other survey design issues that the study faces – e.g., small sample size (about 3,000 respondents) and potential recall and other forms of self-response bias.

A better study, mentioned only briefly in the Globe article, for inference about the effect of insurance coverage on medical debt is the Oregon Health Insurance Experiment (HIE).  This is a randomized control experiment that took advantage of a natural lottery designed to admit people into the Oregon Medicaid program.  The study follows populations of people who won the lottery and those who didn’t, confirming that the groups were similar at baseline to suggest successful randomization.  The groups are compared across a number of dimensions.  The researchers also validated survey responses using credit data, a more objective measure of financial means. This study is optimal for causal inference and suggests that insurance coverage among low income populations significantly reduces financial strain and medical debt.

Finally (and this has nothing to do with research design!), we should query why this Globe story was reported on Sept. 10, given that the underlying survey results were published in January.  The likely answer: politics.  As many are aware in the two months leading up to elections, two pivotal races promote health reform as a solution to individual and family financial woes.  Elizabeth Warren, who is vying for the MA Senate seat against incumbent Scott Brown, supports health reform to reduce health care costs and bankruptcy in the aftermath of serious medical illnesses.  And, of course, President Obama supports the Affordable Care Act (ACA) as a means for more Americans to afford care – by means of exchanges and expanded Medicaid, among other provisions – instead of foregoing it due to inhibitive costs.  With potentially 32 million more people covered under the ACA’s individual mandate, a central purpose of health insurance to protect individuals against catastrophic risks could be fulfilled.  True, the Globe study admits that other factors could have played a role in the lack of reduced medical debt among survey respondents.  But the article undercuts the idea that increased access and affordability can reduce medical debt and promote financial well-being by failing to report robustly on the Oregon HIE and by neglecting to identify the MHRS as a study not appropriate for causal inference.