Two-Phase Occupational Screening in Coal-Fired Power Plant Workers: Anamnestic Risk Stratification and Targeted Low-Dose CT Findings.

Abstract

Background: Employees at coal-fired power plants experience cumulative occupational exposure to respirable particulates alongside a high prevalence of smoking, both contributing to chronic respiratory disease and elevated lung cancer risk, and organized-screening implementation data remain scarce in this specific industrial setting outside asbestos-exposed cohorts. We evaluated a two-step occupational screening strategy integrating an anamnestic questionnaire with targeted low-dose CT (LDCT) imaging, assessing risk distribution, structured imaging findings, and the discriminative value of questionnaire-based triage relative to its underlying continuous predictors. Methods: In this cross-sectional study, all workers at a coal-fired power plant completed a structured questionnaire (F1) and were classified as high-risk or non-high-risk using a predefined eligibility rule (respiratory symptoms, or age ≥ 50 years with current/former smoking). All workers were invited to voluntary LDCT (F2), interpreted using Lung-RADS v2022 with structured emphysema and coronary artery calcium (CAC) assessment. Associations were assessed using Fisher's exact tests, multivariable logistic regression, and ROC analysis with DeLong's test; voluntary LDCT uptake was characterized with a multivariable propensity model. Results: Among 870 screened workers, 364 (41.8%) were high-risk. LDCT uptake was higher among high-risk workers (130/364, 35.7%) than non-high-risk workers (56/506, 11.1%), yielding 186 examinations. Only 14/186 (7.5%) had Lung-RADS ≥ 3. High-risk status was associated with higher prevalence of any nodule (80.8% vs. 62.5%; OR 2.52) and CAC (46.9% vs. 14.3%; OR 5.30); a composite abnormality outcome was present in 92.3% vs. 69.6% (OR 5.23). Only age (OR 1.11/year) and ever-smoking status (OR 4.38) independently predicted abnormalities on multivariable analysis; dichotomized risk classification showed reduced discrimination for CAC versus a continuous age-smoking model (AUC 0.647 vs. 0.779). Conclusions: Questionnaire-based triage effectively enriched the LDCT-screened population for smoking-related and cardiometabolic risk. Age and smoking history, rather than the full questionnaire, principally drove imaging abnormalities, supporting an integrated occupational model combining lung cancer screening with COPD and cardiovascular risk management.