Background: Previous estimates of the adenoma miss rate with optical colonoscopy (OC) are hindered by the use of OC as its own reference standard.
Objective: To evaluate the frequency and characteristics of colorectal neoplasms that are missed prospectively on OC by using virtual colonoscopy (VC) as a separate reference standard.
Design: Prospective, multicenter screening trial.
Setting: 3 medical centers.
Participants: 1233 asymptomatic adults who underwent same-day VC and OC.
Measurements: Colorectal neoplasms (adenomatous polyps) missed at OC before VC results were unblinded.
Results: Fourteen (93.3%) of 15 nonrectal neoplasms were located on a fold; 10 (71.4%) of these were located on the backside of a fold. Five (83.3%) of 6 rectal lesions were located within 10 cm of the anal verge.
Limitations: Estimation of the OC miss rate depended on polyp detection on both VC and second-look OC and therefore underestimates the true OC miss rate, particularly for smaller polyps.
Conclusions: Most clinically significant adenomas missed prospectively on OC are located behind a fold or near the anal verge. The 12% OC miss rate for large adenomas (>or=10 mm) when state-of-the-art 3-dimensional VC is used as a separate reference standard is increased from the previous 0% to 6% estimates derived by using OC as its own reference standard.