The antidepressant-debunking study

[substitute]

There was a news release today about a study that appears to show the uselessness of popular antidepressants.

This was reported in the Guardian, among other places. The publication can be read here.

There are problems, as summarized:

1. PlOS is not an academic peer-reviewed journal. edit: They are in fact peer-reviewed, based on better information I have received by comments. Read the threads. They say they are peer-reviewed, but when you read their FAQ, you'll see this: "We involve the academic community in our peer review process as much as possible. After professional staff have determined that the paper falls within the scope of the journal, and is of a minimum acceptable quality, decisions on whether to send a paper out for in-depth review are made via a collaboration between experienced, professional editors who work full time at PLoS, and academic editors who are experts in their field."
   
   I'm not saying this is Wikipedia, but it's not the same thing as a traditional journal, either.
   
   
2. It's one study. Beware of an equivalency between "one metastudy showed that these three or four drugs didn't show a good outcome under these conditions" and "antidepressants don't work."
   
   
3. The study measured outcomes at six weeks. That isn't very long in a depression treatment, whether you're using Prozac or a trampoline.

That having been said, anything that keeps family doctors from throwing the best-advertised drug at every problem is going to be helpful at this juncture. And using any kind of medication (except possibly the trampoline) without counseling is, well, crazy.

Comments

[etb.livejournal.com]

Not knowing about the practices of a new publishing medium, I want to know for sure that the good things about the old model haven't been tossed out with the bad.

My impression is that the biggest change in medical publishing isn't the "real" open-access journals like PLoS's—called "gold OA"—but open-access repositories, called "green OA". These haven't done very well, because scientists don't bother to post to repositories unless someone makes them, even if the journal explicitly allows it (and many toll-access journals do). This, despite pretty good evidence that open access, not surprisingly, increases exposure and therefore increases "impact". Scientists are insanely conservative. The high-energy physicists may have started xxx.lanl.gov because it was the smart thing to do, but I'm sure the only reason they keep using it is habit.

However, the NIH is about to mandate that all journal articles resulting from NIH-funded research be posted to PubMed. In the long term, funder mandates mean that the old model will work even more badly than it already does (serials crisis? what serials crisis?), because there's no need for institutions to subscribe to journals if the articles are freely available. The journals will have to convert to "author pays" (which will really mean "funder pays") to survive. But that really has nothing to do with peer review.

[substitute.livejournal.com]

I really like the idea of Green OA.

[etb.livejournal.com]

You might find Stevan Harnad's rants interesting, though the one I linked to looks somewhat restrained.