Estimating Causal Effects for Adherers in a Randomized Trial Using Multiple Imputation

• Activity Number: 488 - Estimands and Imputation Methods
• Type: Contributed
• Date/Time: Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Junxiang Luo* and Yongming Qu and Steve Ruberg
• Companies: Sanofi and Eli Lilly and Company and Analytix Thinking, LLC
• Keywords: Counterfactual effect; adherence causal estimator; multiple imputation; tripartite estimands; confidence intervals

Abstract

Randomized controlled trials are a gold standard to evaluate the treatment effect (estimand) for efficacy and safety. According to the recent ICH-E9 addendum (R1), intercurrent events need to be considered in defining estimand and principal stratum is one of the 5 strategies to handle ICEs. Qu, et al (2020, Statistics in Biopharmaceutical Research 12:1-18) proposes adherence causal estimators (ACEs) for estimating the treatment difference for those who adhere to one or both treatments based on the causal-inference framework and demonstrated the consistency of those estimators. However, no variance estimation formula is provided. In addition, it is difficult to evaluate the performance of the bootstrap confidence interval (CI) due to computational intensity in the complex estimation procedure. This research implements ACEs with the method of multiple imputation (MI). Simulation is conducted to evaluate the coverage probability of the CI constructed by combining inferences from MI (Barnard and Rubin, Biometrika 1999:948–955) and the bootstrap CI. The application of the proposed MI-based method for a real data example is provided.