📝 SummarXiv

Abstract

Externally controlled trials (ECTs) compare outcomes between a single-arm trial and external controls drawn from sources such as historical trials, registries, or observational studies. In survival analysis, a major challenge arises when the time origin (index date) differs across groups, for example, when treatment initiation occurs after a delay in the single-arm trial but is undefined in the external controls. This misalignment can bias treatment effect estimates and distort causal interpretation. We propose a novel statistical method, Index Date Imputation (IDI), that imputes comparable index dates for external control patients using the estimated distribution of treatment initiation times from the single-arm cohort. To address population-level confounding, IDI is combined with propensity score weighting or matching, yielding balanced and temporally aligned cohorts for survival comparison. We detail diagnostics for covariate balance and truncation bias, and evaluate performance via extensive simulations. Applying IDI to a randomized oncology trial, we demonstrate that the method recovers the known treatment effect despite artificial index date misalignment. IDI provides a principled framework for time-to-event analyses in ECTs and is broadly applicable in oncology and rare disease settings.