Over 6 weeks the COMPare team analysed 67 trials in the top 5 medical journals to assess the ongoing prevalence of outcome switching in clinical trials, and to find out why it persists. Our methodology can be found here. In this series we will be highlighting the salient lessons learned from this process by describing the analysis of several trials in detail. Each study highlights different ways in which outcomes can be misreported, leading to undetected flaws in the evidence presented in clinical trial reports.
Here we describe our coding of the ERASE trial . The context and overview of this study is described in this editorial. This study was 1 of 13 randomised control trials published in JAMA that we analysed. Of these, 11 contained outcome switching and required a letter of correction. All were rejected by JAMA’s editors; more on that, including full correspondence with JAMA’s editors, here.
Our summary analysis of outcome reporting in the ERASE trial was that:
3/3 pre-specified primary outcomes were correctly reported
21/37 pre-specified secondary outcomes were correctly reported
16/37 pre-specified secondary outcomes were not reported at all
6 additional outcomes were reported that were not pre-specified