Causal Survival Analysis in Platform Trials with Non-Concurrent Controls

2026-03-11

AI Summary

This study addresses the lack of a clear causal objective and identification conditions for leveraging non-concurrent controls (NCCs) to improve statistical efficiency in platform trials, a challenge particularly acute in survival analysis. The authors propose an estimator-oriented causal survival analysis framework that focuses on treatment-specific counterfactual survival curves—and their functionals such as restricted mean survival time—for the concurrent population. They establish, for the first time, causal identifiability conditions for NCC use, formalize the requisite pooling assumptions, and systematically evaluate the bias–efficiency trade-offs between covariate-adjusted outcome regression (OR) and doubly robust (DR) estimators. Theoretical and empirical analyses based on ACTT data show that OR benefits from pooling only when the pooling assumption holds and models are correctly specified, whereas DR may offer no efficiency gain in certain settings. The study recommends prioritizing covariate-adjusted DR estimation using concurrent controls alone.

Technology Category

• Reasoning under Uncertainty: Causality
• Search and Optimization: Non-convex Optimization
• Machine Learning: Causal Learning

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems
• Responsible Web: Consent frameworks and practices on the web

Abstract

Platform trials allow treatment arms to enter and exit over time while maintaining a shared control arm, yielding concurrent and non-concurrent controls (NCC). Pooling NCC is often motivated as a strategy to improve statistical efficiency, but it is unclear which estimand is targeted, what assumptions justify identification and estimation, and when precision gains are achievable; these questions are further complicated by time-to-event/survival data. Motivated by the Adaptive COVID-19 Treatment Trial (ACTT) platform trial with time to recovery as the primary endpoint, we develop an estimand-first causal survival framework targeting the treatment-specific counterfactual survival curve in the concurrent population and the corresponding functionals including the concurrent restricted mean survival time (RMST). We give nonparametric identification results and formalize conditions that justify pooling using NCC. We study covariate-adjusted outcome-regression (OR) and doubly robust (DR) estimators for the concurrent RMST, comparing concurrent-only versions to pooled-control versions. Pooling improves precision for OR estimators only when the pooling assumption holds and parametric hazard models are correctly specified; otherwise, pooling can induce bias. Moreover, in certain settings, pooling NCC yields no efficiency gain for the DR estimator. Overall, the most robust route to improve precision is to target concurrent causal survival estimands and use a covariate-adjusted DR estimation that uses only concurrent controls. An ACTT application corroborates these results.

Problem

• causal survival analysis
• platform trials
• non-concurrent controls
• restricted mean survival time
• time-to-event data

Innovation

• causal survival analysis
• platform trials
• non-concurrent controls
• doubly robust estimation
• restricted mean survival time