Answers are grounded in this paper and cite their sources.
Hypothetical strategy allows to define an estimand for the pure treatment effect subjected to the original randomized treatments on overall survival (OS) without impact from effective subsequent therapies. Novel statistical methods including Inverse-Probability-of-Censoring Weights analysis (IPCW) and Two-Stage estimation (TSE), have been proposed for handling initiation of subsequent treatment. However, OS can vary across different types of subsequent therapies, which is ignored by these methods. Motivated by the GADOLIN trial, we propose modified methods (m-IPCW, m-TSE) to account for two different types of subsequent therapy.RCT data were simulated with various scenarios. The occurrence of subsequent therapies was simulated to vary with individual patient’s characteristics and randomized treatment. The performance of different methods was evaluated using bias and root-mean-square deviation compared to the true log hazard ratio in OS without subsequent therapies.Across simulated scenarios, both modified and standard TSE and IPCW methods demonstrated superior performance to naive censoring approach. The m-TSE outperformed its standard counterpart, as confirmed by the application in the GADOLIN trial. Our results suggested that in the presence of different effective subsequent therapies, statistical methods differentiating subsequent therapy types might yield more valid estimators for investigational treatment effect on OS under the hypothetical strategy.
No takes yet. Share an insight, caveat, or question.
Huang et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: