Enrique Marino, MS, Michelle Korslin, BS, Victoria Raymond, MS, Kathryn Lang, MBBS; Guardant Health, Redwood City, CA
Introduction: In a cancer screening program, the number of people needed to screen (NNTS), and more importantly the number needed to invite to screening (NNTI), provides an assessment of the screening intervention’s clinical impact. While NNTS is heavily influenced by the test performance (e.g. sensitivity, specificity), NNTI incorporates an additional patient compliance factor. We developed a model that assessed the impact of compliance in a colorectal cancer (CRC) screening program using varying test profiles and compliance estimates. Methods: We developed a Markov model to simulate the NNTI required to detect one CRC. Two test profiles, 90% sensitivity/85% specificity for CRC (Test A) and 75% sensitivity/85% specificity for CRC (Test B) were iterated with 90%, 75%, and 50% compliance rates. CRC incidence and prevalence were estimated from SEER data. Modelling approaches were programmed to begin at age 45 years. In the simulation, each patient is invited to screening every 5 years and censored at CRC diagnosis. We modelled two scenarios for each test profile. First, each subject is offered only the first test; then, to better model true clinical decision making, each subject is immediately offered a second test (with an equal or greater compliance rate) if non-compliant with the first test. Results: For each scenario we modelled 1,000 simulations of a 30,000-subject cohort with 0.72% CRC incidence. In the single test offering (Table 1), a screening test with 90% sensitivity and 50% compliance would need to invite 218 subjects to screening in order to detect one CRC. A test with 75% sensitivity and 90% compliance would need to invite 184 subjects. In sequential test offering (Table 2), Test A with 50% compliance followed by Test B with 90% compliance would invite 272 subjects to screening. Offering sequential testing with two 90% compliance modalities (Test B then Test A) would result in 177 invitations. Discussion: When evaluating cancer screening technologies, often the focus is on test performance metrics. However, a screening program’s success is heavily dependent on patient uptake of the screening invitation. We mathematically demonstrate a method incorporating patient compliance with test performance metrics, NNTI. We show that compliance is of equal, if not greater importance to performance metrics alone, emphasizing the need to maximize compliance at the first testing opportunity to ensure patients are up to date with CRC screening.
Disclosures: Enrique Marino: Guardant Health – Employee, Stockholder/Ownership Interest (excluding diversified mutual funds). Michelle Korslin: Guardant Health – Employee, Stockholder/Ownership Interest (excluding diversified mutual funds). Victoria Raymond: Guardant Health – Employee, Stockholder/Ownership Interest (excluding diversified mutual funds). Kathryn Lang: Guardant Health – Employee, Stockholder/Ownership Interest (excluding diversified mutual funds).