Keywords: estimands, intercurrent event strategies, death, quality of life, composite endpoints
In clinical trials where a non-trivial number of deaths are expected to occur during the study evaluation period (e.g., in trials of heart disease or cancer), evaluation of treatment effect on continuous measures of efficacy or quality of life (QoL) requires special attention. The measures of interest are typically undefined for deceased subjects after death and dealing with them requires different considerations compared to data missing for other reasons. If the study treatment is expected to affect both the measure of interest and duration of survival, different estimands may be of interest to different decision-makers: these estimands may involve evaluating the continuous measure and the survival measure separately or as a combined endpoint. We will discuss three types of estimand strategies that can be deployed in this setting: a composite, principal stratification, or while-on-treatment strategy may be used. We will also discuss some possible estimators (analysis methods) for each type of estimand. For an estimand that uses a composite strategy, we will discuss in more detail the challenges and approaches when defining and analyzing the composite endpoints based on a continuous measure and survival.