Clinical trials are the best method for determining which treatments are effective. Although some treatments have very obvious benefits, most have modest effects. It is in these common circumstances that randomised controlled trials (RCTs) are needed to distinguish real effects from the influence of biases. The recent hormone replacement therapy story is one of many examples in which non-randomised studies have proved to be seriously misleading.1

Effective healthcare clearly requires health professionals to be fully informed of the evidence from RCTs. Unfortunately, access to the results of all RCTs remains limited, despite the efforts of the Cochrane Collaboration. A recent review found that only about half of the RCTs presented at conferences went on to be published, even though controlled trials were more likely to be published than other types of studies.2 The main reason for this appears to be triallists’ failure to submit, based on their perception of whether or not the work is likely to be accepted.3 Even if the work is submitted, publication may be delayed for years if the results are negative,4 contributing to cycles of early excitement followed by disappointment as all the evidence emerges.5

Failure to publish can also lead to unacceptable delays in uncovering harmful therapies. For example, the use of class 1 antiarrhythmics after myocardial infarction received FDA approval in the early 1980s, despite the lack of evidence that these drugs reduced mortality. By 1993, evidence had accumulated that they actually increased mortality,6 prompting the first publication of a negative trial conducted 13 years earlier.7 The increased mortality was originally thought to be a chance finding, but thousands died in the resulting interim before publication.8

For these reasons, failure to publish has been described as another form of scientific misconduct.9 However, even if trials are published, locating them can be difficult. A handsearch of the NZMJ found 143 RCTs published after 1965, only 89 (62%) of which were identifiable as trials on Medline.10 Similar findings are seen for other journals.11

Prospective registration, with free public access to that information, is the only means of detecting potentially biased under-reporting, and reducing difficulties in locating RCTs. Prospective trials registration would minimise the impact of failure to submit by authors, and failure to publish by journals. A register of RCTs would also:

The best strategy for trials registration in New Zealand would involve research ethics committees. These committees are already notified of all trials, and none can proceed without their approval (unlike alternatives, such as MedSafe or the proposed Therapeutic Goods Agency, which is notified only of trials of unregistered drugs or devices; and local research funding bodies, which fund only a fraction of the RCTs conducted in New Zealand). It would be a relatively simple matter for ethics committees to require an International Standardised RCT Number (ISRCTN) to be included in ethics applications as evidence of the registration of a trial. An ISRCTN is currently available through the meta-register of RCTs at www.controlled-trials.com. With a little more effort, a simple Internet-based local site could deliver details to the meta-register, while providing triallists with ISRCTNs and New Zealand with a national trials registry. The technology is secure, affordable and robust.

The issue of commercial sensitivity is a common challenge to the registration of trials, but it can be overcome by requiring only broad details. Debate should focus on the level of detail available, not the principle of trial registration. Notably, the Association of the British Pharmaceutical Industry advocated that from 2001 its members should register all their trials.12 Trial registration is fast becoming recognised as good research practice.13 The UK Medical Research Council has, since 1998, required all new RCTs to be registered as a condition of releasing grants awards; the NHS Research Governance Framework now requires registration of all clinical trials undertaken within the NHS; and the US National Institutes of Health has developed ClinicalTrials.gov to provide members of the public with current information about clinical research studies. New Zealand punches above its weight in terms of performing RCTs – it is time to join the crowd, register these trials, and make the information publicly available.

Author information: Andrew Jull, HRC Foxley Fellow, Clinical Trials Research Unit, University of Auckland; Iain Chalmers, Editor, James Lind Library, Oxford, UK; Anthony Rodgers, Co-director, Clinical Trials Research Unit, University of Auckland, Auckland.

Correspondence: Andrew Jull, Clinical Trials Research Unit, Department of Medicine, University of Auckland, Private Bag 92019, Auckland. Fax: (09) 373 1710; email: a.jull@ctru.auckland.ac.nz