Inter-Physician Variation in Follow-Up Colonoscopies after Screening Colonoscopy

BACKGROUND AND AIMS: Surveillance is an integral part of the colorectal cancer (CRC) screening process. We aimed to investigate inter-physician variation in follow-up procedures after screening colonoscopy in an opportunistic CRC screening program. METHODS: A historical cohort study in the German statutory health insurance system was conducted. 55,301 individuals who underwent screening colonoscopy in 2006 in Bavaria, Germany, and who were not diagnosed with CRC were included. Utilization of follow-up colonoscopies performed by the same physician (328 physicians overall) within 3 years was ascertained. Mixed effects logistic regression modelling was used to assess the effect of physicians and other potential predictors (screening result, age group, and sex) on re-utilization of colonoscopy. Physicians were grouped into quintiles according to individual effects estimated in a preliminary model. Predicted probabilities of follow-up colonoscopy by screening result and physician group were calculated. RESULTS: The observed rate of follow-up colonoscopy was 6.2% (95% confidence interval: 5.9-6.4%), 18.6% (17.8-19.4%), and 37.0% (35.5-38.4%) after negative colonoscopy, low-risk adenoma and high-risk adenoma detection, respectively. All considered predictors were statistically significantly associated with follow-up colonoscopy. The predicted probabilities of follow-up colonoscopy ranged from 1.7% (1.4-2.0%) to 11.0% (10.2-11.7%), from 7.3% (6.2-8.5%) to 35.1% (32.6-37.7%), and from 17.9% (15.5-20.6%) to 56.9% (53.5-60.3%) in the 1(st) quintile (lowest rates of follow-up) and 5(th) quintile (highest rates of follow-up) of physicians after negative colonoscopy, low-risk adenoma and high-risk adenoma detection, respectively. CONCLUSIONS: This study suggests substantial inter-physician variation in follow-up habits after screening colonoscopy. Interventions, including organizational changes in CRC screening should be considered to reduce this variation.