Randomization in clinical trials with small sample sizes using group sequential designs

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Statistical Inference

Background:

• Group sequential designs allow early trial stopping for efficacy or futility.
• Balanced sample sizes are often desired in group sequential trials, limiting randomization choices.
• The impact of randomization procedures on trial outcomes in group sequential designs is not fully understood.

Purpose of the Study:

• To investigate the influence of randomization procedures on type I error probability and power in small sample clinical trials using group sequential designs.
• To assess the reporting of randomization methods in published group sequential trials.

Main Methods:

• Literature review on randomization reporting in group sequential trials.
• Simulation studies evaluating type I error and power across different group sequential designs (Pocock, O'Brien-Fleming, Lan-DeMets, inverse normal combination tests) and randomization procedures.

Main Results:

• Many published group sequential trials lack detailed randomization reporting.
• Poor randomization implementation can inflate type I error rates.
• Certain combinations, like inverse normal tests with specific randomization, can reduce statistical power.
• Lan-DeMets approach is robust to allocation ratio deviations; inverse normal tests require caution with permuted block randomization.

Conclusions:

• A framework is proposed for selecting appropriate group sequential designs and randomization procedures for small trials.
• Improved reporting of randomization methods in group sequential trials is essential.
• The validity of some group sequential designs depends on correct randomization procedures, which are often poorly reported.