Compulsory registration of clinical trials

“The case for registering all clinical trials—first advanced a decade ago1—is now unanswerable.”2 Editors of the BMJ and the Lancet made this statement in 1999. Five years of industry resistance, government impotence, and public confusion followed. Medical journals persisted with noble intentions and wise words but were themselves in part resistant, impotent, and confused about how to enforce registration. Some journals, including the BMJ, tried an amnesty for unpublished trials, with little success.3 The BMJ also considered asking for compulsory registration, but it seemed to us that trial registries were too diverse, disorganised, and easily disregarded to insist on registration before submission. Nor did we want fees for registering trials to be a barrier to researchers wanting to publish in the BMJ.

The world has changed. Undisclosed trials and duplicate and selective publication sting government agencies, clinicians, researchers, and journals ever more frequently and painfully. Crucially, this form of misconduct, which Iain Chalmers identified in 1990, distorts the scientific record.4 By suppressing negative findings and exaggerating positive ones, by downplaying harms and talking up benefits, healthcare decisions are based on incomplete data and ultimately harm the patients.

A BMJ theme issue last year spelt out the myriad entanglements between doctors and the drug industry, arguing that a new relationship based on transparency and trust was needed to protect patients.5 This year's speed of reform has put the drug industry on the back foot. Most notably, Eliot Spitzer, New York's attorney general, won a landmark legal case against GlaxoSmithKline forcing disclosure of undisclosed trial data on paroxetine in adolescents, data that support a long campaign to highlight its harmful effects.6

This week, many of the world's leading medical journals—those belonging to the International Committee of Medical Journal Editors (ICMJE)—announced a tough stance on trial registration. Trials that begin enrolment of patients after 1 July 2005 must register in a public trials registry at or before the onset of enrolment to be considered for publication in those journals. Trials that began patient enrolment on or before 1 July 2005 must register before 13 September 2005 to be considered for publication.7 This is a policy we support, and we will use the same deadlines for consistency—albeit with a couple of important differences.

Firstly, we believe it is wrong to be too prescriptive about the choice of registry because registries for trials are in an early stage of development. The ICMJE statement proposes a set of essential criteria and says that it “does not advocate one particular registry,” although only one registry meets its criteria: http://www.clinicaltrials.gov/, a registry sponsored by the US National Library of Medicine. This exclusive endorsement by the ICMJE journals is perplexing because clinicaltrials.gov offers registration only to “federal agencies sponsoring the clinical research studies (both interventional and observational trials), private sponsors that have submitted an Investigational New Drug Application (IND) to the Food and Drug Administration (FDA), such as pharmaceutical companies, and organizations representing IND sponsors.”8 These restrictive entry criteria will not be met by many trials worldwide. An inquiry from one of the BMJ's editorial board members about registering non-drug, non-National Institutes of Health trials emanating from low or middle income countries was met with this response: “The sorts of trials you have described are not eligible for registration in clinicaltrials.gov at this time.”

The BMJ's criteria for a suitable registry are listed in the box. As registries become more sophisticated we may have to revise these criteria. For example, because hundreds of registries exist there is already a role for a database—a metaregister—that aggregates trial information in individual registries. Individual registries should allow their contents to be aggregated. A way of recording or linking to the results of trials might become an essential criterion, as might information on competing interests and ethical approval.

Examples of registries that meet our proposed initial criteria are the International Standard Randomised Controlled Trial Number (ISRCTN) registry, managed by Current Controlled Trials—a commercial company that also owns BioMed Central and a metaregister (http://www.controlled-trials.com/)—clinicaltrials.gov, and the UK National Register of Cancer Trials (http://212.219.75.236/ukcccr). The ISRCTN registry is not publicly owned—public ownership is one of the criteria in the ICMJE statement—but its content is freely available to the public. Registration of trials in the ISRCTN registry is free for trials that are “sponsored, funded, and carried out in developing and transitional countries, in line with HINARI (Health InterNetwork Access to Research Initiative).” The registration fee for all other trials is £80 ($144; €117). Our view is that although public ownership is valuable, free public access suffices as an essential criterion—whether ownership is public or private—and that an international registry is preferable to one focusing on the research needs of one nation. This is not to undermine the ICMJE journals, but we believe that these modified criteria are more equitable and inclusive.

Will this move by journal editors work? On their own, journals can change little quickly, but with the public and governments (particularly the United States) whipping up a storm around the drug industry—and the researchers who conspire with industry to suppress findings that may harm patients and profits—journals have just done the public a service. The real question is how will industry respond? Missing out on the chance of publishing favourable findings in prestigious journals by not complying with this policy is undesirable for industry but the battleground will shift to monitoring of trials and ensuring that results—good, bad, or indifferent—are reported.