Network and covariate adjusted response-adaptive design for binary response.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Health Informatics

Background:

• Randomization is crucial for unbiased clinical trial efficacy assessment.
• Existing methods face challenges in flexibility, efficiency, and managing network interference and covariates.
• Adaptive designs offer potential solutions for more dynamic trial management.

Purpose of the Study:

• To introduce the network and covariate adjusted response-adaptive (NCARA) design.
• To address challenges in maximizing patient benefit, balancing social networks, and ensuring covariate balance concurrently.
• To evaluate the performance of the NCARA design against alternative randomization methods.

Main Methods:

• The proposed NCARA design incorporates response-adaptive randomization with network and covariate adjustments.
• Simulations were conducted using diverse network structures and parameter settings.
• Real-world data from two clinical trials were analyzed to implement and validate the NCARA design.

Main Results:

• The NCARA design outperformed four alternative randomization designs in simulations.
• It demonstrated comparable statistical power and Type I error rates for treatment effect detection.
• Real data analysis showed increased power, higher assignment to superior treatments, and improved network/covariate balance compared to equal randomization.

Conclusions:

• The NCARA design offers a high-quality and efficient approach to clinical trial conduct.
• Its advantages are particularly notable in trials with small sample sizes and high network interference.
• The NCARA design effectively balances treatment allocation, network ties, and covariate distribution.