Doctors often tell people starting on an SSRI such as fluoxetine or paroxetine that they have a 55-70% chance of responding to their medication. However, a recent paper by Turner et al in the New England Journal of Medicine (358(3), 252-260) suggested that the effectiveness of SSRIs is greatly exaggerated as a result of a bias in the publication of research studies. Of 74 studies registered with the United States Food and Drug Administration (FDA), 37 of 38 studies with positive results were published in academic journals. In contrast, only 14 of 36 studies with negative results were published in academic journals, and 11 of these were published in such a way as to convey a positive outcome. Thus, whilst 94% of published studies conveyed a positive outcome, only 51% of all studies (published and unpublished) actually demonstrated one.

Another paper by Kirsch et al in Public Library of Science Medicine combined 35 studies submitted to the FDA before the licensing of four antidepressants, including the SSRIs fluoxetine and paroxetine. The authors of the study found that, whilst the antidepressants performed better than a placebo, the effect size was very small for all but very severe cases of depression. Furthermore, the authors attributed this increased effect size in very severe cases of depression not to an increase in the effect of the antidepressants, but to a decrease in their placebo effect.

If, as these studies suggest, the efficacy of SSRIs has been greatly exaggerated, their cost-benefit urgently needs to be re-evaluated. In any case, there can be little doubt that at least some of the benefit of an antidepressant is attributable to its placebo effect.