Open access to industry's clinically relevant data

Last month GlaxoSmithKline announced that it would publish summaries of all its clinical trials of a new product once it had been launched.1 This decision followed news of a lawsuit brought by New York State alleging that the company had concealed the results of trials of paroxetine because they might have spoilt marketing plans. GSK said it had been considering the move for some months. A similar sounding policy was announced by Glaxo Wellcome in 19982 but seems to have been quietly abandoned in 2000 after the merger with Smith KlineBeecham.

The arguments for free public access to all clinically relevant data on a company's drug have been stated many times: clinicians, patients, and the institutions that pay for health services all need the data to make good choices and to use drugs in the best ways, maximising their benefits and minimising harms. That is true not only for individual drugs and treatments, but also for the more efficient and speedy management of knowledge. Systematic reviews of treatments (both of individual treatments and of a range of treatment options for a problem) are bound to be biased if important studies are kept secret, and future research is misrouted or impeded if lessons from hidden studies cannot be assimilated.

The International Committee of Medical Journal Editors is considering requiring all clinical trials submitted for publication to be listed in a registry. The American Medical Association has decided to urge the Department of Health and Human Services to establish a comprehensive registry for all clinical trials and require every trial to have a unique identifier. Over 300 clinical trial registries currently exist: they are hard to use and not comprehensive.

Industry's argument against free access has always been that it has paid for the research and therefore owns the results, which are “commercially sensitive,” a sweeping notion that covers trade secrets, “might help competitors,” “could affect the share price.” Competition thus rewards those who best keep secrets, not those who have the best drug and the data to prove it. It also means that anything to do with harms, which are sensitive commercially, tends to remain buried, and authors will feel less accountable. This denies the public interest and ignores the contributions of the participants, investigators, and the institutions where the work was done. Industry is beginning to recognise that secrecy is not “perceived” to be in the public interest. The Association of the British Pharmaceutical Industry (ABPI) launched its voluntary clinical trial register in May 2003. The register holds outline details of phase 3 trials of a licensed medicine in UK patients. So far only six member companies have registered 93 trials,3 and the information about each is sparse. The ABPI seems satisfied, saying: “Since the launch it has been possible to refute a lot of the criticism about the perceived secrecy of the industry in the clinical trial area.”

The US industry association, Pharmaceutical Research and Manufacturers of America (PhRMA), has acknowledged criticism in its statement on public disclosure of clinical trial results,4 just updated.5 This notes, however, that exploratory studies (“early phase or post-marketing”) are often highly proprietary to the sponsoring company and of low statistical power, so that sponsors do not commit to publish the results of every exploratory study, nor “to make the… clinical trial protocols available at inception, as in a clinical trials registry.” But now the momentum for meaningful disclosure has moved to a higher plane—partly fuelled by a succession of unexpected and worrying revelations about the harm caused by selective serotonin reuptake inhibitor (SSRI) antidepressants.

Even if other companies copy the policy announced by GSK we would lack much of the information we need. The data produced by the company would be abstracts that may well be incomplete and biased. Too many published clinical trials, whether industry-sponsored or not, do not comply with the CONSORT (consolidated standards of reporting trials) guidelines (http://www.consortstatement.org/), and most are seriously deficient in detecting and reporting adverse events,6 a problem to be addressed in the next revision of CONSORT.

To bring the reporting, interpretation, and dissemination of clinical trials into the 21st century will need a lot of work. Journal editors and participants in the Cochrane Collaboration are trying hard to improve matters, but others who should be concerned have shown little interest. Research ethics committees (institutional review boards) could contribute by pressing study investigators and sponsors to undertake to make the results—whether positive, negative, or inconclusive—publicly accessible within a reasonable time of completion (or abandonment). Regrettably, the new UK clinical trial regulations may prevent their making it a condition of approval.7

Drug regulatory agencies in particular could do most to help progress. They receive huge volumes of data to digest at speed. This may help to explain why the US Food and Drug Administration approves a drug if two studies of sufficient size establish its superiority over placebo, but then discounts the results from studies that show no evidence of drug efficacy.8 Regulators should recognise that the data on which they base their decisions must be made available to the scientific community, clinicians, and the public as soon as a drug can be prescribed. They behave as if pharmaceutical companies were their primary customers. That will continue as long as industry funds drug regulation: government should fund it fully and independently as a vital part of public health, and separate bodies should deal with licensing and pharmacovigilance. UK law still forbids regulatory officials to disclose any information “obtained by or furnished to [them] in pursuance of” the Medicines Act, but this provision will be repealed next year. “Commercial confidentiality” should be confined to details of manufacture and formulation, not to clinical trial methods, data, or results. In the words of a former chairman of the Committee on Safety of Medicines, “the major proportion of individual [licence] applications, could, with little loss to anyone, be made publicly available.”9