A Comparison of the Results of Intent-to-Treat, per-Protocol, and g-Estimation in the Presence of Nonrandom Treatment Changes in a Time-to-Event Noninferiority Trial

• Activity Number: 501
• Type: Topic Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yutaka Matsuyama*+
• Companies: The University of Tokyo
• Keywords: g-estimation ; Inten-to-treat ; non-compliance ; non-inferiority trials ; per-protocol ; randomization-based analysis

Abstract

It has often been said that ITT analysis tends to be anti-conservative in demonstrating non-inferiority, suggesting that PP analysis may be preferable. We propose using randomization-based g-estimation analyses that more effectively preserve the integrity of randomization. Simulation studies were conducted to investigate the impacts of different types of treatment changes on the conservatism or anti-conservatism of analyses using ITT, PP, and g-estimation in a time-to-event outcome. ITT results were anti-conservative for all simulations. For outcome-dependent treatment change, PP analysis was either conservative or anti-conservative. G-estimation analysis maintained type I error near the nominal level even for outcome-dependent case. Randomization-based g-estimation analyses should be used to supplement the more conventional ITT and PP analyses, especially for non-inferiority trials.