Bayesian response adaptive randomization design with a composite endpoint of mortality and morbidity

Area of Science:

• Biostatistics
• Clinical Trial Design
• Medical Informatics

Background:

• Sequential adaptation of treatment allocation in clinical trials can optimize patient outcomes by minimizing failures or maximizing benefits.
• Organ support-free days (OSFD) is a composite endpoint combining mortality and morbidity, crucial for intensive care unit (ICU) patient assessment.
• Traditional methods often treat OSFD as an ordinal variable, potentially limiting the analysis of its complex nature.

Purpose of the Study:

• To develop and evaluate a Bayesian response-adaptive randomization (RAR) design for composite endpoints like OSFD in ICU patients.
• To improve upon existing adaptive randomization rules by better allocating patients to more effective treatment arms.
• To ensure robust statistical performance, including adequate power and type I error rate control.

Main Methods:

• Development of a novel Bayesian mixture model incorporating Markov chain Monte Carlo (MCMC) sampling.
• Estimation of the posterior probability distribution for OSFD to guide treatment allocation ratios at interim analyses.
• Simulation studies comparing the proposed RAR design against various randomization rules and alpha spending functions.

Main Results:

• The proposed Bayesian RAR design demonstrated superior patient allocation, directing more participants to better-performing treatment arms.
• The design effectively maintained adequate statistical power across diverse clinical scenarios.
• Type I error rate control was successfully assured, validating the design's reliability.

Conclusions:

• The novel Bayesian RAR design offers an effective strategy for optimizing treatment allocation in clinical trials with composite endpoints.
• This approach enhances patient benefit by adaptively assigning more participants to superior treatments.
• The methodology provides a robust framework for future clinical trial designs, particularly in critical care settings.