Poor correlation between clinical impression, the small colonic polyp and their neoplastic risk

Background and Objectives: Significance of the small colonic polyp is unclear and its removal is frequently determined by the proceduralist's clinical impression. Our aims were to determine if clinical discernment is accurate, and the likelihood that lesions < 10 mm are histologically advanced. Method: We prospectively collected 1988 lesions from 854 subjects (2215 consecutive colonoscopies). Lesion size, location, patient age, sex and the colonoscopist's clinical impression was recorded. Results: Clinical assessment for neoplasia had a sensitivity of 87.4%, specificity of 65.0%, positive predictive value of 76.0% and negative predictive value of 80.2%, resulting in an accuracy of 73.4%. Factors predictive of correct clinical impression were polyp size, location in the rectum and being pedunculated, but not the patient's age, sex or the endoscopist's experience. Of the 1434 lesions <= 5 mm in size, 44.5% were neoplastic and 3.5% were histologically advanced. Of the 266 lesions 6-9 mm, 79.3% were neoplastic, 19.9% were histologically advanced, five demonstrated high-grade dysplasia and three were malignant. Only two patients with an adenocarcinoma or high-grade dysplasia in a polyp < 10 mm had a lesion >= 10 mm elsewhere in the colon. Of the 288 lesions >= 10 mm in size, 92.7% were neoplastic, 29.5% had a villous component, 6.9% demonstrated high-grade dysplasia and 29.2% were malignant. Factors predictive of neoplasia were patient age, polyp size and sessile nature of the lesion. Conclusion: Polyps < 10 mm had a significant risk of neoplasia and advanced histology and, in general, clinical impression correlated poorly with neoplasia. Removal of all lesions proximal to the rectum, regardless of size, should therefore be considered.