CP function: an alpha spending function based on conditional power.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Methods

Background:

• Group sequential designs are crucial for adaptive clinical trials.
• Alpha spending functions and stochastic curtailment are common but have limitations.
• Stochastic curtailment struggles with controlling Type I error rates, though conditional power (CP) is clinically intuitive.

Purpose of the Study:

• To develop a new spending function, the CP function, integrating benefits of alpha spending and stochastic curtailment.
• To enhance flexibility and control Type I error rates in group sequential designs.
• To offer a clinically understandable approach to adaptive trial design.

Main Methods:

• Developed a novel two-parameter CP function based on conditional power.
• Conducted simulation studies to evaluate the CP function's performance.
• Compared the CP function against established methods like Pocock, O'Brien-Fleming, and quadratic spending functions.

Main Results:

• The CP function demonstrates flexibility, comparable or superior power to traditional methods with a proper CP threshold (ρ0).
• The CP function effectively controls the overall Type I error rate.
• It overcomes the Type I error control limitations of standard stochastic curtailment.

Conclusions:

• The CP function is a promising tool for group sequential trial design, offering improved Type I error control and flexibility.
• It provides a statistically sound and clinically interpretable alternative to existing methods.
• Proper selection of the CP threshold (ρ0) is key to maximizing trial power.