Trajectories of cigarette smoking and exposure to welding fumes and their impact on lung cancer risks: a latent class modelling approach
Menée à partir des données de 2 études allemandes portant sur 3 539 témoins et 3 498 patients atteints d'un cancer du poumon, cette étude estime le risque de développer la maladie en fonction des trajectoires d'exposition aux fumées de soudage et au tabagisme
Objectives: Traditional epidemiological approaches usually assume a constant relationship between cumulative exposure and disease, which implies that exposure duration and intensity contribute equally to the studied outcome. But individuals with the same cumulative exposure but different temporal exposure patterns may show different risks. Trajectory classification is a good way to assess exposure–risk associations and leads to a better understanding of lifetime variability in exposure levels. Therefore, this study aimed to estimate lung cancer risk according to the exposure trajectory classes on welding fumes and cigarette smoking.
Design: Two population-based German case–control studies.
Participants: 3498 male lung cancer cases and 3539 male control subjects.
Methods: Separate latent class mixed models (LCMM) were determined to identify profiles of exposure trajectories of cigarette smoking and occupational exposure to welding fumes. To investigate the risk of lung cancer by class membership, ORs with 95% CI were estimated via multiple logistic regression analyses.
Results: LCMM each identified four latent classes of smoking and welding-fume exposure. Classes of smokers showed much higher risk of lung cancer compared with never smokers or subjects exposed to welding fumes. Smokers in one class characterised with the highest exposure over the past 10 years had the highest adjusted lung cancer risk (OR=39; 95% CI 29 to 53). For welding, the highest lung cancer risks were found for the class in which exposure to welding fumes in the past 10 years prior to the diagnosis of lung cancer was highest and the duration of welding was also quite high (OR=1.71; 95% CI 0.92 to 3.15).
Conclusions: In summary, LCMM opens a new perspective on dose–effect relationships and could be employed to complement established epidemiological methods.No data are available. The data underlying this study are not publicly available due to privacy restrictions. Data sharing is therefore not possible.
BMJ Open , article en libre accès, 2026