Comparison of the spending function method and the Christmas tree correction for group sequential trials

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Methods

Background:

• Sequential designs allow for continuous monitoring of clinical trials.
• Real-world application often involves discrete analyses, deviating from continuous Brownian motion models.
• Corrections are needed to maintain statistical integrity under discrete monitoring.

Purpose of the Study:

• To compare two proposed methods for correcting discrete monitoring in sequential trial designs.
• To evaluate the performance of these correction methods in maintaining desired error rates.
• To assess accuracy for O'Brien and Fleming and triangular test procedures.

Main Methods:

• The study adapted two correction methods for discrete monitoring.
• Procedures analogous to the O'Brien and Fleming and triangular tests were developed.
• Error rates were compared between the correction methods and the original test procedures.

Main Results:

• For O'Brien and Fleming type tests, the spending function method demonstrated superior accuracy in achieving target error rates compared to the Christmas tree correction.
• For triangular tests, both the spending function method and the Christmas tree correction performed as intended, maintaining planned error rates.
• The study highlights differential performance of correction methods based on the specific sequential design used.

Conclusions:

• The spending function method is recommended for correcting discrete monitoring in O'Brien and Fleming sequential designs due to its accuracy.
• Both tested correction methods are suitable for triangular sequential designs.
• Accurate error rate control in discrete sequential monitoring is crucial for valid clinical trial outcomes.