Propensity Score--Based Approaches for Imputation in Time-to-Event Noninferiority Trials

• Activity Number: 582
• Type: Topic Contributed
• Date/Time: Wednesday, August 7, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Susan Wang*+ and Carrie Li
• Companies: Boehringer Ingelheim Pharmaceutical Inc. and Boehringer Ingelheim Pharmaceutical Inc.
• Keywords: propensity score ; time to event ; superiority ; noninferiority ; ITT

Abstract

In time to event outcome clinical trials, patients are usually followed until the occurrence of a specific event or until study termination, even if patients discontinue their medications prematurely. The outcome events are then analyzed using the ITT approach including outcome events that have occurred after study drug discontinuation. While this may seem to be an acceptable approach for a superiority trial, it may not be appropriate for a non-inferiority trial, especially in the setting where patients who have been randomized to the test drug are put on the active control drug after their premature discontinuation of study medication. Therefore additional exploratory analyses are recommended. One easy and direct approach is to censor patients at the time of premature study drug discontinuation. This approach bears the assumption that the censoring is at random. In this presentation, we consider the propensity score based methods to impute time to events for patients who terminate the study drug prematurely as an exploratory analysis. We present results from simulation studies and results from applying this approach to a real clinical trial.