A proper statistical inference framework to compare clinical trial and real-world progression-free survival data.

The past decade has witnessed an increasing trend in utilizing external control data in clinical trials, especially in the form of synthetic control arms (SCA) derived from real-world or historical trial data. Including such data in clinical trial analysis can improve trial feasibility and efficiency, provided the issues caused by non-randomization and systematic differences are appropriately addressed. Current methodology development in this area focuses on establishing the comparability of patient baseline characteristics between arms, and more research is needed to ensure comparability of other elements such as endpoints. Motivated by the comparative analysis of SCA progression-free survival (PFS) and trial arm PFS, we aim to address another important but little discussed issue for external time-to-event (TTE) data that depend on disease assessment schedules (DAS). Since DAS are generally inconsistent across different data sources, we propose a proper statistical inference framework that harmonizes the DAS through data augmentation by multiple imputation. We demonstrate through extensive simulations that the proposed framework is unbiased in estimating median TTE and hazard ratio, well controls the type I error and achieves desirable power for log-rank test, while the unadjusted analysis can be biased and suffer from severe type I error inflation or power loss depending on the direction of the bias. Given the desirable performance, we recommend the proposed framework for comparative analysis using external DAS-based TTE data in clinical trials.