Comparison of dynamic block randomization and minimization in randomized trials: a simulation study.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Randomization Techniques

Background:

• Achieving covariate balance is crucial for precise treatment effect estimation in clinical research.
• Dynamic randomization methods aim to balance multiple baseline characteristics.
• Limited empirical data exists comparing these methods' balance and efficiency.

Purpose of the Study:

• To compare dynamic block randomization and minimization for covariate balance and statistical efficiency.
• To evaluate simple randomization as a reference method.
• To inform the choice of randomization strategy in clinical trials.

Main Methods:

• A simulation study utilizing data from a prior randomized controlled trial.
• Comparison of balance statistics across randomization methods.
• Assessment of hypothesis testing accuracy and power.

Main Results:

• Dynamic block randomization consistently yielded superior balance and higher statistical power.
• Minimization showed improved balance and power over simple randomization, but less so than dynamic block randomization.
• These findings held true even after post-adjustment of baseline covariates.

Conclusions:

• Dynamic block randomization outperformed minimization in achieving balance and maximizing efficiency.
• Differences among the three randomization strategies (dynamic block, minimization, simple) were modest.
• The choice of randomization method should consider statistical advantages alongside practical implementation factors like sample size and time constraints.