Randomized clinical trials in oncology stopped early for benefit (RCTSEB)

Abstract

6039 Background: Randomized clinical trials (RCT) that stop earlier than planned because of apparent benefit often receive great attention. However, how often these early results get confirmed by other subsequent meta-analysis/randomized studies, and how often results get included in clinical guidelines is not known. We conducted systematic review of oncologic RCTSEB to further determine if the results from RCTSEB are confirmed by the subsequent studies and/or included in guidelines. Methods: RCT of any intervention reported as having stopped early because of the results favoring the intervention were previously identified (JAMA.2005;294:2203).Trials with interventions in prevention and treatment of solid and hematologic malignancies were included for this review. Citation search was conducted by two reviewers to identify meta-analysis (MA), systematic reviews (SR) and RCTs that used similar trial design as the RCTSEB. National Comprehensive Cancer Guidelines were reviewed to identify the impact of the results on recommendations on clinical practice. Results: 33 trials were identified, and 5 were excluded because they did not fit the inclusion criteria. MA/SR were available for the assessment of interventions reported in 15/28 RCTSEB. Results of subsequent MA confirmed 12/15 findings, while 3 trials were contradicted. Of the remaining 13 trials, for 7 there was at least one subsequent RCT identified. 6/7 were confirmed as beneficial in the subsequent RCTs and 1 was contradicted. For 6/28 RCTSEB we did not find any subsequent MA/SR/RCTS. We also looked at the inclusion of interventions from RCTSEB in the clinical guidelines. 18/28 interventions from RCTSEB were mentioned in the formulation of clinical guidelines.16/18 were recommended as true positive recommendations. The other 2 interventions which were addressed in the guidelines were either not confirmed (n = 1) or incorrectly recommended (n = 1). Conclusions: In this set of oncology of RCTSEBs the findings were subsequently confirmed in 64% (18/28) cases. Guidelines developers used 2/28 (7%) of RCTSEB for which reliable confirmatory data were available. However, caution should be applied in the interpretation of our results because the same RCTSEB was also included in confirmatory MAs which possibly could have introduced bias. No significant financial relationships to disclose.