Cost-effectiveness analysis for determining optimal cut-off of immunochemical faecal occult blood test for population-based colorectal cancer screening (KCIS 16)

Objectives We aimed to determine the optimal cut-off of the immunochemical faecal occult blood test (iFOBT) by using cost-effectiveness analysis. Methods A total of 22,672 subjects aged 50 years or older were invited to have an uptake of iFOBT. We collected data from screen-detected cases for the cut-off above 100 ng/mL and obtained interval cancers from a nationwide cancer registry for a cut-off below 100ng/mL. We found a total of 65 colorectal cancer (CRC) cases, including 43 detected by screen and 22 diagnosed between screens (interval cases). The optimal cut-off was first determined by receiver operating characteristics (ROC) curve analysis. Formal economic evaluation was further applied to identifying the optimal cut-off by assessing the minimum incremental cost-effectiveness ratio (ICER), an indicator for cost per life year gained (effectiveness), given a series of cut-offs of iFOBT, ranging from 30 to 200 ng/mL compared with no screening. Results ROC curve analysis found the optimal cut-off of iFOBT to be 100ng/mL at which the sensitivity, false-positive and odds of being affected given a positive result were 81.5% (70.2%-89.2%), 5.7% (5.4%-6.0%) and 1.24 (1.19-1.32), respectively. The area under ROC curve was 0.87 (0.81-0.93). In economic appraisal, the screening programme irrespective of any cut-off dominated (less cost and more effectiveness) over the control group. The optimal cut-off (the lowest ICER) was 110 ng/mL at which an average of 0.054 life year was gained and that of 950 ($US) was saved. Conclusions We used cost-effectiveness to identify 110 ng/mL as the optimal cut-off of iFOBT in a Taiwanese population-based screening for CRC. Our model provides a useful approach for health policy-makers in designing population-based screening for CRC to determine the optimal cut-off of iFOBT when cost and effectiveness need to be taken into account.