07/17/2013 04:49 pm ET Updated Sep 16, 2013

A Look at the Oregon Medicaid Experiment

Do you have health insurance? If so, why? (I could also ask "if not, why not?" but that's a subject for another time.)

This is not a rhetorical question. As we approach implementation of the cornerstone of the Affordable Care Act--mandatory health insurance coverage, including an expansion of the Medicaid program--we are likely to hear its value debated.

So, why have health insurance? For peace of mind? So you can see a doctor without worrying that it will break the bank? Maybe because you know that medical debt is a contributing factor in more than 62 percent of individual bankruptcies? Or is it because you need to lower your blood pressure and get your cholesterol under control?

While you're thinking about that, let me tell you about a recent randomized controlled trial that pondered a similar question. It followed two groups in Oregon: those who were given the opportunity to enroll in the state's Medicaid program through a limited expansion and those who were not. For researchers, this kind of data--a truly random, blind sample of several thousand people--is pretty rare, and the study's potential excited policy and public health wonks around the country.

Happily for the uninsured in Oregon, the state can now offer everyone eligible access to Medicaid, so the study left researchers with just two years of data to play with.

Those results were written up in the New England Journal of Medicine back in May, and since then pundits and academics alike have been having a field day with them. The findings, in report-speak: "This randomized, controlled study showed that Medicaid coverage generated no significant improvements in measured physical health outcomes in the first 2 years, but it did increase use of health care services, raise rates of diabetes detection and management, lower rates of depression, and reduce financial strain."

The first part of that sentence refers to measured blood-pressure, cholesterol, and glycated hemoglobin ("GH," a measure of blood sugar control) levels, which did improve among the Medicaid "lottery winners," but not by a statistically significant amount. Forget that the other improvements were statistically significant--you can guess how the anti-Obamacare camp responded. The blogosphere was full of those denouncing public health interventions, especially Medicaid.

NHeLP dug into the numbers and found that while the Oregon Medicaid study suffered from weaknesses that make drawing any conclusions from its two years of data a tricky proposition, its results favor health insurance coverage and suggest areas in which our existing system can be improved.

First, let's talk about two significant shortcomings in the report:

It did not compare people with health insurance and people without health insurance. Instead, the study followed those who were given the opportunity to enroll in Oregon's Medicaid program and those who were not. Only 42 percent of those given the opportunity to enroll actually did so, and in the meantime many of those who were "lottery losers" found other ways to access health insurance, including Medicaid. The result? In the end the study compared a group in which 47 percent had insurance to a group in which 36 percent had insurance, a difference in insurance coverage of only 11 percent.

To demonstrate whether Medicaid has an impact on managing diabetes, blood pressure or cholesterol, the study had to be bigger or the people sicker. Even though individuals in the "lottery winner" group were less likely to have diabetes (by 18 percent), high cholesterol (17 percent), or high blood pressure (8 percent) than the control "loser" group, these results were not considered "statistically significant" in large part because so few people in the study had these conditions to begin with. For example, only 5 percent of the lottery losers had diabetes; as a consequence, economists have calculated that the study needed to be 23 times larger to get an accurate picture of Medicaid's impact on GH.

By contrast, more than 30 percent of those in the control group suffered from depression, and all of the control group members were economically vulnerable, so results in these areas are more likely to be accurate. And here the findings are statistically significant: the study found much lower rates of depression in the "lottery winner" group and, perhaps relatedly, that reports of catastrophic out-of-pocket health care costs virtually disappeared in this population.

Perhaps even more telling? Lottery winners were also more likely to report that their health was the same or better than the previous year, a key predictor of long-term health.

Medicaid is not a perfect system, and to the extent that the Oregon study suggests areas for improvements we ought to pay attention. But to suggest that the program is worthless because certain health indicators did not improve enough over two years reflects a very limited view of the purpose of insurance.

Of course, it is easy to throw stones from a safe distance. I wonder how many Medicaid critics would actually give up their own insurance if their cholesterol did not improve enough over two years.