Enhanced doubly robust estimation in clinical trials accounting for intercurrent events

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Causal Inference

Background:

• The ICH E9 (R1) estimand addendum highlights the need for robust methods to handle intercurrent events (ICE) in clinical trials.
• Causal inference is critical for mitigating bias, especially with imbalanced covariates or imperfect randomization.

Purpose of the Study:

• To develop a time-varying covariate model for asymptotically unbiased treatment effect estimation.
• To utilize enhanced doubly robust estimators (eDRE) for causal inference under the estimand framework, accounting for ICE.

Main Methods:

• Development of a novel enhanced doubly robust estimator (eDRE).
• Application of a time-varying covariate model within a causal estimation framework.
• Analysis of asymptotic properties of the proposed eDRE.

Main Results:

• The proposed eDRE provides an asymptotically unbiased estimate of the treatment effect.
• Simulation studies demonstrate the method's superior performance compared to existing parametric and naive estimates.
• The method was successfully applied to a real-world clinical trial dataset.

Conclusions:

• The enhanced doubly robust estimator offers a statistically sound approach for estimating treatment effects in the presence of intercurrent events.
• This method improves causal inference in clinical trials by adjusting for time-varying covariates and utilizing all randomized patients.