A simple, flexible, and effective covariate-adaptive treatment allocation procedure

Area of Science:

• Biostatistics
• Clinical Trial Design
• Health Services Research

Background:

• Sequential treatment allocation is crucial in clinical trials and observational studies.
• Existing methods may require outcome models or struggle with covariate balance.
• Propensity score methods are well-established for covariate adjustment in observational data.

Purpose of the Study:

• To present a new method for sequential treatment allocation based on propensity score theory.
• To demonstrate the method's ability to balance discrete and continuous covariates without outcome modeling.
• To evaluate the performance of the proposed method against existing sequential allocation techniques.

Main Methods:

• Developed a sequential allocation method leveraging propensity score principles.
• Explored specific deterministic and randomized versions of the proposed method.
• Compared the new method with established sequential randomization and allocation procedures.
• Utilized simulations to assess covariate balance, efficiency, and selection bias.

Main Results:

• The deterministic version achieved excellent covariate balance and near-optimal efficiency with minimal assumptions.
• Simulations confirmed the robustness of the method across various covariate distributions and numbers.
• Randomized versions exhibited favorable properties regarding optimality and selection bias.

Conclusions:

• The proposed propensity score-based sequential allocation method offers a robust and efficient approach.
• It effectively balances covariates without requiring outcome models, applicable to diverse study designs.
• The method shows promise for improving treatment allocation in sequential clinical and observational studies.