Adaptive biased urn randomization in small strata when blinding is impossible

Area of Science:

• Biostatistics
• Clinical Trial Design

Background:

• Adaptive biased urn randomization provides balance in clinical trials, even with small strata.
• Maintaining unpredictability in treatment assignment is crucial to prevent selection bias when blinding is not feasible.

Purpose of the Study:

• To describe a more flexible adaptive biased urn model for clinical trial randomization.
• To offer investigators customizable assignment probabilities based on group imbalance and trial size.

Main Methods:

• The proposed urn model allows probabilities to adapt based on group size imbalance.
• The model incorporates tendencies towards complete randomization as group sizes increase.
• It enables setting a maximum allowable difference in group sizes.

Main Results:

• The flexible urn model guarantees good balance between group sizes, even in small strata.
• It allows for unpredictable treatment assignments, mitigating selection bias.
• The model can maintain group size differences below a specified maximum.

Conclusions:

• The enhanced adaptive biased urn model offers greater flexibility and control in clinical trial randomization.
• It accommodates various randomization strategies, including complete randomization and permutations.
• The model can be extended to promote unequal group sizes and incorporates aspects of randomized minimization.