Douthat and generalizing the OHIE


H/T The Incidental Economist

If you have the opportunity this Sunday (a rainy, cool one in my neighborhood) to read Ross Douthat’s column on the most recent Oregon Heath Insurance Experiment (OHIE), the subject of which I covered in my last post, please do so because it illustrates a few very important points that many Very Serious People are missing.

First, though, here’s Douthat (emphasis mine):

Needless to say, experts have been arguing about what the RAND results mean ever since. But the basic finding — that more expensive health insurance doesn’t necessarily lead to better health — just received a major boost. The state of Oregon expanded its Medicaid program via lottery a few years ago, and researchers released the latest data on how health outcomes for the new Medicaid users differed from those for the uninsured. The answer: They didn’t differ much. Being on Medicaid helped people avoid huge medical bills, and it reduced depression rates. But the program’s insurance guarantee seemed to have little or no impact on common medical conditions like hypertension and diabetes.


This is true. But it’s also true that the health care law was sold, in part, with the promise (made by judicious wonks as well as overreaching politicians) that it would save tens of thousands of American lives each year. There was so much moral fervor on the issue, so much crusading liberal zeal, precisely because this was not supposed to be just a big redistribution program: it was supposed to be a matter of life and death.

And his preferred alternative:

There are a variety of ways this could be accomplished — a bigger child tax credit for struggling families, a payroll tax cut to boost workers, an expanded earned-income tax credit to raise wages at the bottom, health savings accounts that roll over money left unspent. In each case, the goal would be to help people rise by giving them more money and more options for what to do with it, rather than just expanding 1960s-vintage programs that pay medical bills and only medical bills.

The second point is especially irritating because Douthat should know better. If ten people out of every 10,000 — which should satisfy any ‘believability’ factor — who receive Medicaid coverage stay alive as a result then at the estimated 21.3 million who would enroll after 2014 (assuming, theoretically, all states accepted expansion) that would account for 21,300 saved lives. Ten is just arbitrary number, of course, and could be much higher but it highlights how easily it would be to rebut such a silly criticism.

As to the first and third points, as I wrote earlier, this was a study covering fewer than 13 thousand people in one city in America for two years. That this is a randomized controlled experiment means we can start discussing causality, but there are actually serious reasons to doubt the causal applicability of these “common medical conditions” studied. Which is to say, the study may have been underpowered to determine statistical significance for these conditions. In other areas there was substantial differences in mental health; positively for those with Medicaid coverage. Now a drastic reduction in depression is a very important health outcome, a fact of which seems (while meriting some mention) to be dropped to the wayside on the journey to declaring the program as it exists now, all things considered, a waste of the public’s time and resources.

For obvious reasons I find that line of thinking distasteful, and I don’t write that to assert that Douthat considers Medicaid an absolute waste of time, per se, but he obviously believes it could be done better in way that satisfies his ideological priors. Well, so do I. The difference, and this is important, is that my priors are being currently being satisfied compared to the conservative alternative — one founded not in his or a handful of right-leaning wonk’s preferred alternatives but the elected Republican vision of moving the poor’s healthcare concerns off the federal agenda and chipping in drastically less to states to cover fewer people. That is the reality we’re dealing with right now. Douthat recognizes this, nominally, but that hasn’t stopped him (or indeed, most others on the right) from taking this one study and making broad and sweeping generalizations about a program that covers 62 million Americans. Or from deciding that expanding access to another 21.3 million people ‘just isn’t worth it.’

Just to recap here:

  • It’s well-within reason to believe that expanding access to Medicaid coverage would, in fact, save “tens of thousands of American lives.”
  • The OHIE does not, however many caveats you include, prove that having Medicaid means your health outcomes will be indistinguishable from the low-income uninsured.
  • The choice isn’t between Medicaid as it exists now and the utopian ideals of a handful of people, but between current Medicaid and blocking the expansion while writing more columns about the ideal.

I’m not sure which one of those deligitimizes the reactionary bomb-throwing on the right from the OHIE most, and it probably doesn’t matter, but I’d put my choice on the middle bullet-point. Without such an (even then, misinterpreted) ‘proven’ leg to stand on there isn’t much for pundits and partisan analysts to prognosticate. In the end it doesn’t really matter. This study hasn’t, and won’t, change anyone’s mind on any fundamental questions despite Douthat’s insistence in the NYT that no, really, it should. Further, that while conservatives just need a few photogenic champions, it’s actually liberals that must now rethink their entire ideological [strawman] approach to healthcare .

No thank you, Ross. I’d rather keep calm and collect more data.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s