Calibrated dynamic borrowing using capping priors.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Bayesian Inference

Background:

• Multisource exchangeability models (MEMs) offer a Bayesian approach to enhance randomized controlled trial (RCT) efficiency by integrating data from multiple sources.
• Integrating supplementary trial data, especially when significantly larger than the primary trial, requires careful management to prevent bias.

Purpose of the Study:

• To propose a novel Bayesian method, "capping priors," to control information borrowing in MEMs.
• To ensure that large supplementary datasets do not disproportionately influence primary trial results.

Main Methods:

• Development of the "capping priors" technique, which imposes an a priori limit on the effective sample size of supplementary data.
• Simulation studies to evaluate the performance and behavior of the capping priors method.
• Application of the method to real-world data from four clinical trials involving very low nicotine content cigarettes.

Main Results:

• The capping priors method effectively regulates the influence of supplementary data in MEMs.
• Demonstrated control over the degree of information borrowing, preventing the primary trial from being overwhelmed.
• Successful application in a practical setting, showcasing its utility in nicotine cigarette trials.

Conclusions:

• Capping priors provide a robust mechanism for dynamic data integration in clinical trials.
• This method enhances the efficiency and reliability of multisource exchangeability models, particularly when dealing with heterogeneous trial sizes.
• The approach is validated for use in complex clinical trial scenarios, such as those evaluating smoking cessation interventions.