Response-adaptive decision-theoretic trial design: operating characteristics and ethics.

Area of Science:

• Clinical Trials
• Biostatistics
• Medical Ethics

Background:

• Adaptive randomization enhances clinical trial statistical efficiency.
• A common belief is that adaptive randomization inherently improves ethical treatment of subjects.

Purpose of the Study:

• To develop Bayesian, decision-theoretic clinical trial designs using response-adaptive randomization.
• To contrast designs with and without a cost for poor subject outcomes in the loss function.

Main Methods:

• Bayesian decision-theoretic framework for clinical trial design.
• Response-adaptive randomization to optimize subject allocation.
• Incorporation of subject outcome costs into the loss function.

Main Results:

• Designs without subject outcome costs showed efficiency gains but ethically concerning allocations.
• Designs including subject outcome costs achieved a better balance between competing trial needs.
• Controlling Type I and II error rates negated benefits of incorporating competing needs into the loss function.

Conclusions:

• Adaptive randomization efficiency gains are limited to explicitly stated goals in the loss function.
• Ethical goals not in the loss function may be compromised, challenging the assumption that all adaptive trials are more ethical.
• Explicitly defining trial goals, including ethical considerations, within the decision-theoretic framework is essential.