Information on all worldwide active and completed randomized controlled clinical trials should be made available through a central database to avoid unnecessary duplication of medical research, participants at an international meeting held in London on 4 October insisted.

Such a database would prevent unethical testing of patients, whose enrollment in trials might add little value to current knowledge; prevent wastage of public research funds; promote research cooperation; and flush into the open negative results that now escape publication.

The detrimental effect of publishing only positive medical information is widely acknowledged. In an example of biased under-reporting of negative data, Iain Chalmers, director of the UK Cochrane Centre, Oxford, cited the case of the anti-arrhythmic drug lorcainide. A 1980 study that revealed increased mortality of the drug went unpublished for 13 years because researchers thought that the increased death rate "was an effect of chance." This class of drug was used in the intervening period and has been associated with thousands of deaths.

The UK has 3,000 trials involving half a million people taking place at any one time. Databases are maintained by the Cochrane center, the Medical Research Council and the largest by the National Health Service. The UK's Chief Medical Officer, Liam Donaldson, voiced his support for a centralized database, but did not say who would fund or coordinate the effort.

Such a database would be incomplete without trial information from private companies, and so far only two pharmaceutical companies—Schering Plough and Glaxo Wellcome—have committed to making data on licensed products available. It is unclear whether other companies will volunteer similar information.

According to Alexa McCray from the US National Library of Medicine (NLM), the Food and Drug Administration Modernization Act of 1997 requires that a registry of trials for both private and federally funded work in the area of "serious and life-threatening conditions" is created by the NLM. This database will be publicly accessible in January 2000, but will only contain information on active trials in which patients may wish to enroll. It will not contain results from completed trials.