RENEE MONTAGNE, Host:
NPR's Joanne Silberner reports.
JOANNE SILBERNER: The study had its start several years ago when psychiatrist Erick Turner worked for the Food and Drug Administration. He was evaluating a drug company application to market a new antidepressant. He had access to all the research that had ever been done on the drug.
ERICK TURNER: In going through some of that, I was noticing quite a few negative studies, studies where the drug did not beat placebo to an extent that was statistically significant.
SILBERNER: At first, that surprised him. He figured he was up on things and would have known about any negative research - studies showing the drug didn't work. Then he realized the problem.
TURNER: The reason, of course, that I hadn't almost because I was relying like any physician does on a published literature and was realizing these studies were not getting published.
SILBERNER: What he found confirmed what he suspected - 94 percent of the studies that showed the drugs worked made it to publication. However...
TURNER: For the studies that were not positive, only a minority of those studies were published. It turned out to be only three out of 36, which is eight percent.
SILBERNER: Jeff Drazen is the editor-in-chief of the New England Journal of Medicine, which published Turner's study. Drazen wasn't surprised by the finding.
JEFFREY DRAZEN: This is about what we had suspected, but we didn't have these kinds of good quantitative data to show the magnitude of the problem.
SILBERNER: He now have a policy. Drug companies have to announce that they're starting a drug trial at the very beginning before they know what the results will be. Otherwise, the major journals will refuse to publish the results.
DRAZEN: Having at least an understanding of the trials that have not been reported allows you to compensate when you are examining the effect of a drug for those unreported trials. If you don't know about them, you can't compensate.
SILBERNER: The major drug company trade group says that many of its members have been voluntarily posting results of unpublished studies. Drazen says that's the least researchers can do for their patients who volunteer for experimental drug trials.
DRAZEN: Because they put themselves at risk, we, as a profession, owe them the dignity of making sure that their period at risk is not hidden away, so the company can make money.
SILBERNER: Joanne Silberner, NPR News.