Background: Screening for colorectal cancer with sigmoidoscopy benefits from the fact that distal findings predict the risk of advanced proximal neoplasms (APNs). This study was aimed at comparing the existing strategies of postsigmoidoscopy referral to colonoscopy in terms of accuracy and resources needed.
Methods: Asymptomatic individuals aged 50-69 years were eligible for a randomized controlled trial designed to compare colonoscopy and fecal immunochemical test. Sigmoidoscopy yield was estimated from results obtained in the colonoscopy arm according to three sets of criteria of colonoscopy referral (from those proposed in the UK Flexible Sigmoidoscopy, Screening for COlon REctum [SCORE], and Norwegian Colorectal Cancer Prevention [NORCCAP] trials). Advanced neoplasm detection rate, sensitivity, specificity, and number of individuals needed to refer for colonoscopy to detect one APN were calculated. Logistic regression analysis was performed to identify distal findings associated with APN. All statistical tests were two-sided.
Results: APN was found in 255 of 5059 (5.0%) individuals. Fulfillment of UK (6.2%), SCORE (12.0%), and NORCCAP (17.9%) criteria varied statistically significantly (P < .001). The NORCCAP strategy obtained the highest sensitivity for APN detection (36.9%), and the UK approach reached the highest specificity (94.6%). The number of individuals needed to refer for colonoscopy to detect one APN was 6 (95% confidence interval [CI] = 4 to 7), 8 (95% CI = 6 to 9), and 10 (95% CI = 8 to 12) when the UK, SCORE, and NORCCAP criteria were used, respectively. The logistic regression analysis identified distal adenoma ≥10 mm (odds ratio = 3.77; 95% CI = 2.52 to 5.65) as the strongest independent predictor of APN.
Conclusions: Whereas the NORCCAP criteria achieved the highest sensitivity for APN detection, the UK recommendations benefited from the lowest number of individuals needed to refer for colonoscopy.