It’s not possible for me to get all my thoughts in order in a few hours, while taking care of my kids. So, last night’s post by Aaron and me left something out. It probably left a few things out, but I’ll just hit one here.
Some readers may recall my blog-to-blog battle with Avik Roy and others about whether Medicaid harms health relative to being uninsured. He (and others) said it did and pointed to some studies that could only show an association and, hence, suffered from reverse causality. I (and others) said it didn’t and pointed to studies with designs that could more plausibly reveal the causal effect of Medicaid on health. (You’ll find plenty on this under the Medicaid-IV tag. See also our NEJM publication.)
What should settle the matter once and for all is an RCT. Now we have one. It may not have shown that Medicaid is as beneficial as I might have thought, though given all the limitations of the study it is also possible that’s too tall an order. However, it also did not show that Medicaid harms health. Far from it. The benefits of Medicaid on mental health are clinically and statistically significant. The benefits of Medicaid on some process measures were not statistically significant, which is not the same thing as a lack of benefit. More to the point, it is also not the same thing as showing harm.
You know me well enough by now that I’m not comfortable relying on one study for all things. It is also not necessary for me that Medicaid be the best thing since sliced bread. I can entertain the idea that Medicaid, indeed many other forms of insurance, is poorly designed and does not provide as much bang for the buck as we would hope. I do not think the latest Oregon study necessarily settles every question about Medicaid.
But one thing I do think it settles, in conjunction with the body of other good work suitable for causal inference, is that Medicaid offers benefits and there is no evidence of harm, relative to being uninsured. Until I see a study with an equally strong research design demonstrate otherwise, I will not believe for a moment that it is so. There has not been a single study designed for causal inference that has shown Medicaid harms patients. Now we have an RCT that does not show that.
Can we slay this zombie idea now?