Randomization by minimization for unbalanced treatment allocation.

Area of Science:

• Clinical Trials
• Biostatistics
• Medical Research Methodology

Background:

• Minimization is a dynamic randomization technique used in clinical trials to balance prognostic factors.
• Traditional minimization is primarily applied in settings with equal treatment allocations.
• A gap exists in published minimization procedures for clinical trials with unequal treatment allocations.

Purpose of the Study:

• To present novel strategies for applying minimization methodology in clinical trials with unequal treatment allocations.
• To propose and compare two new minimization techniques: naïve minimization and biased-coin minimization (BCM).

Main Methods:

• Two minimization techniques were developed: naïve minimization and biased-coin minimization (BCM).
• Probability calculations and simulation studies were used to compare the performance of the two methods.
• The performance was evaluated across various trial settings, including different numbers of treatments, prognostic factors, and sample sizes.

Main Results:

• Biased-coin minimization (BCM) demonstrated superior performance compared to naïve minimization.
• BCM effectively achieves an 'unbiased' target allocation ratio by incorporating allocation ratios into probability assignments.
• The choice of distance metrics slightly impacts minimization performance but can be optimized for specific trial features.

Conclusions:

• Biased-coin minimization (BCM) is the preferable method for randomization in clinical trials with unequal treatment allocations.
• The proposed BCM method offers an effective solution for balancing prognostic factors in complex trial designs.
• Further optimization of distance metrics can enhance minimization performance based on trial specifics.