Background: Colonoscopic screening for colorectal cancer has been suggested because sigmoidoscopy misses nearly half of persons with advanced proximal neoplasia.
Objective: To create a clinical index to stratify risk for advanced proximal neoplasia and to identify a subgroup with very low risk in which screening sigmoidoscopy alone might suffice.
Design: Cross-sectional study.
Setting: A company-based program of screening colonoscopy for colorectal cancer.
Patients: Consecutive persons 50 years of age or older undergoing first-time screening colonoscopy between September 1995 and June 2001.
Measurements: A clinical index with 3 variables was created from information on the first 1994 persons. Points were assigned to categories of age, sex, and distal findings. Risk for advanced proximal neoplasia (defined as an adenoma 1 cm or larger or one with villous histology, severe dysplasia, or cancer) was measured for each score. The index was tested on the next 1031 persons from the same screening program.
Results: Of 1994 persons, 67 (3.4%) had advanced proximal neoplasia. A low-risk subgroup comprising 37% of the cohort had scores of 0 or 1 and a risk of 0.68% (95% CI, 0.22% to 1.57%). Among the validation group of 1031 persons, risk for advanced proximal neoplasia in the low-risk subgroup (comprising 47% of the cohort) was 0.4% (upper confidence limit of 1.49%). Application of this index detected 92% of persons with advanced proximal neoplasms and, if applied following screening sigmoidoscopy, could reduce the need for colonoscopy by 40%. The marginal benefit of colonoscopy among low-risk persons was small: To detect 7 additional persons with advanced proximal neoplasia, 1217 additional colonoscopies would be required.
Conclusions: This clinical index stratifies the risk for advanced proximal neoplasia and identifies a subgroup at very low risk. If it is validated in other cohorts or groups, the index could be used to tailor endoscopic screening for colorectal cancer.