Estimating treatment effects on duration with disease: a principal stratification framework

Area of Science:

• Clinical Epidemiology
• Biostatistics
• Causal Inference

Background:

• Estimating average treatment effects is crucial in clinical research.
• Individual variations in treatment effects necessitate subgroup analysis.
• Principal strata offer a refined focus for causal inference.

Purpose of the Study:

• To develop a framework for causal inference in a principal stratum for duration outcomes.
• To estimate the average treatment effect within a defined principal stratum.
• To assess the impact of potential assumption violations using a sensitivity parameter.

Main Methods:

• Focuses on causal inference for duration outcomes within a principal stratum.
• Utilizes a multi-state model with pseudo-observations to handle censoring.
• Introduces a sensitivity parameter to assess the robustness of findings.

Main Results:

• The proposed method offers greater statistical power compared to conventional group comparisons.
• Demonstrates a framework for identifying and estimating average treatment effects in principal strata.
• Provides a method for sample size calculation in relevant studies.

Conclusions:

• The framework enables robust causal inference for duration outcomes in principal strata.
• The multi-state model approach enhances statistical power in clinical trial analysis.
• The methodology is applicable to cancer recurrence studies and other clinical settings.