Quackbusters
Editors rarely have much to delight them on cold Friday mornings in Cambridgeshire so here at PLoS-UK HQ there was a frisson of excitement over Ben Goldacre’s latest blog. It seems faintly ridiculous that around 50 years after the first properly randomized trials were done (and please don’t quote me on specifics here, because I’m not a trials historian – check out the James Lind Library if you want more precision), anyone has to defend the importance of scientific rigour, appropriate controls, and dispassionate appraisal of all the available evidence. And yet with respect to the practice of homeopathy, someone has to go to the trouble of laying out the basic principles of evidence-based medicine: just read the blog to find out why.
Ben also talks about publication bias. We have no idea whether this is a more serious problem for homeopathy trials than for other types of study, but it has been estimated that around the results from around half of all trials never make it into the literature. As pointed out in the blog, clinical trial registries are a crucial way of addressing this problem. The essential idea behind registries is that 20 basic pieces of information about every trial get deposited into a publicly-accessible database before any patients are recruited. Then, when a trial has been completed, everyone can match up the results with the original plan for the study, and it will also be obvious if a trial (or specific outcomes from a trial) have not been reported.
As I’ve said probably far too many times, registries are a crucial step forwards, but ultimately don’t go far enough. We also need a way to ensure that the results are eventually reported. This is an area within which the World Health Organization’s group which developed the International Clinical Trials Registry Platform has been investigating international norms and standards. At PLoS we are also committed to publishing the results of all correctly conducted and reported trials, irrespective of outcome, within PLoS ONE, and making these available via our Clinical Trials Hub. And, of course, PLoS Medicine publishes rigorously conducted trials which the editors think will significantly advance clinical management or public health policy.
U.S. Public Law 110-85, Section 801 (“Expanded Clinical Trial Registry Data Bank”) addresses some of these issues. One part expands the scope of the existing registration data set items. More information about the expanded registry is availble at http://prsinfo.clinicaltrials.gov/fdaaa.html.
Another part of Section 801 requires submission of “basic results” for completed trials, whether published or not, to ClinicalTrials.gov. Demographic and baseline characteristics, primary and secondary outcomes, adverse events, and other data will be available to the public. Thus, the combination of (1) more comprehensive prospective clinical trials registration and (2) basic trial results reporting — with enforcement via penalties for noncompliance — should help to mitigate publication bias, together with efforts such as the PLoS Hub for Clinical Trials and the International Committee of Medical Journal Editors registration policy.
Absolutely, this is a very encouraging step forward – and it will be interesting to see how the results reporting component within ClinicalTrials.gov is eventually implemented. One question that I’m sure interested parties are asking is, how will quality control work; how can we be sure that all data has been reported?