Purpose: Quantitative fecal occult blood tests (FOBTs) allow to specify the positivity threshold. We compared colorectal cancer (CRC) screening strategies with quantitative FOBTs, varying the positivity threshold and adapting the screening interval accordingly (longer intervals for better sensitivity).
Method: We used a Markov state-transition model of CRC to calculate life-years and the lifetime number of screening-related tests (FOBTs and follow-up/surveillance colonoscopies; the number of both procedures were combined using their US cost ratio as weighting factor) for a cohort of US 50-year-olds to whom FOBT screening is offered. We compared 2 strategies: 1) FOBT with per-test specificity of 95% in combination with a screening interval of 1 year (FOBT95-1y) and 2) FOBT with per-test specificity of 80% in combination with a screening interval of 5 years (FOBT80-5y). We selected specificity and screening interval combinations such that both strategies had a similar chance that an individual would experience a false positive FOBT result during the screening program. Per-test sensitivities for FOBT95-1y and FOBT80-5y were assigned according to ROC curve projections (15% and 30% for small precursor lesions, 35% and 50% for large precursor lesions, and 70% and 90% for CRC, respectively). We assumed perfect adherence in the base-case analyses. In sensitivity analyses, we used recent US data on longitudinal adherence with FOBT screening in community practice that showed that of those who attended FOBT screening, 42%, 44% and 14%, respectively, received 1, 2-3, and 4 or more FOBTs during the 5-year study period.
Result: In the base-case analyses, FOBT95-1y saved 22% more life-years and the number of screening-related tests was 29% higher compared to FOBT80-5y. When using observed data on longitudinal adherence with FOBT, the effectiveness of FOBT95-1y decreased by one third compared with the base-case analyses. The FOBT80-5y strategy was now more effective (given the higher per-test sensitivity), saving 19% more life-years, while requiring 17% more screening-related tests compared with FOBT95-1y.
Conclusion: Taking into account that regular adherence with yearly FOBT screening is low, the potential benefit of this screening strategy is not realized in practice. Lowering the positivity threshold of quantitative FOBT (yielding a higher per-test sensitivity and a lower per-test specificity) in combination with an extended screening interval could be a pragmatic approach to optimize FOBT screening in view of real-life adherence patterns.