All Times EDT
Keywords: Longitudinal trials, Missing data, Hypothetical estimands, Bayesian methods
When a primary objective is to assess treatment effect at the last visit time point in longitudinal trials, specifications are required on how to handle intercurrent events (ICE) such as discontinuation of study therapy or early dropout and on how to deal with missing data. To completely defining the estimands, specific assumptions are made for outcomes after ICEs not only on the missing data but also on the potential treatment condition. In this talk, we will first discuss the ambiguity points on defining estimands and estimators under those situations, and then explore a few hypothetical strategies that assess the theoretical or attributable efficacy effects. Common missing data assumption such as missing at random, control-based imputation, and return-to-baseline are considered. To account for the uncertainty of the potential outcomes after ICEs, we investigate Bayesian approaches to obtain corresponding point and interval estimates and conduct sensitivity analysis. The methods are illustrated with applications to clinical trials.