Evaluated Screening Strategies

In total, 363 screening strategies were evaluated, each considering different combinations of screening starting and stopping ages, risk-prediction model used to estimate age-specific lung cancer incidence risk, and risk threshold for screening eligibility (Box 1). Upper bounds on screening stopping ages were enforced, as otherwise-eligible individuals would continue screening at ages with limited life expectancy. Lower bounds on screening starting age were enforced because the risk-prediction models were developed in populations consisting of individuals older than 45 years and may be unsuitable for younger individuals (3,20). At each age, a person’s screening eligibility was determined (ie, whether the person’s estimated risk at that age exceeded the risk threshold). Screening eligibility was assumed to be free of misclassification error (ie, risk at each age was correctly estimated, and ineligible individuals were not screened). In total, 120 screening strategies were considered per risk-prediction model. In addition, three screening strategies were used to evaluate the USPSTF criteria at different stopping ages. Perfect screening adherence was assumed. For each strategy, the following outcomes were evaluated: lung cancer deaths averted, life-years gained, proportion of individuals ever screening eligible, computed tomography screens required, and overdiagnosis (both the absolute number of overdiagnosed cases and percentage of screen-detected cases that is overdiagnosed, ie, number of overdiagnosed casesnumber of screen-detected cases*100%. Screening outcomes were counted from ages 45–100 years (maximum age in all models). All outcomes were compared with no-screening results, and standardized to the number of individuals alive at age 45 years. Results were summarized as means across CISNET models, along with the lower and upper ranges across models (CISNET model range [CMR]). Two sensitivity analyses were performed. The first considered hypothetical perfect life expectancy assessments, excluding individuals from further screening when non–lung cancer death occurred within 5 years. The second considered a 1960 birth cohort, representing smoking patterns and life expectancies that are more contemporary.

Overview of evaluated screening strategies

LCDRAT = Lung Cancer Death Risk Assessment Tool; PLCO = Prostate Lung, Colorectal, and Ovarian Cancer Screening Trial, USPSTF: United States Preventive Services Task Force.