662 Pre-trial registration as accountability in manuscript publication

The reproducibility crisis raises concerns about scientific rigor in randomized controlled trials (RCTs). Pre-trial registration has been proposed to improve rigor by requiring pre-defined methodology and outcomes and is required by the International Committee of Medical Journal Editors. We aimed to evaluate pre-trial registration compliance, and concordance between registry and publication for outcomes and sample size, in obstetrics and gynecology (OBGyn) RCTs. We performed PubMed and manual searches to identify all OBGyn RCTs published in three OBGyn (Obstetrics and Gynecology, AJOG, BJOG) and three non-OBGyn (NEJM, JAMA, The Lancet) journals between July 2017 and June 2019. We excluded non-human and non-parallel arm RCTs. Data were collected in duplicate from the published study and corresponding pre-trial registry. We determined concordance between registry and publication for outcomes and sample size. Discrepancies were recorded. Pre-trial registration and outcome concordance was also compared between OBGyn and non-OBGyn journals. Of 170 manuscripts included, 166 reported pre-trial registration (98%) and were analyzed. Only 125 studies (75%) were confirmed to have pre-registered prior to the start date of the trial. Concordance between registry and publication was identified for 77% of primary outcomes with differences most frequently attributable to omitting the registered primary outcome (8%) or reporting it as secondary (8%). Registry and published secondary outcomes were concordant in only 32% with differences most frequently attributable to new secondary outcome adoption in the publication (48%, Table 1). Sample size was discrepant in 40% of studies with 91% providing no explanation. Pre-trial registration was higher in non-OBGyn journals compared to OBGyn journals (86% v 70%, p=0.02) but outcome concordance and sample size were comparable across journal type (Table 2). Despite the adopted requirement, not all published OBGyn trials are pre-registered and major discrepancies between registry and publication were identified.View Large Image Figure ViewerDownload Hi-res image Download (PPT)