Design of Bayesian Clinical Trials With Clustered Data.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Bayesian Methodology

Background:

• Assessing operating characteristics (power, type I error) is crucial for clinical trial design.
• Current Monte Carlo simulation methods are computationally intensive, especially for clustered data and complex models.
• Efficient evaluation is needed for Bayesian trials with clustered data.

Purpose of the Study:

• To propose an efficient method for assessing operating characteristics and determining sample sizes in Bayesian trials with clustered data.
• To leverage theoretical results for faster computation and improved sample size recommendations.
• To demonstrate the methodology's applicability in a Bayesian cluster-randomized trial.

Main Methods:

• Developed theoretical results modeling posterior probabilities as a function of the number of clusters.
• Utilized these functions to assess operating characteristics across various sample sizes using limited simulations.
• Quantified the impact of simulation variability on sample size recommendations.

Main Results:

• The proposed method significantly reduces computational burden compared to traditional Monte Carlo simulations.
• Theoretical functions enable accurate assessment of operating characteristics with fewer simulations.
• The approach provides robust sample size recommendations, accounting for simulation variability.

Conclusions:

• The novel methodology offers an efficient and theoretically grounded approach for Bayesian clinical trials with clustered data.
• This method enhances the practical assessment of operating characteristics and sample size determination.
• The findings are applicable to complex trial designs, including Bayesian cluster-randomized trials.