The Report Card on Antidepressants

March 4, 2008— -- A study released last week reported that depressed patients put on antidepressants get better at only a slightly higher rate than those treated with a placebo, which is essentially a sugar pill. This is, of course, a potentially concerning finding for people like me, who prescribe antidepressants, and for those of you readers who take them.

The results, reported in the open-access journal PLoS Medicine, come from a combined analysis of 35 studies with more than 5,000 patients, evaluating four of the newer antidepressants -- fluoxetine (Prozac), venlafaxine (Effexor), nefazodone (Serzone) and paroxetine (Paxil).

The new twist is that this report includes data from many studies that were conducted by the drug companies and filed with the Food and Drug Administration but never published in academic journals.

What the PLoS study found is that, on average, patients taking medication had a substantial response as measured by a depression rating scale. Whereas they started with a score of roughly 26 (on a 54-point scale, where higher means more depressed), they dropped after six weeks of treatment to about 16.

But the patients taking placebos also, on average, had a strong response, dropping to about 18.

However, the placebo picture was a bit more complicated. Unlike the antidepressant response, which did not vary based on how initially depressed patients were, the placebo response was strongest in the least-depressed folks, dropping off in those whose illness was more severe.

So here's take-home message No. 1: Antidepressants showed a significant advantage over placebos for the severely depressed patients but not for the mildly and moderately depressed patients.

Maybe that's not really so surprising.

I had a mild depressive episode myself once while I was working in the Peace Corps in West Africa. It occurred in the context of such challenges as acclimating to a new language (Wolof) and culture (Senegalese), getting a well-digging project started and romantic turmoil.