BAKERSFIELD, Calif. -- In 2006, there were more prescriptions written for antidepressants than any other class of medication in the United States. But a new analysis has suggested that the effectiveness of antidepressants may be overstated in published medical literature.

Doctors from the Portland Veterans Affairs Medical Center compared the results of published studies on antidepressants with the results of unpublished studies. They found that published studies were overwhelmingly favorable, with 94 percent reporting positive effects of the drugs.

In contrast, 96 percent of unpublished studies found questionable or no benefits to antidepressants.

Researchers said that this bias is common in medicine -- drug companies are more interested in publishing positive findings, and medical journals prefer to publish research that demonstrates a treatment benefit.

But this bias can paint a misleading picture of a drug's effectiveness. The full analysis also revealed that similar antidepressants had different levels of demonstrated benefit.

Paxil and Celexa both had roughly equal amounts of positive and negative findings, while studies on Prozac and Effexor were much more positive than negative.