Two-stage adaptive enrichment design for testing an active factor.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Inference

Background:

• Identifying active factors requires methods sensitive to effects varying across subpopulations.
• Traditional designs may lack efficiency when treatment effects are heterogeneous.
• Adaptive strategies can improve power by focusing resources on promising subgroups.

Purpose of the Study:

• To propose and evaluate an adaptive enrichment approach for testing active factors.
• To enhance the efficiency and power of clinical trials with heterogeneous treatment effects.
• To develop a robust statistical test applicable to different outcome types.

Main Methods:

• A two-stage play-the-winner design is implemented.
• Subjects in the second stage are selected from subpopulations with the highest observed effect in the first stage.
• A weighted Fisher's combination of Hotelling's test (stage 1) and noncentral chi-square test (stage 2) is recommended.
• The method is extended for binary and time-to-event outcomes.

Main Results:

• The proposed adaptive enrichment approach increases statistical power for detecting active factors.
• The two-stage design efficiently allocates resources to subpopulations with the largest effects.
• The recommended weighted combination test provides a robust framework for analysis.
• The extension to binary and time-to-event data broadens the applicability of the method.

Conclusions:

• Adaptive enrichment strategies are effective for testing active factors in heterogeneous populations.
• The proposed two-stage play-the-winner design offers an efficient and powerful approach.
• The recommended statistical tests and extensions provide a versatile tool for clinical research.
• This method can optimize resource allocation and improve the detection of treatment effects.