Furious Seasons

December 04, 2007

Researcher Attempts To Answer Critics Of Suicide, Anti-Depressant Prescription Study

Most of you are aware that in September Robert Gibbons, a researcher at the University of Chicago, authored a study asserting that a 2004 FDA warning issued about suicide risk connected with anti-depressant use led to a drop in anti-depressant prescriptions and an uptick in suicide statistics that year. The study was challenged by many commentators, myself included.

Now, Gibbons attempts to answer critics of the study in this month's American Journal of Psychiatry. The authors of three separate letters challenge him on different points. Gibbons answers them here.

I won't even attempt to summarize the discussion as it's almost too technical to be illuminating. One point Gibbons doesn't address, however: in 2005, the suicide rate in America edged downwards slightly despite a drop in anti-depressant use that year. Which sort of makes his assertions about 2004 a moot point, but a beautiful example of religion in the world of science.

Let me be clear about a connected issue: The evidence on suicides and suicidality and anti-depressant use is mixed. There are messy studies such as Gibbons' from which you cannot draw too many conclusions. There are population-based studies that show no increase in suicides alongside increased anti-depressant use (not that I think population studies offer proof on this question due to the many confounding environmental factors which such studies cannot correct for). There are tightly-controlled clinical studies which show an elevated risk of suicidality (anywhere from 4 percent to 14 percent) connected with anti-depressant use, there is the FDA black box warning which is itself based on real-world data, then there are many tens of thousands of anecdotal reports (if not more) of suicides and suicidality connected with anti-depressant use, not to mention actual suicides during clinical trials of anti-depressants (in some cases, the trials weren't even for depression).

Straight-up: anti-depressant use is risky and, as far as I can tell, the drugs are at best a weak anti-suicide technology. Use with caution.

As exhausting as I find this endless debate, I'll continue to address it as the need arises. Besides, someone wrote something remarkably stupid about suicide in the current issue of a major magazine, and you just know I am not going to let that pass.

Posted by Philip Dawdy at December 4, 2007 12:01 AM

del.icio.us Digg it reddit

Comments

Philip, The issue is, as you rightly point out, about study design: the Gibbons study design is known as ecological - it takes two variables (numbers of prescriptions written and suicide deaths) and tries to make some sort of statement about the connection between the two. The commentors are telling him all the ways in which the connections are not logical. Of course, he rebuts these arguments so that his research can go forward.

As far as the weight you give this evidence; the study design dictates that it is tenuous, maybe a little bit interesting, but complete speculation. The authors themselves acknowledge this in their article. In science, though, this is where you start, not end, a discussion. Unfortunately, what gets translated to the public (and prescribers which is really unfortunate) is that this actually MEANS something important. I've had more shrinks, policymakers, etc tell me that they are prescribing (or advocating for) MORE prescribing of questionable drugs to kids after this article than I care to comment further about.

Posted by: Liz at December 4, 2007 09:17 AM

Right, Liz, and of course this is exactly what the purpose of publishing these articles is -- to frighten docs back into prescribing the drugs. The science, the logic, behind the conclusions in these articles, even if the data isn't fudged (or, more charitably, incomplete, which I'm sure it is) is pure BS. It's a lot better science to observe how people in clinical trials are reacting to antidepressants (by becoming more suicidal) than it is to go out and look at epidemiological data on prescription and suicide rates and say that demonstrates a drug effect.

Posted by: Sara at December 4, 2007 10:53 AM