The New England Journal of Medicine published an analysis of the effects of Medicaid coverage, and it’s no surprise that both sides of the Medicaid expansion debate claimed to have found support for their positions. Opponents of the health reform legislation, and the Medicaid expansion in particular, argue, “See? We told you Obamacare won’t work.” Proponents look at the same data and say, “Medicaid works.” And some observers see that each side is only focusing on one aspect of the data.
What exactly did the results show? Two years after winning the lottery, the Medicaid beneficiaries had gotten more preventive screenings, were more likely to have been diagnosed with diabetes and to be on medication for it, they reported less depression and were free of the strain of medical debt. But there were no significant differences in blood pressure, cholesterol levels, or glycated hemoglobin between the two groups. The same percentage of subjects in each group had glycated hemoglobin levels exceeding 6.5 percent, which would signify diabetes.
Apparently, the investigators had hoped for a demonstrable improvement in physical health, and the results didn’t support that interpretation. The authors and others have offered several possible explanations for this unexpected result.
- Having Medicaid doesn’t eliminate all barriers to access to care
- Medicaid beneficiaries get poorer quality health care than either the privately insured or the uninsured because of the low reimbursement rates
- It takes time for the effects of preventive care to become apparent
- “Lifestyle” changes are necessary to lower blood cholesterol or blood pressure, and it is hard to quit smoking, change one’s eating habits, and exercise.
Isn’t elimination of crushing medical debt are the purpose of health insurance, including Medicaid? It stands to reason that removing the debt would decrease the likelihood of depression. And it’s well established that one’s state of mind is an important factor affecting one’s physical health.
In Forbes, Avik Roy compared the Oregon Study to the kind of clinical trial that the Food and Drug Administration would require to test a new drug. The study was flawed because it wasn’t double-blinded, meaning that the subjects knew whether they were on Medicaid. And there wasn’t a specified endpoint, a specific result that would show whether the medication was effective at treating the disease. Is that a fair comparison? The investigators took advantage of an opportunity to see what happened when some people on a waiting list had the chance to apply for Medicaid and others did not. But what the two groups had in common was poverty, not any particular medical condition.
In the aggregate, the lottery winners and the control group were similar demographically. But that doesn’t mean they all started out in the same state of health. To look for changes in the clinical conditions, the investigators focused on the subjects who reported having a diagnoses of hypertension, diabetes or high cholesterol before March 2008, when the first lottery was held. The subjects who were chosen in the lottery, applied, and were found eligible for Medicaid may have been more motivated to follow through because they had a greater need for health care. To know what difference Medicaid made in the lives of beneficiaries, it would be useful to know how many of them were newly diagnosed with hypertension, diabetes, or high cholesterol during the study period. The investigators also asked about diagnoses of asthma, but the analysis of that data has not been published.
The results at two years are based on interviews held between September 2009 and December 2010. The average length of time on Medicaid was 17 months, not two years. Not all the Medicaid beneficiaries stayed in the program for the entire two years, and we cannot assume that everyone in the control group was uninsured for the entire time, either. The study documentation states that some of the people in the control group got Medicaid benefits through a later lottery. The article states that winning the lottery increased a subject’s chances of having Medicaid coverage by about 25 percent. Apparently, the population who actually got Medicaid after winning the lottery was not as similar to the control group as the larger group of lottery winners, so the investigators used winning the lottery as the “instrument” for Medicaid coverage. In other words, many people in the “Medicaid group” actually did not receive Medicaid coverage; if that is true, it’s no surprise that Medicaid coverage had little effect. The investigators’ plans for further analysis will focus more closely on the group who enrolled in Medicaid.