For the second time in six years, a metanalysis led by psychologist Irving Kirsch asserts that antidepressants are no more effective than placebos (see November/December 2002 Networker). Like his earlier analysis, Kirsch analyzes both published clinical trials and the raw data that pharmaceutical companies are required to submit to the Food and Drug Administration. This time, Kirsch asks whether the same antidepressants he studied earlier—Prozac, Effexor, Serzone, and Paxil—work better than placebo on severe depression. The answer, he says, is still no.
Kirsch insists that using the raw data enables him to avoid the well-known bias toward positive results inherent in published studies of antidepressants. A study in the January 17 New England Journal of Medicine, for example, reports that 94 percent of the published trials were positive, whereas only about 50 percent the unpublished studies were. Raw data, unlike published studies, aren't methodologically tweaked to insure positive results.
Like Kirsch's earlier study, the new one, in the February online journal Public Library of Science Medicine, has generated considerable controversy. Much of the criticism focuses on his decisions about what constitutes clinical effectiveness, which his detractors say were arbitrary and ignored the considerable number of individuals who show marked improvement. Using the same data as Kirsch, psychiatrist Eric Turner and psychologist Robert Rosenthal conclude that each drug is indeed superior to placebo in the March 8 online edition of the British Medical Journal.
Nevertheless, in the wake of Kirsch's study, British Health Secretary Alan Johnson has announced plans to train 3,600 more therapists, hoping that with more therapists available, doctors won't reach as quickly for their prescription pads.