Dynamic Borrowing From Historical Controls via the Synthetic Prior With Covariates in Randomized Clinical Trials.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Bayesian Inference

Background:

• Rheumatoid arthritis clinical trials often require large control groups, increasing costs and timelines.
• Borrowing information from historical trials can optimize new trial designs but requires robust statistical methods.
• Existing methods may not adequately balance trust in historical data with new trial specifics.

Purpose of the Study:

• To introduce a novel Bayesian method, synthetic prior with covariates (SPx), for efficient clinical trial design.
• To enable borrowing information from historical trial data to reduce the required control group size in new trials.
• To provide a flexible approach that dynamically adjusts reliance on historical data.

Main Methods:

• Developed SPx, a Bayesian method utilizing synthetic priors with covariates.
• Employed Bayesian model averaging to manage uncertainty in the relationship between historical and new trial data.
• Required only trial-level summary statistics, enhancing data accessibility.

Main Results:

• Simulations demonstrated SPx's ability to substantially reduce control group sizes.
• Application to a rheumatoid arthritis trial confirmed the method's practical utility.
• The method maintained desirable Frequentist properties, ensuring statistical validity.

Conclusions:

• SPx offers a powerful tool for optimizing clinical trial design by reducing control group requirements.
• The method's reliance on summary statistics makes it broadly applicable.
• SPx facilitates more efficient and cost-effective clinical research, particularly in areas like rheumatoid arthritis.