Abstract / Summary
Abstract Purpose: Patient-reported outcomes (PROs) are central to evaluating treatment benefit-risk in oncology randomised clinical trials (RCTs). However, high mortality rates impact PRO data analysis and interpretation. Current methodologies often inadequately account for death as an intercurrent event (ICE), leading to potential biases. This manuscript explores strategies for addressing death using the estimand framework to ensure clarity and consistency in PRO analyses. Methods: A motivating example is used to provide examples of potential strategies for addressing death as an ICE when estimating treatment effects in the context of a RCT, and a case-study example illustrates details of three strategies for addressing death as an ICE: hypothetical, while alive, and composite. Results: The different strategies can produce markedly different treatment effects, particularly as the proportion of deaths increases. The case study emphasized the importance of defining the timeframe of interest for the analysis when selecting the appropriate method to account for hypothetical timepoints after death. Conclusion: Robust PRO analysis requires clear research objectives, pre-specified estimands, and transparent incorporation of ICEs like death. SISAQOL-IMI acknowledges that there is no single optimal strategy for addressing death in PRO analyses in RCTs and recommends justifying the chosen strategy, which should match the predefined PRO objective of the clinical trial. Clear estimands, aligned statistical analyses, and transparent presentation of death events over time are essential to enhance credibility and impact of PRO data in oncology trials.