Demand for Colonoscopy in Colorectal Cancer Screening Using a Quantitative Fecal Immunochemical Test and Age/Sex-Specific Thresholds for Test Positivity
Cancer Epidemiology, Biomarkers and Prevention
Despite age and sex differences in fecal hemoglobin (f-Hb) concentrations, most fecal immunochemical test (FIT) screening programs use population-average cut-points for test positivity. The impact of age/sex-specific threshold on FIT accuracy and colonoscopy demand for colorectal cancer screening are unknown. Methods: Using data from 723,113 participants enrolled in a Taiwanese population-based colorectal cancer screening with single FIT between 2004 and 2009, sensitivity and specificity were
... timated for various f-Hb thresholds for test positivity. This included estimates based on a "universal" threshold, receiveroperating-characteristic curve-derived threshold, targeted sensitivity, targeted false-positive rate, and a colonoscopy-capacityadjusted method integrating colonoscopy workload with and without age/sex adjustments. Results: Optimal age/sex-specific thresholds were found to be equal to or lower than the universal 20 mg Hb/g threshold. For older males, a higher threshold (24 mg Hb/g) was identified using a 5% false-positive rate. Importantly, a nonlinear relationship was observed between sensitivity and colonoscopy workload with workload rising disproportionately to sensitivity at 16 mg Hb/g. At this "colonoscopy-capacity-adjusted" threshold, the test positivity (colonoscopy workload) was 4.67% and sensitivity was 79.5%, compared with a lower 4.0% workload and a lower 78.7% sensitivity using 20 mg Hb/g. When constrained on capacity, age/ sex-adjusted estimates were generally lower. However, optimizing age/-sex-adjusted thresholds increased colonoscopy demand across models by 17% or greater compared with a universal threshold. Conclusions: Age/sex-specific thresholds improve FIT accuracy with modest increases in colonoscopy demand. Impact: Colonoscopy-capacity-adjusted and age/sex-specific f-Hb thresholds may be useful in optimizing individual screening programs based on detection accuracy, population characteristics, and clinical capacity. Cancer Epidemiol Biomarkers Prev; 27(6); 704-9. Ó2018 AACR.