Prospective Analysis of Withdrawal Times and Polyp Detection Rates During Screening Colonoscopy and a Comparison to US Multi-Society Task Force Recommendations: A Community-Based Continuous Quality Improvement Study

Abstract

Background: A US Multi-Society Task Force on colon cancer suggests that the withdrawal time of a screening colonoscopy should be at least 6-10 minutes and expected to yield an adenoma detection rate of >25% in men and >15% in women. Aims: To assess the association between adenoma detection and colonoscopy withdrawal time as a surrogate measure of quality. Methods: A series of videotaped colonoscopies were reviewed in a group training session of 9 experienced gastroenterologists to develop a consensus of bowel prep quality measures, based upon a previously validated instrument. 506 screening colonoscopies were performed in an ambulatory endoscopy center. The withdrawal, cleaning, and total time was recorded by an assistant. Resected polyps were assessed for size, histology, and location. Results: Cecal intubation was achieved in 99.8% of exams. The mean patient age was 60.7+/−8.9 years, 47.9% were male, and 87.9% white. Among completed exams, a similar proportion of men and women had polyps (58.3% vs. 52.5%), though adenoma were more common in men (36.8% vs. 27.4%, p = 0.02) and tended to occur in older patients (p = 0.08). Isolated adenomas proximal to the splenic flexure were seen in 11.7% and isolated diminutive adenomas (<5 mm) were observed in 16%. Mean withdrawal time (excluding cleaning time) was 4.2 min (range: 0.8-9.8) with a mean total cleaning time of 2.6 min (range: 0-14.7). The proportion of patients with identified adenomas varied significantly by provider (p = 0.01) with a range of 10-50%. In univariate analysis, withdrawal time was significantly correlated with detection of polyps (p = 0.001) and with detection of adenomas (p = 0.04), and varied significantly by provider (p = 0.0001). The bowel prep quality was rated lower in those with adenomas (p = 0.02). In a multivariate logistic regression model accounting for provider clustering and including withdrawal time, bowel prep, body mass index, gender and age, only male gender (OR 1.52, 95% CI 1.02-2.26) and bowel prep (OR 1.08, 95% CI 1.00-1.17) were significantly associated with adenomas. Two patients required surgical resection, for a high grade flat dysplastic lesion in the cecum and a 4 cm tubulovillous adenoma in the ascending colon, respectively. Conclusions: A community-based gastroenterology practice did achieve the recommended adenoma detection goals, although there was significant variation across providers. Withdrawal time was associated with adenoma detection in univariate analysis, though it was not significant in a multivariate model. Further investigation is warranted to determine key predictors for high quality exams.