Selection of individuals for lung cancer screening based on risk prediction model performance and economic factors – The Ontario experience. (June 2021)

• Record Type: Journal Article
• Authors: Tammemägi, Martin C.; Darling, Gail E.; Schmidt, Heidi; Llovet, Diego; Buchanan, Daniel N.; Leung, Yvonne; Miller, Beth; Rabeneck, Linda
• Is Part Of: Lung cancer. Volume 156(2021)
• Page Start: 31
• Page End: 40
• Publication Date: 2021-06
• Subjects: CMS Centers for Medicaid and Medicare Services -- IELCAP International Early Lung Cancer Action Program -- LDCT Low-Dose Computed Tomography -- Lung-RADS Lung CT Screening Reporting & Data System -- NELSON NEderlands-Leuvens Longkanker Screening ONderzoek -- NLST National Lung Screening Trial -- OCR Ontario Cancer Registry -- PCP Primary Care Provider -- PLCO Prostate Lung Colorectal and Ovarian Cancer Screening Trial -- USPSTF The United States Preventive Services Task Force; Lung screening -- Risk model -- Eligibility -- Selection -- PLCOm2012; Lungs -- Cancer -- Periodicals; Lung Neoplasms -- Abstracts; Lung Neoplasms -- Periodicals; Poumons -- Cancer -- Périodiques; Lungs -- Cancer; Periodicals; Electronic journals; Electronic journals; 616.99424
• DOI: 10.1016/j.lungcan.2021.04.005
• Languages: English
• ISSNs: 0169-5002

Abstract

Highlights: Lung cancer screening eligibility determined by Health Technology Assessment (HTA) and MISCAN microsimulation model vs. PLCOm2012 model. PLCOm2012 had significantly higher sensitive and PPV compared to the preferred cost-effective MISCAN model. Expert panel choose the PLCOm2012noRace model for deciding eligibility into the Ontario Health - Cancer Care Ontario lung screening Pilot. Navigators PLCOm2012 risk assessed efficiently. Doctors preferred it. Participants rated it highly. Navigators reported no problems. Pilot PLCOm2012 led to significantly higher cancer detection than in NLST and NELSON, with significant shift to early stage. Abstract: Introduction: Randomized controlled trials have shown that screening with computed tomography reduces lung cancer mortality but is most effective when applied to high-risk individuals. Accurate lung cancer risk prediction models effectively select individuals for screening. Few pilots or programs have implemented risk models for enrolling individuals for screening in real-world, population-based settings. This report describes implementation of the PLCOm2012 risk prediction model in the Ontario Health (Cancer Care Ontario) lung cancer screening Pilot. Methods: In the Pilot's Health Technology Assessment, 576 categorical age/pack-years/quit-years scenarios were evaluated using MISCAN microsimulation modeling and cost-effectiveness analyses. A preferred model was selected which provided the most life-years gained per cost. The PLCOm2012 was compared to the preferred MISCAN scenario at a threshold that yielded the same number eligible (risk ≥2.0 %/6-years). Results: The PLCOm2012 had significantly higher sensitivity and predictive value (68.1 % vs 59.6 %, p < 0.0001; 4.90 % vs 4.29 %, p = 0.044), and an Expert Panel selected it for use in the Pilot. The Pilot cancer detection rate was significantly higher than in the NLST (p = 0.009) or NELSON (p = 0.003) and there was a significant shift to early stage compared to historical Ontario Cancer Registry statistics (p < 0.0001). Pre- and post-Pilot evaluations found that conducting quality risk assessments were not excessively time consuming or difficult, and participants' satisfaction was high. Conclusions: The PLCOm2012 was efficiently implemented in the Pilot in a real-world setting and is being used to transition into a provincial program. Compared to categorical age/pack-years/quit-years criteria, risk assessment using the PLCOm2012 can lead to effective and efficient screening.