Bayesian methods for assessing transplant program performance.

Area of Science:

• Biostatistics
• Medical Informatics
• Transplant Surgery

Background:

• Current transplant program-specific reports (PSRs) use conventional statistical methods.
• These methods can yield implausible estimates, wide confidence intervals, and underpowered tests.
• Statistical summaries are prone to misinterpretation.

Purpose of the Study:

• To adopt Bayesian methods for assessing transplant program performance.
• To improve the accuracy and interpretability of program-specific reports (PSRs).
• To replace current statistical methods with a more robust Bayesian approach.

Main Methods:

• Utilizing Bayesian statistical inference.
• Employing posterior probability distributions to reflect data and prior beliefs.
• Calculating program-specific hazard ratios to assess performance.

Main Results:

• Bayesian methods assume most programs perform near average, requiring strong evidence for extreme performance conclusions.
• Posterior distributions will indicate if a program is likely performing better or worse than average.
• New Bayesian-derived PSRs will be available for preview in mid-2014.

Conclusions:

• Bayesian methods offer a more reliable approach to transplant program performance assessment.
• The transition to Bayesian-derived PSRs is planned for early 2015.
• This shift aims to provide clearer and more accurate performance metrics for transplant programs.