Journal of IiME Volume 9 Issue 1 than half had failed to post results; and even though the FDA is entitled to issue fines of $10,000 a day for transgressions, no such fines have ever been levied [3]. In the face of such slow progress, this commentary sets out some practical suggestions for auditing, performance tables, accountability, codes of conduct, and better data that should help to drive up standards and prevent trial reports being withheld from those who need them most. What Should Trials Transparency Look Like, and How Do We Achieve It? The WHO statement calls for summary results to be both posted on a registry and submitted to a journal within 12 months. However, it is worth noting that academic journal publication may ultimately prove to be a red herring, as an indicator of transparency. Academic publishing decisions can be arbitrary, and introduce lengthy delays in access to knowledge. Furthermore, there is a growing body of evidence demonstrating that journals often fall short of the basic expected standards for reporting of clinical trials. It is commonplace to find that primary outcomes have been switched, for example [7]; findings are routinely “spun” [8]; and compliance with reporting standards such as CONSORT is highly variable. When compared with the long and formal structured Clinical Study Reports created for all industrysponsored trials, academic papers have been shown to be incomplete and inconsistent [9]. However, since all clinical trials are fundamentally similar—when compared, for example, with the myriad study designs in molecular biology—it has been possible to develop reporting standards and operationalise these. Reporting results onto a structured database, such as the results tab of clinicaltrials.gov [10], has many preferable features: there is minimal delay, there is compulsory reporting of features that are required; and there is no possibility to switch pre-specified outcomes or other forms of May 2015 reporting misconduct. Put simply, there is a box to report the pre-specified primary outcome, and it has to be filled. Recent research has shown that academic journal reports are inconsistent with those on clinicaltrials.gov [2] and contain less complete information on methods, results, and adverse events [11]. Furthermore, International Committee of Medical Journal Editors (ICMJE) member journals have explicitly stated that they will not reject trial reports on the grounds that the results have already appeared on clinicaltrials.gov, and that they do not regard registry results reporting as prior publication [12]. Lastly, clinicaltrials.gov is clear that they will accept results on any trial, from any era, on any treatment, from any territory. This negates a key defence commonly cited by trialists and sponsors when facing calls for greater transparency: that journals reject “negative” results. All trials can now be reported, immediately, using clinicaltrials.gov as a first or last resort, if the trialist is willing. The question remains: how can we ensure this is done? The Need for Audit One key element is likely to lie in medicine’s most basic research tool. Audits are routinely conducted on local service issues, such as infection rates, or waiting times, but rarely on broader structural issues such as publication bias, even though the impact of the latter on patient care is likely to be greater and global. Indeed, it is peculiar that for many years trial registration was considered an end in itself, when in reality registration is only of value as the raw material for publication audit. The basic structure for a routine ongoing audit of results reporting is simple: using a register, identify trials that completed more than 12 months ago; establish, through whatever means, whether results from the trial have been reported; and post the date of results appearing to the register. From this, it is trivial to derive performance metrics for individual companies, Invest in ME (Charity Nr. 1114035) www.investinme.org Page 13 of 57
14 Publizr Home