Weighted Approach for Estimating Effects in Principal Strata With Missing Data for a Categorical Post-Baseline Variable in Randomized Controlled Trials

This research was motivated by studying anti-drug antibody (ADA) formation and its potential impact on long-term benefit of a biologic treatment in a randomized controlled trial, in which ADA status was not only unobserved in the control arm but also in a subset of patients from the experimental treatment arm. Recent literature considers the principal stratum estimand strategy to estimate treatment effect in groups of patients defined by an intercurrent status, that is, in groups defined by a post-randomization variable only observed in one arm and potentially associated with the outcome. However, status information might be missing even for a nonnegligible number of patients in the experimental arm. For this setting, a novel weighted principal stratum approach, namely weighted imputation regression (WRI), is presented: Data from patients with missing intercurrent event status were re-weighted based on baseline covariates and additional longitudinal information. A theoretical justification of the WRI method is provided for different types of outcomes, and assumptions allowing for causal conclusions on treatment effect are specified and investigated. Simulations demonstrated that the WRI method yielded valid inference and was robust against certain violations of assumptions. The method was shown to perform well in a clinical study with ADA status as an intercurrent event.

Automated summary

This research introduces a novel weighted principal stratum approach called weighted imputation regression (WRI) to address missing status information in the experimental treatment arm. The method is theoretically justified for different types of outcomes and demonstrated to yield valid inference in simulations and a clinical study.