(A-105) Transporting the effects of lung cancer screening from the National Lung Screening Trial (NLST) to a nationally representative US screening-eligible population

Background: The National Lung Screening Trial (NLST) demonstrated that annual low-dose computed tomography (LDCT) reduces lung cancer mortality. However, subsequent studies have indicated that the general US screen-eligible population differs in age, racial composition, and comorbidity burden, raising concerns about the net benefits of lung cancer screening in real-world clinical practice.

Objectives: Estimate benefits and harms of lung cancer screening with LDCT in a screening-eligible US population using NLST and survey data.

Methods: We used individual-level data from the NLST, including 53,454 individuals from August 2002 through April 2004. Participants were 55-74 years old, had a smoking history of at least 30 pack-years, and, if they had formerly smoked, had quit within the past 15 years.

We identified the target population using the 2022 Behavioral Risk Factor Surveillance System (BRFSS), a nationwide, phone-based survey that includes questions on demographics, health (and smoking) history, and use of LDCT. We applied the same age and smoking history eligibility criteria as in the NLST. We computed inverse odds weights (IOWs) to reweight the NLST to reflect the target population’s covariate distribution, incorporating BRFSS survey sampling weights.

We estimated the effects of LDCT screening for lung cancer and all-cause mortality using an IOW-weighted Poisson model. We also compared the false-positive rates of LDCT in the weighted and unweighted trial populations.

Results: Our analysis included 15,779 screening-eligible BRFSS respondents (weighted to 7,265,314) and all NLST participants for mortality estimation. For false-positive rate estimation, we included only NLST participants who underwent LDCT (n=26,722). Compared to the NLST, the screening-eligible BRFSS population was older, more racially and ethnically diverse, more likely to smoke currently, and more likely to have respiratory comorbidities.

We estimated a similar effect of LDCT on lung cancer mortality in the weighted (rate ratio (RR) (95% CI): 0.83 (0.72, 0.96)) as in the unweighted (0.80 (0.70-0.91)) trial population. The benefit of LDCT on all-cause mortality was also similar (RR (95% CI): 0.95 (0.89, 1.01) vs. 0.93 (0.89-0.98), respectively). Estimated false-positive rates in the LDCT arm were higher in the weighted versus unweighted trial population across all three screening exams (14.25%, 6.20%, and 5.73% vs. 12.87%, 5.35%, 5.05%, respectively).

Conclusions: We estimated similar benefits and slightly increased harm from LDCT in the weighted trial population compared to NLST. Our findings assume that screening adherence and diagnostic work-up were similar between trial and real-world practice; future work should leverage alternative methods to address these assumptions.