Adaptive treatment allocation for comparative clinical studies with recurrent events data

Area of Science:

• Clinical Trials
• Biostatistics
• Medical Research

Background:

• Recurrent event data analysis is crucial for comparing treatment efficacies in long-term clinical studies.
• The negative binomial model is commonly used for recurrent event data, accounting for patient heterogeneity.
• Balanced treatment allocation (equal sample sizes) is standard but may be suboptimal if one treatment is superior.

Purpose of the Study:

• To develop a sequential response-adaptive treatment allocation procedure for clinical trials.
• To reduce the number of subjects assigned to less-effective treatments.
• To maintain comparable statistical power to balanced designs.

Main Methods:

• A sequential response-adaptive treatment allocation procedure was derived.
• The procedure is based on the doubly adaptive biased coin design.
• The method was illustrated by redesigning a clinical study.

Main Results:

• The proposed allocation schemes effectively reduce the number of subjects receiving inferior treatments.
• The statistical power of the proposed method is comparable to that of balanced designs.
• The redesign demonstrated practical advantages of the new procedure.

Conclusions:

• Sequential response-adaptive allocation offers an efficient alternative to balanced designs in clinical trials with recurrent events.
• This approach optimizes resource allocation by minimizing exposure to less effective treatments.
• The method provides a statistically sound way to improve clinical trial efficiency and patient outcomes.