The British Medical Journal announced today that it and more than 100 other journals will urge medical researchers to register their unpublished clinical trials. A list of unpublished trials, to be made available on a Web site, will allow scientists planning new trials or reviewing drug effectiveness to find out about all previous trials, not just those in the published literature. Details of the plan will be announced at next week's International Conference on Biomedical Peer review in Prague, Czech Republic.
The goal is to offset a bias in the medical literature: "Doctors like to publish the good news, that a treatment works; they don't like to publish the bad," says Ian Roberts, director of the Child Health Monitoring Unit at The Institute of Child Health in London, which is also participating. "They are three times more likely to publish the results of a clinical trial if they're significant than if they're not." As a result, patients may be enrolled in studies whose questions have already been answered. And worse, doctors may be prescribing medicines that some researchers have found to be useless or even harmful.
The tendency to publish only positive results can also skew systematic reviews of the efficacy of a drug or medical technique. In these reviews, researchers collect all pertinent studies of the intervention and then assess the data collectively. The validity of these systematic reviews--sometimes known as meta-analyses--is undermined if negative or inconclusive trials have not made their way into the literature, says David Naylor, a clinical epidemiologist with the Institute of Clinical Evaluative Sciences in Ontario, Canada. "The net result will be an unduly inflated and excessively precise estimate of a treatment's effectiveness."