Estimation of treatment effects in presence of differential use of post-randomization concomitant medication with time-to-event outcomes

2026-05-07

AI Summary

This study addresses a critical bias in randomized controlled trials where more frequent use of cardioprotective concomitant medications in the placebo group leads to underestimation of the true treatment effect in intention-to-treat analyses. To resolve this, the authors propose a novel estimator within a causal inference framework that balances exposure to time-varying concomitant medications through a designed stochastic intervention. This approach mitigates violations of the positivity assumption and alleviates data sparsity by integrating targeted minimum loss-based estimation (TMLE) with flexible modeling of time-varying covariates, yielding efficient and robust estimates of the target causal effect. Applied to both simulation studies and the LEADER cardiovascular outcomes trial, the method successfully disentangles the effect of liraglutide from that of concomitant medications on cardiovascular events, substantially improving the accuracy of efficacy estimation.

Abstract

In placebo-controlled randomized trials, the post-randomization use of concomitant medications may be higher in the placebo arm than in the treatment arm. This may dilute the full benefits of the randomized drug as estimated by the intention-to-treat analysis. We focus on cardiovascular outcomes trials in type-2 diabetes patients of glucose-lowering treatments where patients in the placebo arm are more likely to add other glucose-lowering agents with established cardio-protective properties. As a supplement to the intention-to-treat analysis, we propose a class of estimands within a causal framework that isolates the specific impact of the treatment being studied from that of concomitant treatment use. These estimands are defined under time-dependent treatment interventions to balance exposure to additional medications across intervention arms. We advocate for specific stochastic interventions to achieve this balance while minimizing positivity violations, which arise when certain treatment combinations or characteristics are not sufficiently represented in the data. We employ targeted minimum loss-based estimation (TMLE) to optimize the estimation procedure for our estimands while allowing for flexible adjustments for time-dependent covariates from follow-up visits. Finally, we demonstrate the application of the methods through a simulation study and a real-world example from the LEADER cardiovascular outcomes trial, which assessed cardiovascular risk for liraglutide versus placebo.

Problem

• treatment effect estimation
• concomitant medication
• time-to-event outcomes
• randomized trials
• intention-to-treat analysis

Innovation

• causal inference
• time-dependent treatment
• stochastic intervention
• targeted minimum loss-based estimation
• positivity violation