A systematic review of noninferiority margins in oncology clinical trials.

A systematic literature review was conducted to identify and characterize noninferiority margins for relevant end points in oncology clinical trials. Randomized, controlled, noninferiority trials of patients with cancer were identified in PubMed and Embase. Of 2284 publications identified, 285 oncology noninferiority clinical trials were analyzed. The median noninferiority margin was a hazard ratio of 1.29 (mean: 1.32; range: 1.05-2.05) for studies that reported time-to-event end points (n = 192). The median noninferiority margin was 13.0% (mean: 12.7%; range: 5.0-20.0%) for studies that reported response end points as absolute rate differences (n = 31). Although there was consistency in the noninferiority margins' scale, variability was evident in noninferiority margins across trials. Increased transparency may improve consistency in noninferiority margin application in oncology clinical trials.