New law may be having some effect on publication bias.

P ositive trials get reported more often than negative trials, according to many studies. And despite efforts in the past decade to combat publication bias, it remains a substantial problem, according to two articles published last fall. One found that results had been published from fewer than half of the trials registered in the federal database, clinicaltrials.gov. The second article found that approximately one-third of the trials eventually published in journals changed the trial endpoint prior to publication. Despite the disconcerting results, experts are cautiously optimistic that a federal law may be having an effect. The law requires not only registration of trials but also that researchers enter results into the database. Journal policies had previously encouraged trial registration, but the new law adds clout. “The only way trial registration was going to have an effect was if someone was going to enforce it, and this is the enforcement that is needed. I think it could well address publication bias,” said Virginia Barbour, M.D., chief editor of the open-access journal PLoS Medicine, in the Public Library of Science in London. Although experts look forward to evaluating the infl uence of the law, no one — including Barbour — expects that it can entirely cure the problem. Other efforts are needed, experts say, including increased scrutiny of trial protocols and registration information during the peer-review process.