Large trials that go unreported threaten the credibility of available, published evidence and are detrimental to the scientific process, according to a brief research report and an editorial published in Annals of Internal Medicine.
Researchers identified the 500 largest preregistered randomized controlled trials (RCTs) with a start date after June 1, 2007, and completion date before June 1, 2012 that lacked reported results on clinicaltrials.gov and remained unpublished more than 46 months after completion. Nonrandomized, cluster-randomized, and non-preregistered trials were excluded from the analysis. The researchers evaluated whether these trials were subsequently published or had posted any results through January 2019. Whenever possible, investigators were contacted by email to verify publication status.
The investigators found that 60 of the 146 large RCTs without main publications listed or results posted on clinicaltrials.gov as of April 2016 had in fact been published by that time. Of the 84 RCTs that were indeed unreported as of April 2016, only 14 (17%) were published by January 2019. Searches at clinicalstudydatarequest.com for the 70 unreported trials resulted in no data. The trials that went unreported were typically the largest to assess a specific treatment comparison for a specific disease or condition. As an example of the significance of this lack of reporting, the researchers found 5 long-unreported large trials on antidepressants; the largest meta-analysis on the same topic found data on only 32 trials of equal size. The investigators conclude that large trials that go unreported for 4 or more years after completion are unlikely to be published, resulting in a loss of evidence from almost 90,000 study participants.
The study is limited by the use of only 1 registry, and the researchers could not determine the exact proportion of total evidence represented by the unreported trials.
In the accompanying editorial, the authors express concern that the credibility of published evidence is threatened by large unreported trials when extensive results remain undisclosed. Failure to report clinical trial results also threatens the public’s trust in research and the integrity of medical literature. The authors believe that such failures of disclosure should be considered academic misconduct at the individual and institutional levels. Participants who volunteered to take part in clinical trials are potentially exposing themselves to interventions of unknown safety and efficacy profiles under the assumption of trust that the evidence generated will inform clinical science. When research findings are not consistently disseminated years after their study completion, “patients, institutional review boards, clinicians, researchers, and the public must rely on incomplete evidence, which may lead to misconceptions about the efficacy and safety of interventions.”
Tatsioni A, Karassa FB, Goodman SN, Zarin DA, Fanelli D, Ioannidis JPA. Lost evidence from registered large long-unpublished randomized controlled trials: a survey [published online May 7, 2019]. Ann Intern Med. doi: 10.7326/M19-0440
Wallach JD, Krumholz HM. Not reporting results of a clinical trial is academic misconduct [published online May 7, 2019]. Ann Intern Med. doi: 10.7326/M19-1273