Identification of physicians with unusual performance in screening colonoscopy databases: a Bayesian approach

Area of Science:

• Gastroenterology
• Public Health
• Biostatistics

Background:

• Adenoma detection rate (ADR) is a key quality indicator for screening colonoscopies.
• Identifying performance variations among physicians is crucial for quality assurance.

Purpose of the Study:

• To demonstrate a method for identifying physicians with unusual adenoma detection rates (ADR) in screening colonoscopy databases.
• To evaluate the effectiveness of a statistical approach in flagging performance outliers.

Main Methods:

• Bayesian random-effects modeling and Winsorization were used to create a robust model.
• Funnel plots were utilized for visualizing analysis steps and assessing divergence.
• Data from 422 physicians and 69,738 screening colonoscopies in Bavaria, Germany, were analyzed.

Main Results:

• The overall adenoma detection rate (ADR) was 26%.
• An initial model identified 62 physicians (15%) as potential outliers.
• A refined model confirmed 10 physicians (16% of outliers) with significantly lower than expected ADR at a 5% false discovery rate.

Conclusions:

• The statistical approach effectively identifies unusual performance in screening colonoscopy data.
• This method can aid in evaluating and enhancing the quality of colonoscopies within population-based colorectal cancer screening programs.