The reference standard used in most reviews was colonoscopy {10,15,17,22}. Burch et al reported that one study was a diagnostic cohort study in which 3090 patients underwent colonoscopy and an unspecified immunochemical FOBT: for the detection of all neoplasms, sensitivity was 53% and specificity 99.6%; for the detection of CRC, sensitivity was 52.6% and specificity 87.2% {17}.

We found one systematic review reporting accuracy measures for colonoscopy among a total of five colorectal cancer screening methods {22}. It included 130 articles in total (of these 20 were relative to colonoscopy) and the reported mean ± standard deviation sensitivity of colonoscopy for cancer and for large polyps (≥10mm) was respectively 94.7 ± 4.6 % and 92.5± 6.2 % (compared to FOBTs 45.7 ± 26.5% and 18.5 ± 11.8%). Instead the overall specificity of colonoscopy for detecting CRC was 99.8 ± 0.2% (compared to FOBTs 87.6 ± 11.4%). Colonoscopy has the highest sensitivity and specificity of the selected screening methods.

In the remaining reviews the results were not clear. Nevertheless, one of them{15} reported results about one study also included in this review. Park 2010 et al{13} implemented a prospective study in a large population of average- risk people in which everyone underwent colonoscopy after having FIT and gFOBT: confirmed with better evidence the observations of others that FIT has a higher sensitivity for detecting advanced colorectal cancers than gFOBT, and has an acceptable specificity that significantly reduces the need for colonoscopic evaluation in the screened population. FIT results were compared directly with colonoscopy results: the positive rate of gFOBT in patients with adenomas did not differ from the patients with normal colonoscopies (8.0 % vs. 7.3 %); however, the positive rate of FIT at the 75 and 100 ng/ml thresholds was higher in patients with adenomas compared with that of patients with normal colonoscopies (16.9 % vs. 7.5 %, and 14.6 % vs. 7.3% ( P < 0.001 and P = 0.002), respectively.