Abstract Background A common aim of lung health research is to determine whether an exposure or intervention is associated with longitudinal change in lung function. This study sought to (i) characterize longitudinal spirometry change in adults by age, sex, and baseline spirometry, and (ii) evaluate modelling strategies with respect to estimation efficiency. We hypothesized that (i) longitudinal change in spirometry would differ by sex, baseline age and spirometry, and (ii) modelling strategy would impact the sample size required to demonstrate that current cigarette smoking is a risk factor for lung function decline. Methods The Canadian Longitudinal Study on Aging (CLSA) is an ongoing population-based cohort of adults 45 years and older that performed standardized anthropometry, questionnaire assessments, and spirometry at up to three study visits. Annualized change in forced expiratory volume in 1-second (FEV1) between study visits was computed as FEV1 at visitn+1 - FEV1 at visitn divided by the time interval between visits, resulting in one to two annualized FEV1change estimates per participant. The annualized FEV1 change by sex was described, and the relationship of annualized FEV1 change to baseline age and FEV1 were assessed via locally weighted smoothing regression. Next, generalized estimating equations were used to fit various linear models of annualized FEV1 change that included sex, baseline age, and baseline FEV1 to assess the minimum sample size required to infer that current cigarette smoking status was harmful. Secondary analyses evaluated annualized forced vital capacity change, baseline height and used linear mixed effect modelling. Results Among 16,692 participants with at least two spirometry visits (mean±SD age: 61.3±9.6 years; 53.0% female; 5.3% current daily cigarette smoking), the mean±SD annualized FEV1 change was -33±60 mL/year. Annualized FEV1change differed significantly by sex, baseline age, and baseline FEV1 (Figure a-c). Current daily cigarette smoking status was associated with accelerated FEV1decline in all models evaluated, but the sample size required to identify a statistically significant estimate differed by up to 4.9-fold (Figure d-h). Results were similar for forced vital capacity, with additional adjustment for baseline height, and when using linear mixed effects models. Conclusion Longitudinal spirometry change in adulthood is associated with sex, baseline age and baseline spirometry. Accounting for these factors may improve efficiency in risk factor epidemiology and clinical trials evaluating lung function change. This abstract is funded by: CIHR
Collins et al. (Fri,) studied this question.