Tuesday, October 03, 2006

If You're Looking for A Small Benefit -- Try Effexor for PTSD

The Archives of General Psychiatry had another interesting piece this month. In a rather large double-blind trial, venlafaxine (Effexor) was compared to placebo over a six-month period in the treatment of PTSD. Why so large (161 in the Effexor group and 168 in the placebo group)? Statistically, when you are likely to only find a small effect of treatment, make sure to have a large sample, so that you can find statistical significance.

How small the effect? The authors are to be commended for providing effect size data in Table 4, which indicate, across 17 measures, the highest effect size was .35 (in terms of Cohen’s d). I calculated the average of the effects across the 17 measures and came up with an average d of .256. So across 17 measures, the mean effect size was .256. This puts the effect just above small (generally recognized as d = .20), and well short of moderate (generally considered d = .50). In other words, not much to get excited about.

The authors receive credit for briefly (and I mean one sentence) mentioning that the effects were in the “low to modest” range, “in line with earlier reports for sertraline and paroxetine (p. 1163).” Using remission as an outcome, the authors found a number needed to treat of 8, meaning that for every eight people treated with venlafaxine, one additional patient would show remission relative to placebo treatment. This is described as a “moderate effect,” which seems like a stretch, but probably short of utter B.S. (p. 1164).

To sum up, venlafaxine (Effexor) showed a small advantage over placebo that was unlikely to be due to chance. The difference is so small that it is unclear why psychotherapy would not be indicated as opposed to pharmacotherapy in cases of PTSD since psychotherapy has a much lower incidence of side effects and likely longer lasting benefits. The longer lasting benefits guess is an extrapolation from research on treating anxiety disorders with medication, where once medication is discontinued, symptoms tend to recur, whereas the effects of psychotherapy tend to linger once treatment is discontinued (though not completely). The authors make not a single reference to nonpharmacological treatments.

Ghostwriter Watch: Dr. Michael P. Rennert is acknowledged for his “editorial assistance” with the paper. A simple Google query shows and ad of his placed in the American Medical Writers Association Journal in 2002 (Vol 17, No. 3). One can only guess who wrote what in this article, but it’s probably safe to believe that Dr. Rennert made a substantial contribution to the paper’s text.

Duplicate Presentation Watch: Right before the references sits a section titled “Previous Presentations.” This acknowledges that this study was formally presented at 13 conferences. THIRTEEN! In psychology, the general guide is to present findings at one conference, not a baker’s dozen. Maybe if it's a big study, split the findings between two or maybe three conferences at most. Apparently in psychiatry, anything goes in terms of duplicate presentations! Talk about marketing like crazy from a pretty meager finding. Perhaps this can make its way into multiple publications as well? Let's hope not.

No comments: