Publication bias skews efficacy of therapy

It seems this is a big problem in the research around treatments for depression. In 2008, researchers looked at studies on antidepressants for a paper in the New England Journal of Medicine. They found that the medicine’s benefits had been exaggerated in the scientific literature because of publication bias.

For a new study in the journal PLOS One, researchers zeroed in on psychotherapies for depression. These including talking therapies, like counseling, group or family therapy, and cognitive behavioral therapy. To find out whether their effects have also been exaggerated by publication bias, the researchers went back to the source, identifying grants that the National Institutes of Health gave out for randomized trials looking at psychological treatments for depression.

Of the 55 grants that helped launch studies, about a quarter were never published. (That’s a little better than the estimated half of medical studies that are never published.) When the researchers got their hands on the unpublished data and pooled those results together with the research that had already been published, they found that psychotherapy looked a lot less promising. (Compared with the studies in journals, the efficacy of psychotherapy dropped by about 25 percent.)