Exact permutational tests for group sequential clinical trials

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Computing

Background:

• Group sequential methods are crucial for interim analyses in clinical trials.
• Accurate stopping boundaries are essential for ethical and efficient trial conduct.
• Existing methods may lack reliability with limited early-stage data.

Purpose of the Study:

• To develop an efficient numerical algorithm for computing precise stopping boundaries in group sequential clinical trials.
• To provide a method that is reliable even with limited data in early trial stages.
• To derive stopping boundaries from the exact joint permutational distribution of linear rank statistics.

Main Methods:

• The algorithm computes the exact boundary generating function using the joint permutational distribution.
• Ranks are assigned after pooling all patients, with permutations within newly arrived blocks.
• The method accommodates continuous/categorical, censored/uncensored data, and adaptive randomization.

Main Results:

• The algorithm yields exact stopping boundaries, improving reliability in early trial phases.
• It is applicable for an arbitrary number of monitoring times, adaptable to study needs.
• Demonstrated utility through a group sequential analysis of an Eastern Cooperative Oncology Group study.

Conclusions:

• The developed algorithm offers an efficient and accurate approach to determining stopping boundaries for group sequential trials.
• This method enhances the reliability of interim analyses, particularly when data is scarce.
• The technique is versatile, supporting various data types and adaptive randomization strategies.