Follow-up on medical school rank and narcotic prescriptions

I've had several conversations with fellow docs about the NBER paper from last week showing a relationship between medical school rank and narcotic prescriptions. Naturally, the responses I got were pretty skeptical. A recurrent theme I ran across was that the authors didn't pay enough attention to practice setting. 

1. Higher-ranked medical schools tend to produce more students who eventually land in academia, often at higher-ranked medical schools or residency programs.

2. Because of the culture of those high-performing places, characterized by greater access to subspecialists, a slower pace of practice, the presence of trainees to keep you honest and whatnot, you practice more conservatively and without fear of patients firing you for not giving them narcotic prescriptions.

 3. Perhaps people graduating from lesser-ranked schools end up in private practice, where there is more pressure to write a narcotic prescription to all those patients with back pain, just to keep them from 1) firing you, and 2) telling all their friends and family what a crappy, uncaring doctor you are.

4. So the speculated end result is that people graduating from higher-ranked medical schools end up writing fewer narcotic prescriptions than their peers from lower-ranked schools.

I pointed out to several people that the study took into account specialty and location; the relationship held for people in the same specialty and same county. The investigators pontificated on practice setting (noting that DO graduates often take care of a more rural, white population, for instance). But they didn't control for it. So I held my hands up ¯\_(ツ)_/¯ and moved on with my life.

But then I listened to a segment about it on the Weeds podcast today (start at 1:01:25):

UPDATE: I'm having some trouble embedding the audio, so if you can't get it to play, go here: http://bit.ly/2uLk2lY

Sarah, Ezra, and Matthew made note that the relationship held within the same clinic, so I re-visited the paper. I found this: 

 Check the footnote. *slaps forehead*

Check the footnote. *slaps forehead*

So the investigators did account for practice setting, but since it was hidden in a footnote, I missed it. 

The take-home from this paper is still kind of hard to identify. I'm not in favor of necessarily checking the LinkedIn page of every doctor you see to check what med school she went to, but there is clearly some kind of relationship between the culture of medical schools and the prescribing behavior of graduates.