Systematic review and bivariate/HSROC random-effect meta-analysis of immunochemical and guaiac-based fecal occult blood tests for colorectal cancer screening
Launois, R., Le Moine, J-G., Uzzan, B., Fiestas Navarrete, LI., & Benamouzig, R. (2014). Systematic review and bivariate/HSROC random-effect meta-analysis of immunochemical and guaiac-based fecal occult blood tests for colorectal cancer screening. European Journal of Gastroenterology & Hepatology, 26(9), 978-989. https://doi.org/10.1097/MEG.0000000000000160
Current literature evidences higher accuracy of immunological (iFOBT) vis-à-vis guaiac-based (gFOBT) fecal occult blood tests for colorectal cancer (CRC) screening. Few well-designed head-to-head comparisons exist.
This meta-analysis assesses the performances of two iFOBTs compared with an established gFOBT using colonoscopy as the gold standard.
We mobilized a bivariate and a hierarchical summary receiver operating characteristic (HSROC) model. Positive likelihood ratio (LR+) and negative likelihood ratio (LR?) and diagnostic odds ratios were back-calculated. We constructed bivariate credibility ellipses in the HSROC space and calculated areas under the curve to obtain a global measure of test performance. Estimates are presented at 95% credibility levels.
We included and analyzed 21 studies. OC-Sensor was the best test for CRC screening, with high sensitivity (0.87; 95% credibility interval: 0.73–0.95) and specificity (0.93; 95% credibility interval: 0.84–0.96), optimal LR+ (12.01) and LR? (0.14), and a high diagnostic odds ratio (88.05). Bivariate credibility ellipses showed OC-Sensor’s dominance over Hemoccult (sensitivity: 0.47; 95% credibility interval: 0.37–0.58; specificity: 0.93; 95% credibility interval: 0.91–0.95).
Our findings support the use of OC-Sensor for CRC detection. The diagnostic estimates obtained may be extended to derive model parameters for economic decision models and to offer insight for future clinical and public health decision making. Our findings could influence the future of FOBTs within the CRC screening arsenal.