Calculating power for the Finkelstein and Schoenfeld test statistic for a composite endpoint with two components

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Methods

Background:

• The Finkelstein and Schoenfeld (FS) test is widely used for analyzing prioritized composite endpoints in clinical trials.
• Current power and sample size estimation for the FS test relies heavily on computationally burdensome simulation studies.
• The complexity of simulations increases with more composite endpoints and larger sample sizes.

Purpose of the Study:

• To develop an analytical solution for calculating power and sample size for the Finkelstein and Schoenfeld (FS) test.
• To provide a computationally efficient and practical alternative to simulation-based methods for prioritized composite endpoints.
• To address the computational burden associated with traditional simulation approaches.

Main Methods:

• Derivation of power formulas for two-component hierarchical composite endpoints based on assumed underlying population distributions.
• Development of an analytical solution for power and sample size calculations.
• Validation of the proposed formulas using Monte Carlo simulations.

Main Results:

• The proposed analytical formulas provide a computationally efficient method for power and sample size estimation for the FS test.
• Monte Carlo simulations confirm the consistency and desirable properties of the analytical approach.
• The method demonstrates robustness to mis-specified distributional assumptions.

Conclusions:

• The developed analytical solution offers a practical and efficient alternative to simulation for power and sample size estimation in FS tests.
• This method is applicable to commonly encountered two-component hierarchical composite endpoints.
• The approach was successfully demonstrated in the context of the Transthyretin Amyloidosis Cardiomyopathy Clinical Trial.