The Triumph of Politics Over Public Health: States Opting Out of Medicaid Expansion

Health reform, the Affordable Care Act (ACA), was passed by Congress in 2010 to reduce the number of uninsured people in the United States and make health care more affordable. A provision of the law is expansion of Medicaid eligibility to include currently uninsured low-income individuals. The constitutionality of the Medicaid expansion provisions of the ACA was upheld by the Supreme Court in 2012; however, state participation is optional.

OPTING OUT OF MEDICAID EXPANSION DESPITE ITS BENEFITS

Medicaid expansion is financially beneficial for states. Assuming nationwide expansion, federal funding would cover approximately 93% of the cost of insuring new populations, with states responsible for only 7%.¹ According to investigators at the RAND Corporation, by 2016 the first 14 states committed to opting out of Medicaid expansion would forego $8.4 billion in federal funding and assume increased liability for uncompensated care to the uninsured. There would be 3.6 million fewer insured people in these states.²

The Kaiser Commission on Medicaid and the Uninsured identified 21 states that, as of July 2013, were not proceeding with Medicaid expansion and 6 states in which a decision had not been made. Forty-six percent of uninsured adults in the United States who could become eligible for coverage live in a state that is not planning to expand Medicaid. In the south, the region with the highest rates of uninsured, more than 80% of potentially eligible adults will not gain coverage because of decisions to opt out.³ Because of the importance of having an insured population for public health, it is necessary to explore the reasons for gubernatorial decisions not to expand Medicaid.

My analysis of Kaiser Family Foundation data⁴ revealed significant partisan political trends associated with decisions not to expand Medicaid. Among the 21 states opting out, 17 (80.9%) governors are republican (P < .01). Health reform figured prominently in the 2012 presidential campaign, with Former Governor Mitt Romney having made clear his intention to defund the ACA if elected. Among these 21 states, 19 (90.4%) voted republican in the 2012 presidential election (P < .01).

The decision not to expand Medicaid, however, is not generally presented in partisan terms. Instead, Medicaid is denigrated with allegations of billions of dollars in waste, fraud, and abuse, as Michael F. Cannon, Director of Health Policy Studies at the Cato Institute, underlined in his congressional testimony in 2011.⁵ Furthermore, some of its critics maintain that Medicaid's quality of care and financing are so bad that, as stated by Brian Blase at the Heritage Foundation, Medicaid expansion under the ACA will “likely hurt overall population health.”⁶

THE IMPACT OF MEDICAID COVERAGE AND THE OHIE

The notion that health care reimbursed by Medicaid is of poor or questionable quality has recently been attributed to a public health research study by Baicker et al., the Oregon Health Insurance Experiment (OHIE).⁷ The OHIE was specifically referenced, for example, in the Louisiana Governor’s explanation of his decision not to expand Medicaid.⁸ The OHIE was a randomized controlled trial that allowed comparison of use and outcomes among adults newly enrolled in Medicaid and a demographically matched control group. This was possible because, in Oregon, uninsured low-income adults were randomly selected through a statewide lottery for the opportunity to enroll in Medicaid.

The OHIE generated intense interest among health policy analysts at conservative think tanks. These were the finding that fueled this strong reaction: After two years, compared with the control group, the newly Medicaid-enrolled adults showed “no significant improvement in measured physical health outcomes” (referring to several indicators such as blood pressure and cholesterol, screened for during study participant interviews). Often the phrase “no significant improvement” was misrepresented as “no improvement.” Significant positive findings in the OHIE were minimized or ignored. These positive findings demonstrated success for Medicaid enrollment in meeting public health goals, including increased primary health care use (P < .01), increased receipt of preventive health services including Papanicolaou tests (P < .05) and mammograms (P < .01) for women, increased diagnosis and treatment of diabetes (P < .01), and increased diagnosis of depression (P < .05). Medicaid enrollment was significantly associated with reduced out-of-pocket medical expenditures and debt (P < .01).

OHIE MISREPRESENTATION TO UNDERMINE MEDICAID EXPANSION

Despite positive findings, the OHIE was used by conservative policy analysts to support their established positions against Medicaid. Cannon entitled his response “Oregon Study Throws a Stop Sign in Front of ObamaCare’s Medicaid Expansion,”⁹ claiming that the study showed “no evidence that Medicaid improves the physical health of enrollees.” The results, he wrote, are “a rebuke to those who are pushing states to expand Medicaid” in part because the newly insured would not be “a population that is going to start dying in droves if states decline to participate.” These interpretations seem to have been influenced by Cannon’s previously published position that “Medicaid discourages work and charitable efforts among the taxpayers who fund it, while discouraging self-sufficiency and encouraging dependence among beneficiaries.” Medicaid’s high cost reflects “overuse of medical care” while “giving patients poorer-quality care than they could obtain with private coverage.”¹⁰ A subsequent George Washington University analysis of data from 31 825 primary care visits between 2006 and 2010 showed a slight advantage in care received by Medicaid-insured patients compared with patients with private coverage.¹¹

Another prominent critic of the study was Avik Roy, Senior Fellow at the Manhattan Institute and former health policy advisor to Romney. Roy had previously described Medicaid as “the developed world’s worst health-care system”¹² despite the fact that Medicaid is an insurance type not a health care system. His critique of the OHIE was more nuanced than that of most of his fellow critics.¹³ The finding that Medicaid virtually eliminated catastrophic medical debt, however, was trivialized by his comment that “reduced financial strain didn’t result in better health outcomes.” Roy suggested that the financial benefits of Medicaid could be achieved at a lower cost by giving cash directly to poor people or by subsidizing the purchase of catastrophic health insurance on the open market. He quoted columnist Ross Douthat’s opinion that catastrophic insurance is sufficient because health insurance should be like flood insurance, not compensating for “ordinary repairs” but only compensating after “actual disasters.” Roy concluded that the OHIE results are “consistent with a mountain of clinical evidence showing that Medicaid makes no meaningful difference, at best, compared to being uninsured.” This overlooked evidence to the contrary from the Institute of Medicine among others,³ including a 2012 study by Sommers, Baicker, and Epstein that showed that previous state Medicaid expansions were significantly associated, over the course of a decade, with reduced mortality (P < .01).¹⁴

IMPLICATIONS FOR PUBLIC HEALTH RESEARCHERS

The degree to which the OHIE was used to support positions contradicted by the study’s results and other evidence is instructive for public health researchers working on issues generating ongoing policy debate. First, one must question whether the randomized controlled trial model is the gold standard for public health research. In the OHIE, the Sommers, Baicker, and Epstein study, which had a larger sample size and longer interval for observation of effects than did the OHIE, was dismissed because it was a nonrandomized quasi-experimental model.⁷ The randomized controlled trial model, however, may not effectively eliminate confounding variables in studies conducted in real world settings.¹⁵ It is possible, for example, to control whether patients with a rare and fatal cancer take or do not take an experimental chemotherapy cocktail to test the clinical efficacy of that pharmacological treatment. It is not possible to control whether, over a two-year period, people not selected by lottery to enroll in Medicaid otherwise get health insurance or use health services while uninsured.

Second, outcome variables must be sensitive to the impact of the variable being tested. Much was made by critics of the OHIE about the lack of statistically significant improvement for blood pressure and cholesterol. These may not have been appropriate indicators for this study. Medicaid enrollment does not directly affect blood pressure or cholesterol. Both are multidetermined and reflect lifestyle choices such as diet and physical activity. Expectations regarding the magnitude of change for blood pressure and cholesterol values should have been tempered by the high reported prevalence of obesity in the OHIE control group.⁷ This is likely to have reflected baseline obesity rates for the new Medicaid enrollees.

Third, health services researchers whose findings may become part of politically charged policy debates should be cognizant of the ways that scientific writing may, intentionally or not, be misunderstood or misrepresented by others. Researchers should take this into account when presenting their findings. For example, in the OHIE, the abstract's conclusion begins with a statement that “[t]his randomized, controlled study showed that Medicaid coverage generated no significant improvements in measured physical health outcomes in the first 2 years.”^7(p1713) While accurate, in retrospect this statement was widely used to support negative misrepresentations of the study. It would have been accurate instead to have emphasized the significant positive findings for increased use, receipt of preventive services, diagnosis of chronic conditions, and reduced out-of-pocket medical expenditures, all of which touch on sensitive health insurance issues. The distinctions Baicker et al. made in their carefully worded presentation of OHIE findings may be assumed to be understood by typical consumers of public health research articles but not necessarily by lay readers, let alone political operatives in search of support for their pre-existing positions.