A Scientific Circle Jerk

We (the We being royal) have written on a few occasions about the dubious evidence for the efficacy of antidepressants. (Here's the key post.) Now, in one week, our two leading medical journals have come out with major research reports which are intended to persuade doctors to prescribe more antidepressants. Both journals made a big PR push on these "studies" (we'll get to the scare quotes momentarily), so you may have read about them in your favorite bird-cage liner.

Since, unlike We, you are mere commoners, you can only read the abstracts, but that should be enough. In the New England Journal of Medicine we have Bupropion-SR, Sertraline, or Venlafaxine-XR after Failure of SSRIs for Depression by Rush, et al; and Medication Augmentation after the Failure of SSRIs for Depression, by Trevedi, et al. In JAMA, we have Remissions in Maternal Depression and Child Psychopathology, by Weissman, et al.

The JAMA article finds that children of depressed mothers do better when their mothers' depression improves. (Another open door crashed through.) Their conclusion? "These findings support the importance of vigorous treatment for depressed mothers in primary care or psychiatric clinics and suggest the utility of evaluating the children, especially children whose mothers continue to be depressed." There is only one problem with this conclusion: 100% of the mothers in the study were being vigorously treated. Some of the mothers improved during the course of the study, some did not. The study provides no evidence whatsoever that the treatment made any difference.

Then there are the two NEJM articles. Both of them report on similar strategies: people who were taking antidepressants, who did not improve, were switched to other antidepressants or given additional ones. In both cases, about 30% of them eventually showed some improvement. Again, however, 100% of the patients in both studies were being treated. There was no control group in either study. Depression sometimes remits on its own, given time, and the placebo response in depression is very strong -- as a matter of fact, it's at least 30%. (What a coinkydink!)

In other words, it's no longer considered necessary that research using antidepressants actually provide any evidence that they do any good whatsoever. That is now considered axiomatic. Having assumed that they must be working, you can then write in your conclusions that they ought to be prescribed. Well, think back to your high school geometry class. If you use your axioms to prove your conclusions you are guilty of what? That's right class, circular reasoning.

Remember the study (also in JAMA) that showed that St. John's Wort was no better than placebo? That same study also showed that Zoloft was worse than placebo at alleviating depression, but the authors refused, for no particular reason, to draw that conclusion. How could they? It would have been like claiming that parallel lines meet. It would have violated the axioms of psychiatry.