A Risk Profile for Advanced Proximal Neoplasms on Diagnostic Colonoscopy

Abstract

The capacity for colonoscopy is limited and a method to prioritize patients for diagnostic colonoscopy is needed in health care centers. A retrospective cross-sectional cohort study was carried out in county and community endoscopy centers, which included 1,065 county and 279 community patients aged > or = 40 years undergoing diagnostic colonoscopy. We constructed a risk profile for proximal advanced neoplasms on diagnostic colonoscopy at the county center based on the size of the regression coefficients for independent risk factors from logistic regression. An advanced neoplasm was defined as one of size > or = 1 cm or containing villous histology, high-grade dysplasia, or cancer. In our county colonoscopy population (n = 929 after exclusions), the stepwise logistic regression analysis identified age > or = 60 years (adjusted odds ratio [AOR]: 2.60; 95% confidence interval [CI]:1.14, 6.14), iron deficiency anemia (AOR: 4.74; 95% CI: 2.07, 11.34), and an advanced neoplasm in the recto-sigmoid (AOR: 6.01; 95% CI: 2.02, 16.00) as the statistically significant predictors of an advanced proximal neoplasm. In the county population, the prevalence rates of an advanced proximal neoplasm and proximal high-grade dysplasia/cancer in the low-risk group were 0.71% (95% CI: 0.15, 2.05) and 0.24% (95% CI: 0.01, 1.31), respectively. Avoiding colonoscopy in this group would increase the capacity for colonoscopy by 46% in the higher risk groups. In a disparate community population (n = 237 after exclusions), this scoring system had a goodness-of-fit test showing high concordance (P = 0.51). This clinical profile stratified the risk for an advanced neoplasm proximal to the sigmoid in patients undergoing diagnostic colonoscopy. It identified a large subset of low-risk patients.