Extension of a conditional performance score for sample size recalculation rules to the setting of binary endpoints

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Statistical Modeling

Background:

• Sample size calculation in clinical trials relies on parameter assumptions, introducing uncertainty and risks of underpowering or oversizing.
• Adaptive designs offer a solution by allowing sample size adjustments during interim analyses.
• Existing methods for adaptive designs, like the conditional performance score, are primarily developed for normally distributed endpoints.

Purpose of the Study:

• To extend the conditional performance score methodology to clinical trials with binary endpoints.
• To evaluate the performance and applicability of this extended score for binary outcomes.

Main Methods:

• Developed a one-dimensional score parametrization for the conditional performance score tailored to binary endpoints.
• Conducted a simulation study to assess the operational characteristics of the proposed method.
• Illustrated the practical application of the extended score.

Main Results:

• The conditional performance score theory can be directly extended to binary endpoints without modification.
• The score results are represented by a single distribution parameter.
• A unified effect measure is derived, incorporating both the difference in proportions and the control group proportion.

Conclusions:

• The conditional performance score is successfully extended to binary endpoints, broadening its applicability in clinical trial design.
• This research provides a practical tool for guiding the selection of adaptive designs with sample size recalculation for binary outcomes.