Interim analysis incorporating short- and long-term binary endpoints.

Area of Science:

• Clinical Trial Design
• Biostatistics
• Drug Development

Background:

• Multi-endpoint designs are increasingly used in drug development.
• Interim analyses can incorporate short-term data when long-term endpoints are not yet available.
• Futility stopping rules are crucial for efficient trial management.

Purpose of the Study:

• To compare statistical properties of different estimators in two-stage adaptive designs.
• To evaluate methods for futility stopping based on conditional power and P-values.
• To investigate the impact of combining short-term and long-term endpoints in adaptive trials.

Main Methods:

• Considered two-stage designs with binary endpoints and futility stopping based on conditional power.
• Compared three estimators: long-term endpoint only, short-term endpoint only, and combined data.
• Extended methods to adaptive designs with sample size reassessment using the inverse normal combination function and evaluated two futility stopping rules.

Main Results:

• Equivalent cut-off points were derived for fixed and observed effect conditional power, maintaining overall power.
• Stopping for futility generally decreases, rather than inflates, the type I error rate but results in power loss.
• Compared average sample size, futility stopping probability, and overall trial power across different scenarios and stopping rules.

Conclusions:

• Adaptive designs incorporating interim analyses and futility stopping can be optimized by judiciously combining short-term and long-term endpoint data.
• The choice of futility stopping rules and the weighting of combined data significantly influence trial characteristics like sample size and power.
• These methods provide a framework for more efficient and informative clinical trial conduct.