Estimation of Treatment Effect in Clinical Trials of Continuous Endpoints With Retrieved Dropouts

The estimand framework provides guidance on handling intercurrent events in the analysis of clinical trial data. In this paper, we focus on treatment discontinuation as the intercurrent event of interest. Under ICH E9(R1), the treatment policy (TP) strategy evaluates clinical outcomes regardless of treatment discontinuation and therefore generally requires appropriate consideration of observed post-discontinuation data. Existing approaches include likelihood-based and imputation-based methods for incorporating retrieved-dropout (RD) data. We propose a likelihood-based model for continuous endpoints that jointly represents outcomes among completers, RDs, and subjects lost to follow-up within a unified parametric framework. The approach combines an analysis-of-covariance formulation with a probit model for treatment discontinuation, enabling treatment effects to be formulated explicitly for estimands that apply either the hypothetical or TP strategy to treatment discontinuation. Estimation is carried out via a computationally efficient maximum likelihood procedure. Numerical studies assess the performance of the proposed method relative to several alternative approaches, including settings in which the amount of observed RD data is limited.