• Fellowship Director, Duke Cardiothoracic Radiology Fellowship; Research Director, Duke Lung Cancer Screening Program; Department of Radiology, Duke Health, Durham, North Carolina. Electronic address: tina.tailor@duke.edu.
• Center for Statistical Sciences, Brown University School of Public Health, Providence, Rhode Island; Department of Biostatistics, Brown University of Public Health, Providence, Rhode Island.
• Center for Statistical Sciences, Brown University School of Public Health, Providence, Rhode Island.
• Carver College of Medicine, University of Iowa, Iowa City.
• Department of Radiology, Atrium Health Wake Forest Baptist, Winston-Salem, North Carolina.
• Vice Chair of Academic Affairs and Community Engagement, Department of Radiology and Director, Research on Outcomes and Care Delivery, Center for Innovation in Imaging Biomarkers and Integrated Diagnostics, Columbia University, New York, New York; JACR Editor in Chief; CIBR, Chair.
• Center for Statistical Sciences, Brown University School of Public Health, Providence, Rhode Island; Department of Epidemiology, Brown University School of Public Health, Providence, Rhode Island.

PURPOSE: Although lung cancer screening (LCS) with low-dose chest CT (LDCT) is recommended for high-risk populations, little is known about how clinical screening compares with research trials. We compared Lung CT Screening Reporting and Data System (Lung-RADS) scores between a nationally screened population from the ACR's LCS Registry (LCSR) and the National Lung Screening Trial (NLST).

METHODS: This retrospective study included baseline LDCT examinations from the LCSR and NLST. Patient characteristics (age, gender, smoking status, pack-years, and body mass index) were obtained. NLST LDCT results were recoded to Lung-RADS version 1.1. A multivariable multinomial logistic model was used to examine variations in Lung-RADS scores by screening group (LCSR versus NLST) and patient characteristics.

RESULTS: In all, 686,011 and 26,432 participants from the LCSR and NLST, respectively, were included. Compared with the NLST, the LCSR population was older (mean age [SD]: 64.0 [5.4] versus 61.4 [5.0] years); P < .001) and included more female patients (47.9% versus 40.9%; P < .001), and its patients were more likely to be currently smoking (61.5% versus 48.1%; P < .001). After adjusting for age, gender, smoking history, and body mass index, the LCSR population was more significantly likely to have higher Lung-RADS scores than the NLST (adjusted odds ratio and 95% confidence interval > 1 for Lung-RADS scores 2, 3, 4A, 4B, 4X relative to Lung-RADS 1).

CONCLUSIONS: Lung-RADS scores in clinical LCS are higher than in the NLST, even after adjusting for known confounders such as age and smoking. This would imply higher rates of follow-up testing after LCS and potentially higher cancer rates in the clinically screened population than the NLST.