Withdrawal time, ADR and risk of interval colorectal cancer.

Shaukat A, Rector TS, Church T, Lederle FA, Kim AS, Rank J, Allen JI. Withdrawal time, ADR and risk of interval colorectal cancer. Presented at: American College of Gastroenterology Annual Meeting; 2014 Oct 21; Philadelphia, PA.

Withdrawal times and adenoma detection rates are widely used quality indicators for screening colonoscopy. Higher adenoma detection rates lower the risk of interval cancer. More rapid withdrawal times are associated with undetected adenomas, but their relationship with interval cancer has not been reported. Methods: We analyzed records of 76,810 screening colonoscopies performed in 2004 to 2009 by 51 gastroenterologists practicing in Minneapolis and St. Paul, Minnesota. Each physician's average annual withdrawal times were calculated from negative screening colonoscopies and annual adenoma detection rates were the percentage of all screening colonoscopies that detected at least one pathologically confirmed adenoma. Practice records were linked to the state cancer registry, Minnesota Cancer Surveillance System, to identify interval cancers that were defined as incident colorectal adenocarcinomas that were diagnosed within 5 years following the screening exam. The relationship between the physician's average annual withdrawal times and adenoma detection rates was tested using longitudinal linear regression. Nonlinear and categorical relationships between withdrawal times, adenoma detection rates and other variables and the relative risk of interval cancer were modeled using Poisson regression. Results: The physicians' average annual withdrawal times and adenoma detection rates were 8.6 1.7 minutes and 26 9%, respectively. We identified 56 interval cancers during 249,261 person-years of follow-up, for an annual rate of 0.22/1000 person-years. Physicians' average annual withdrawal times were inversely associated with interval cancers (p < 0.0001) (figure 1). Compared to withdrawal times > 6 minutes, the adjusted incidence rate ratio (IRR) for withdrawal times of < 6 minutes was 2.3 (95% CI 1.5 to 3.4; p < 0.0001). Physicians' adenoma detection rate was not associated with risk of interval cancer (p = 0.40). However, longer average withdrawal times were associated with higher adenoma detection rates (2.5% per minute, 95% confidence interval (CI) 1.5% to 3.5%; p < 0.0001). Conclusions: The overall incidence of interval cancers in a large community practice was quite low. Shorter average annual withdrawal times during screening colonoscopies were independently associated with an increased risk of interval colorectal cancer. Withdrawal times may be a more sensitive indicator of the risk of interval cancer than the adenoma detection rate.