Education and debate

Registering clinical trials

Alison Tonks, papers editor. 

BMJ

atonks{at}bmj.com

Randomised trials of medical interventions are the foundation of evidence based health care, but most are effectively conducted in secret. Few peopleusually an elite network of investigators, funding agencies, and government regulatory bodiesknow about a trial from its inception. Most trials become public knowledge only when the investigators publish their completed project in a journalif they ever do.¹ In the meantime, others may be duplicating the effort or, worse, ignoring early warning signs that an intervention is dangerous. One trial of the class I antiarrhythmic drug lorcainide, for example, went unreported for over a decade even though the data suggested that the drug increased mortality in patients with myocardial infarction. During that time, use of such drugs continued and shortened the lives of up to 70 000 people each year in the United States alone.² A register of clinical trials is one way of opening up the process and avoiding these problems.

The idea was first mentioned 13 years ago³ and has been refined since then by an international group of trialists, academics, and enthusiasts campaigning for a comprehensive, up to date, and searchable archive of ongoing and recently completed randomised trials, including trials done by the pharmaceutical industry. This article reports a recent conference on trial registration hosted in London by the BMJ, the Lancet, and the Association of the British Pharmaceutical Industry. The article was prepared from conference presentations and subsequent debate, discussion with selected speakers and delegates, reviews by those speakers who are running their own registers, and a bibliography of studies prepared by Iain Chalmers, director of the UK Cochrane Centre.

	What is a trial register?

Registering a clinical trial of a new drug or intervention means putting on public record some basic information about the trial from its inception. The aim is to provide reliable intelligence about research in progress to the public, health providers, researchers, and funding bodies (box 1 lists the people and organisations with a stake in trial registration). There is still no consensus about the kind of details that should be registered, but the hierarchy of options is listed in box 2. Comprehensive registration of full protocols and eventual results makes the enterprise richer and more useful but may discourage pharmaceutical companies wanting to safeguard commercially sensitive information. There is also no international agreement about what sort of trials should be included. Randomised trials of new drugs are top of everyone's list, but other interventions, such as surgery or new diagnostic techniques are surely as important.

	Why register trials?

Pioneers of trial registration have been writing and talking about it for decades. Their most powerful argument is that unregistered and unreported trials cannot contribute to the evidence base for healthcare decisions. Only a biased subsample of all trialsthe published onesis included in systematic reviews and meta-analyses. Publication bias is a serious problem that costs lives. Box 3 lists other good reasons to register trials. There are few dissenting voices in this debate. Patients, researchers, funding agencies, governments, publishers, and two leading international pharmaceutical companies (the others are watching and waiting) all agree that clinical trials should be registered in the interests of evidence based medicine and freedom of information.⁴ The remaining question is not whether it should be done but how the effort should be coordinated.

	Who is doing it?

There are already hundreds of clinical trial registers worldwide. There is even a register of registers, begun in 1987 and held at Brown University in the United States, which lists over 500 online registers of clinical trials. Three case studies (a selection of registers presented at the conference) illustrate the kind of initiatives being developed by the pharmaceutical industry, national governments, and science publishers. Clearly, many trials are already registered, but often in registers that are inaccessible even to researchers. Most are managed in isolation, using "stand alone" software, which frustrates the simplest search for information. Recorded details vary dramatically among registers. There is no guarantee that a register is complete, accurate, or comprehensive. Worse, registered trials seem to be a biased subset of all trials: 60% of registers are confined to AIDS or cancer trials, and most cover only drug trials and not other interventions (K Dickersin, personal communication). The existing network of registers is therefore valueless to anyone but a small group of cognoscenti, and only of limited value to them.

Many countries have legislation covering the conduct of clinical trials, and in some, such as the United States and Spain, trials cannot begin legally until they have been cleared and registered with a central body, usually the drug regulatory agency. The US government is leading the charge with its Food and Drug Administration Modernisation Act 1997, which requires the establishment of a prospective database of all trials of new treatments for serious or life threatening diseases.⁵ The system is under construction at the National Library of Medicine, and at least some parts of it should be publicly available soon. In Britain the only legal imperative is that trials of new drugs must be registered with the Medicines Control Agency as part of the licensing process. Details are submitted in confidence by pharmaceutical companies and kept secret by the agency. The Medical Research Council, a leading source of funding for UK research, also requires investigators to register their trials before it releases funds.

	Why isn't everybody doing it together?

The barriers to unifying trial registration across the world are formidable and fall into two main categories. Firstly, there are practical difficulties such as capturing all trials without duplication from a wide variety of sources, finding money to fund free access, agreeing on a minimum dataset, navigating different legal systems, and developing information systems sophisticated enough to search within and between different registers. Maintaining a register is hard work, can be tedious, and is certainly expensivean unpromising trio of attributes.

Secondly, there are trickier human problems such as cultural differences between commercial and research organisations, lack of incentives for researchers to spend time registering trials, competition among organisations running their own registers for profit or influence, questions of ownership and copyright, and the legitimate concerns of the drug industry about releasing sensitive information before a new drug is licensed. From an industry perspective, sharing information means sharing power, and so far only Glaxo Wellcome and Schering UK have signed up. Others will follow their lead only if they see that these companies gain credibility without financial penalty.

	Making it happen

The dialogue about trial registration, which has been gathering momentum for two years, is poised to move from good intention to a new phase of implementation with the help of four influences: the law, the consumer movement, ethical review of research, and advances in information technology.

Linking ethical approval to registration
Capturing trials is a major problem for register organisers. Linking registration to funding is one option, but what about all those trials without an external source of funding? All clinical trials must go through a process of ethical approval, and opinion leaders have suggested that the machinery for this (ethics committees in Britain, institutional review boards in the United States) should be used to guarantee registration. ^{8 9} If research ethics committees policed the system by insisting on registration before granting ethical approval, accrual would leap to almost 100% overnight. Disappointingly, this relatively simple step has not yet been taken in Britain despite clear signals that it would be. The Australian research ethics community also supports the notion of registration before ethical approval. Again, institutional ethics committees have been earmarked for the job.

The law
It is too early to tell whether legislation can be used to force the issue, and in which cultural environments it is best used. The Food and Drug Administration Modernisation Act 1997 has worked in the United States, although its prototype register took two years to develop. The force of law will be of value only if the legislation makes registers open to the public. In Spain legal regulation of drug trials has supported a national trial registry since 1982, but the Spanish government has yet to allow access to researchers and the public. Negotiations between the Spanish Cochrane Centre and the government are continuing about what information should be released and when. Finally, there are plans for a European directive on controlled trials to unify international trial efforts. It states that a European database of trials should be set up.

The consumer movement
Patients are the real consumers of medical research. They buy it through taxation and charitable donation, participate in it, and use the results (mediated by doctors) to improve their health and alleviate disease. They rightly expect that clinical decisions are based on all available knowledge, not just the biased sample that appears in medical journals. A forceful consumer lobby is a powerful ally when it comes to persuading politicians to promote trial registers. Pressure from advocacy groups for patients with breast cancer ultimately led to a change in US law. In Australia the Consumers' Health Forum is one of several consumer organisations that advocate prospective registration of trials. Patients' organisations in Britain, however, have been largely silent on the subject, possibly because public understanding of clinical trials is still inadequate. More could be done to encourage the UK consumer lobby to speak out.

New technology
The recent international explosion of stand alone trial registers and the diversity of organisations running them makes it look increasingly unlikely that a single physical register will ever exist. Fortunately, the need for one is diminishing as internet technology develops to harmonise searching across registers. If all registers were linked, individual organisations would be free to decide how much information to post and who should get itwithin an internationally agreed minimum requirement. An Amish quilt illustrates the idea (see fig). Current Controlled Trials has already implemented such a model in its metaregister of controlled trials (www.controlled-trials.com). A group at Brown University in the United States is also working on a virtual unified register made up of a web of existing registers.

View larger version (104K): [in this window] [in a new window]   "Broken star" quilt illustrating a web of registers based on a minimum data set, with individual outputs for users and flexible levels of access

	Conclusion

Registering a clinical trial is a public declaration of intent by those doing the work, and those paying for it. That such a diverse range of organisations, including commercial companies, are committed to openness about clinical trials is a triumph of common sense over chaos. Many trials are already registered, but there is some way to go before BMJ readers and their patients can search the international research effort for information about trials in progress in the same way that they can search the international literature for results of trials. The problems that remain are largely practical ones, which lend themselves to two practical solutions: passing legislation to outlaw unregistered trials, or convincing ethics committees to withhold approval until a trial is publicly registered.

	Footnotes

   Competing interests: None declared.

	References

1.	Stern JM, Simes RJ. Publication bias: evidence of delayed publication in a cohort study of clinical research projects. BMJ 1997; 315: 640-645[Abstract/Free Full Text].
2.	Yamey G. Scientists who do not publish trial results are "unethical." BMJ 1999; 319: 939[Free Full Text].
3.	Simes RJ. Publication bias: the case for an international registry of clinical trials. J Clin Oncol 1986; 4: 1529-1541[Abstract/Free Full Text].
4.	Horton R, Smith R. Time to register randomised trials. BMJ 1999; 319: 865-866[Free Full Text].
5.	A national resource for information on clinical trials. Natl Forum 1999; 79: 19-21.
6.	Smith R, Roberts I. An amnesty for unpublished trials. BMJ 1997; 315: 622[Free Full Text].
7.	Roberts I. An amnesty for unpublished trials. BMJ 1998; 317: 763-764[Free Full Text].
8.	Blunt J. Research ethics committees must ensure registration of research and accessibility of results. BMJ 1998; 316: 312[Free Full Text].
9.	Savulescu J, Chalmers I, Blunt J. Are research ethics committees behaving ethically? BMJ 1996; 13: 1390-1393.