Oregon and Medicaid

The new Oregon Medicaid study is coming out at a very bad time for me to comment.  But I want to direct you to the Incidental Economist which is doing a banner job of clarifying why it was hard to get good evidence directly on health outcomes from the study.  Further comments are here. 

One thing to note on health outcomes (poached from the comments at TIE) is:

In the case of total cholesterol the rate was reduced by 17% (from 14.1% to 11.7%). If that kind of drop is not detectable by the study then I think it is a problem.

Or this one:

 HgbA1C drops from 5.1% to 4.2%.

I think that this is too harsh, but it does point out that many of the changes were clinically significant but that the study (for a lot of reasons due to enrollment and short follow-up) is not really able to give precise estimates.   Remember, only 25% of the people offered Medicaid took it, so the adherence rate is a lot lower than what we normally think of in an RCT and so the intention to treat estimate is a poor measure of the associations among those who enrolled. 

Kevin Drum discusses this more here. Pay special attention to the PDF at the bottom of the post.