A study published yesterday in PLoS Medicine, a publication of the Public Library of Science, found discrepancies between the information in published drug trials and the trial data that the U.S. Food and Drug Administration reviewed. In addition, following FDA approval, many of the trials remained unpublished, even five years after the trials were complete.
Thus, the scientific literature often contained "incomplete and potentially biased" information, according to the study.
The observational study looked at efficacy trials between 2001 and 2002 to determine whether the trial results listed in New Drug Applications (NDA) for medications submitted to the FDA end up published in medical journals. They also searched for discrepancies between trial data included in the NDAs and articles published between July 2006 and June 2007.
It isn't the first time that studies of drug trials revealed selective reporting, according to the study authors, who found that 25 percent of the efficacy trials in the NDAs went unpublished and that "trials with favorable outcomes were nearly five times as likely to be published as those without favorable outcomes." For example, while both the papers and the NDAs listed 155 primary outcomes, 41 primary outcomes listed in the NDAs were not in the medical literature at all.
As for bias, the researchers found differences in nine out of 99 conclusions between what was in the NDA and what ended up in the published paper. Not surprisingly, the published version was positive about the drug under consideration in every case.