Anti-depressant medications appear to help only very severely depressed people and work no better than placebos in many patients.

Researchers led by Irving Kirsch of the University of Hull reviewed a series of studies, both published and unpublished, on four anti-depressants, examining the question of whether a person's response to these drugs hinged on how depressed they were before getting treatment.

They were Eli Lilly and Co's Prozac, also known as fluoxetine, Wyeth's Effexor, also called venlafaxine; GlaxoSmithKline's Paxil, also called Seroxat or paroxetine, and Bristol-Myers Squibb Co's drug Serzone, also called nefazodone, which it no longer markets in the United States.

They are all so-called selective serotonin reuptake inhibitors, or SSRIs.

The researchers found that compared with placebo, these new-generation antidepressant medications did not yield clinically significant improvements in depression in patients who initially had moderate or even very severe depression.

The study found that significant benefits occurred only in the most severely depressed patients.

Drug-placebo differences in antidepressant efficacy increase as a function of baseline severity, but are relatively small even for severely depressed patients. The relationship between initial severity and antidepressant efficacy is attributable to decreased responsiveness to placebo among very severely depressed patients, rather than to increased responsiveness to medication.
The researchers obtained data on all the clinical trials submitted to the US Food and Drug Administration for the licensing of the four drugs.

Although patients get better when they take antidepressants, they also get better when they take a placebo, and the difference in improvement is not very great. This means that depressed people can improve without chemical treatments.


This analysis has only examined a small subset of the total data available, while regulatory bodies around the world have conducted extensive reviews and evaluations of all of the data available.