Bayesian methods to determine performance differences and to quantify variability among centers in multi-center...

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Health Services Research

Background:

• Assessing variability in clinical trial outcomes across different centers is crucial for quality control.
• Identifying outlier centers can help improve patient care and trial efficiency.

Purpose of the Study:

• To quantify inter-center variability in primary outcomes using a novel statistical approach.
• To develop a guideline for identifying outlier centers in clinical trials.
• To apply these methods to the Intraoperative Hypothermia for Aneurysm Surgery Trial (IHAST).

Main Methods:

• Employed Bayesian hierarchical models to estimate center-specific effects on treatment outcomes.
• Utilized outlier detection methods assuming exchangeable center effects.
• Adjusted analyses for patient demographics, disease characteristics, and treatment variables.

Main Results:

• Center-to-center variation in favorable outcomes was consistent with a normal distribution (posterior sd = 0.538).
• No centers were identified as definitive outliers based on the proposed guideline.
• Center characteristics did not predict outcomes, but patient and disease factors did.

Conclusions:

• Bayesian hierarchical methods effectively assess center performance and identify predictors of outcome, even with small sample sizes.
• The IHAST demonstrated moderate variability between centers, but no centers were found to be outliers.
• These methods enhance the ability to monitor and improve clinical trial conduct across multiple sites.