Positive skew of clinical-trial publication
A news story in Nature Medicine (14, 1133; 2008) discusses an investigation into the publication status of the clinical-trials literature (PLoS Med., doi: 10.1371/journal.pmed.0050191; 2008), which concludes that positive results of clinical trials for drugs or devices have a higher chance of getting published than negative trials. The study further concludes that when the trial sponsors publish the results from 'pivotal' trials only 76% of the time.
Studies reporting a statistically significant difference were more than three times as likely to be published. This 'positive publication bias' is a serious problem, because it can make a drug or device appear in the literature to be more effective than it is. Ida Sim, a co-author of the study, told Nature Medicine: "We have this idea of practicing evidence-based medicine, which is predicated on having a full and complete evidence base. But when the evidence base is skewed, we can't really do this."
According to the Nature Medicine article, a paper in Science (319, 1340–1342; 2008) indicates that "the FDA Amendments Act of 2007 has improved transparency, because the law mandates that sponsors or primary investigators of clinical trials for approved drugs post a summary of their results in a national open-access database. The lead author of the report, Deborah Zarin, oversees the ClinicalTrials.gov registry at the National Library of Medicine of the US National Institutes of Health and is in charge of ensuring the results are posted in compliance with what the new law. According to Zarin, "for the trials that are covered by this law, the results database should have a big impact on disseminating medical knowledge, because the results have to be publicly available."
But not every type of clinical trial is covered by the legislation, nor does it directly affect medical journals. Although Sim applauds the FDA Amendments Act of 2007, she adds that it "doesn't address the issue of not publishing trials in medical journals. They remain one of the most influential and biased sources of information." "
