319P - Selective registration of non-primary endpoints in randomized clinical trials in oncology: a comparison of endpoint reporting between clinical trial protocols and US national clinical trial registration

Abstract

Randomized clinical trials represent the cornerstone of evidence-based medicine. Despite the implementation of clinical trial registration, the completeness of such reporting has not been well studied. We analyzed oncology-based randomized clinical trials (RCTs) conducted in 2012 in the Journal of Clinical Oncology, New England Journal of Medicine, and The Lancet. The primary endpoints and non-primary endpoints from each clinical trial protocol (a required supplement for publication in these journals) and the corresponding listing on the US national clinical trial registry (clinicaltrials.gov) were collected. Secondary, exploratory, correlative, and translational end points were considered non-primary end points. Registry/protocol discrepancies were then quantitatively and qualitatively evaluated. Out of the 58 RCTs, primary endpoint registry/protocol discrepancies occurred in only 2 trials (3%). However, registry/protocol discrepancies in non-primary endpoints occurred in 46 trials (79%). Of these 46 trials, 39 (85%) had non-primary endpoints found in the protocol but not in the registry. Of these 39 trials, 10 discrepancies related to exploratory or translational endpoints missing from the registry, and 6 related to the endpoint of safety missing from the registry. Of the 46 trials with non-primary endpoint registry/protocol discrepancies, 20 (43%) had non-primary endpoints found in the registry that were not found in the protocol. 18 of these 20 trials had de novo endpoints not listed anywhere in the protocol and 2 had endpoints listed in the protocol but not within the respective objective/endpoint section. Registry/protocol discrepancies regarding time-to-event endpoints occurred in 12 trials (21%), with 10 related to omission of such endpoints from the registry. The clinical trial registry is intended to serve as a transparent public database of clinical trials. However, according to our research, the majority of RCTs in high-impact journals demonstrate registry/protocol discrepancies in the listing of non-primary endpoints.