Antidepressant medications appear to help only severely depressed people, a new analysis has found. For most patients, the medications work no better than placebos, the study found.

Researchers led by Irving Kirsch of the University of Hull in Britain reviewed 47 studies, both published and unpublished, on four antidepressants from a class of drugs called selective serotonin reuptake inhibitors, or SSRIs:

• Prozac, (also known as fluoxetine)
• Paxil (also called Seroxat or paroxetine)
• Effexor (also called venlafaxine or Novo-Venlafaxine)
• Serzone (also called nefazodone; no longer available in Canada but is available in the U.K.)

The researchers wanted to know whether a patient's response to the antidepressant depended on how badly depressed they were to start out with.

They conducted a meta-analysis, putting together all data from trials submitted to the U.S. Food and Drug Administration on the drugs - relying on U.S. freedom of information laws to obtain the unpublished studies.

By including data from unpublished as well as published trials, the researchers were hoping to avoid bias that might come from non-publication of disappointing findings.

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

They found significant benefits only in the most severely depressed patients. But they say that these patients did not necessarily respond better to the drug; they simply did not respond as well to placebo as the less depressed patients.

"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," Kirsch said in a statement.

"This means that depressed people can improve without chemical treatments."

Given their findings, the researchers conclude that there is little reason to prescribe these antidepressant medications to any but the most severely depressed patients, unless alternative treatments have been ineffective.

The analysis is published in the journal PLoS (Public Library of Science) Medicine.

A similar study was published earlier this year in the New England Journal of Medicine. It found that antidepressants may be less effective than doctors have been led to believe because of "selective reporting."

The study, by Dr. Erick Turner, an assistant professor of psychiatry at Oregon Health and Science University, found that studies that have found antidepressants do little to help with depression are not being published.

And if studies that have found less-than-positive results have been published, they have been written as if they were positive.

Turner's team found that nearly all the studies (94 per cent) that were published in journals suggested positive treatment results. But FDA data showed that in fact only 51 per cent of the studies were positive.

Turner stressed that he didn't believe that the antidepressants studied were ineffective. In fact, his team found through an analysis of all the studies done on each drug that the medications were superior to placebos.

But after reviewing the FDA data on the drugs, his team found that the published literature often exaggerated the drugs' effectiveness.