News

Scientists who do not publish trial results are "unethical"

Gavin Yamey , BMJ

Failure to report the results of randomised trials constitutes scientific and ethical misconduct, Dr Iain Chalmers, director of the UK Cochrane Centre, told a press conference in London last week. Dr Chalmers and the editors of the BMJ and the Lancet called for an international register of all clinical trials to be established.

Dr Chalmers said that "prospective registration and public access to the results of all randomised trials should be required by organisations responsible for protecting the public." These organisations should include drug licensing authorities and research ethics committees.

Trials should be registered, he argued, to inform patients, clinicians, and other decision makers about trials in which they could participate, toprevent costly research duplication and to promote multicentre trial collaboration. "A substantial problem remains," he said, "because studies with disappointing or negative results are less likely to be submitted for publication." He presented several examples of such biased underreporting of research.

A randomised trial of the class I antiarrhythmic drug lorcainide in patients with acute myocardial infarction, carried out in 1980, remained unreported for 13 years. The increased death rate in the lorcainide group was thought to be a chance finding, and the use of class I antiarrhythmic drugs generally increased.

"At the peak of their use in the late 1980s," said Dr Chalmers, "it has been estimated that class I antiarrhythmic drugs given to people with heart attacks were causing between20 000 and 70 000 premature deaths every year in the United States alone. This yearly total of deaths is of the same magnitude as the total number of Americans who died in the Vietnam war."

Dr Chalmers welcomed the launch of an online register of randomised controlled trials (www.controlled-trials.com) but warned that the comprehensive registration of planned, ongoing, and unpublished controlled trials is unlikely to happen "unless there is legal clout."