"Peer Reviewed Evaluation of Registered End-Points of Randomized Trials (the PRE-REPORT study): Protocol for a Stepped-Wedge, Cluster-Randomized Trial" was published June 1, 2019, in the journal BMJ Open.
It is well-accepted that the most reliable information regarding new medical treatments comes from well-conducted randomized trials. But are all published trials equally reliable? Isn't it possible that some published results are seriously biased and, therefore, unreliable?
Assistant Professor of Statistics Mark Weaver collaborated with Chris Jones from Rowan University and Timothy Platts-Mills from the University of North Carolina at Chapel Hill, along with other colleagues, to design a new study to investigate one practice that could lead to serious bias in trial publications, so-called "outcome switching." Before journals agree to publish the results of any trial, reviewers and editors should consider whether the primary outcome variable described in the manuscript is the same one that the trial investigators initially started off investigating. If it's not, that is, if the investigators changed outcomes during the trial or, even worse, during the analysis of the trial data, the interpretation of the results could be completely unreliable.
With funding from the Office of Research Integrity, Weaver and colleagues designed and initiated a trial among 13 top medical journals to test the effectiveness of a new intervention intended to reduce the prevalence of hidden outcome switching in published trials. The new study used a novel approach called a "stepped-wedge" design. In this design, all 13 journals began in the control condition. Then, each month during the 10-month study, one or two journals will be crossed over to the intervention condition so that by the end of the trial all thirteen journals will be receiving the intervention. The order in which the journals will be crossed over was randomized by Weaver prior to beginning the study.
The novel, yet amazingly simple, intervention developed by Jones and colleagues is to provide peer reviewers with information about the pre-registered primary outcome available from the site ClinicalTrials.gov. The goal is to decreases inconsistencies between prospectively registered and published primary outcomes. A paper detailing the protocol for this study was recently published in the open-access journal BMJ Open.