Forty-nine studies (11,551 participants, range 32 to 2,531) were included in the review. All studies assessed CT colonography and 25 studies (9,223 participants) assessed optical colonoscopy. The main methodological issues identified in the included studies were the awareness of radiologists, in most cases, of the clinical indication for CT colonography (knowledge that patients were symptomatic may have resulted in the expectation of a higher prevalence of disease) and variation in the potential for CT results to influence the reference standard.
Sensitivity of CT colonography for colorectal cancer was 96.1% (95% CI 93.8% to 97.7%; 49 studies). There was no evidence of heterogeneity (I2=0%).
Sensitivity of optical colonoscopy for colorectal cancer was 94.7% (95% CI 90.4% to 97.2%; 25 studies). There was moderate heterogeneity (I2=50%). No variable included in the meta-regression analysis (variables not specified) could account for the observed heterogeneity.
Sixteen of 414 colorectal cancers were missed at CT colonography. Ten of these were located in the rectosigmoid colon, including six in the rectum, and six were located proximal to the splenic flexure. No cancers were missed by CT colonography when both cathartic and tagging agents were used in the bowel preparation.