ABSTRACT
BACKGROUND & AIMS: Capsule colonoscopy is a minimally invasive imaging method. We measured the accuracy of this technology in detecting polyps 6 mm or larger in an average-risk screening population. METHODS: In a prospective study, asymptomatic subjects (n = 884) underwent capsule colonoscopy followed by conventional colonoscopy (the reference) several weeks later, with an endoscopist blinded to capsule results, at 10 centers in the United States and 6 centers in Israel from June 2011 through April 2012. An unblinded colonoscopy was performed on subjects found to have lesions 6 mm or larger by capsule but not conventional colonoscopy. RESULTS: Among the 884 subjects enrolled, 695 (79%) were included in the analysis of capsule performance for all polyps. There were 77 exclusions (9%) for inadequate cleansing and whole-colon capsule transit time fewer than 40 minutes, 45 exclusions (5%) before capsule ingestion, 15 exclusions (2%) after ingestion and before colonoscopy, and 15 exclusions (2%) for site termination. Capsule colonoscopy identified subjects with 1 or more polyps 6 mm or larger with 81% sensitivity (95% confidence interval [CI], 77%-84%) and 93% specificity (95% CI, 91%-95%), and polyps 10 mm or larger with 80% sensitivity (95% CI, 74%-86%) and 97% specificity (95% CI, 96%-98%). Capsule colonoscopy identified subjects with 1 or more conventional adenomas 6 mm or larger with 88% sensitivity (95% CI, 82%-93) and 82% specificity (95% CI, 80%-83%), and 10 mm or larger with 92% sensitivity (95% CI, 82%-97%) and 95% specificity (95% CI, 94%-95%). Sessile serrated polyps and hyperplastic polyps accounted for 26% and 37%, respectively, of false-negative findings from capsule analyses. CONCLUSIONS: In an average-risk screening population, technically adequate capsule colonoscopy identified individuals with 1 or more conventional adenomas 6 mm or larger with 88% sensitivity and 82% specificity. Capsule performance seems adequate for patients who cannot undergo colonoscopy or who had incomplete colonoscopies. Additional studies are needed to improve capsule detection of serrated lesions. Clinicaltrials.gov number: NCT01372878.