Trends in chronic obstructive pulmonary disease and their associations with leisure-time physical activity and sitting time in U.S. adults, 2007–2018

To describe secular patterns in leisure-time physical activity (LTPA), sitting time, and chronic obstructive pulmonary disease (COPD) prevalence and their cross-sectional associations among U.S. adults. We analyzed six repeated cross-sectional NHANES cycles from 2007–2018 among adults aged ≥20 years. The certified source roster contained 34,556 participants, and the formal primary model cohort contained 31,722 participants, including 2,400 participants with self-reported COPD. LTPA was classified as inactive, insufficient, or sufficient, and sitting time as <4, 4–<8 h/day, or ≥8 h/day. We used complex-survey methods, direct standardization to the projected year 2000 U.S. population with five adult age strata, and survey-weighted logistic regression. Only continuous poverty-to-income ratio (PIR) was multiply imputed by predictive mean matching (m=10); each completed dataset was analyzed separately and coefficients and variances were combined with Rubin’s rules. Five-age-group standardized COPD prevalence was 7.3% in 2007–2008, 5.5% in 2009–2010, and 8.5% in 2017–2018 (global categorical P=0.0489; linear P=0.0369; quadratic-component P=0.2782). In the fully adjusted primary model, sufficient LTPA was associated with lower prevalence odds of COPD than inactivity (prevalence OR, 0.726; 95% CI, 0.618–0.852). The corresponding prevalence ORs for sitting 4–<8 h/day and ≥8 h/day versus <4 h/day were 1.052 (95% CI, 0.910–1.217) and 1.131 (95% CI, 0.954–1.339). In a model-prediction exercise for 2017–2018, setting LTPA to the sufficient category for all model records gave a predicted prevalence of 7.64% versus 9.03% under observed exposures; the absolute difference was 1.392 percentage points (95% CI, 0.581–2.203) and the relative difference was 15.42% (95% CI, 7.09%–23.75%). The low-sitting scenario had an absolute difference of 0.355 percentage points (95% CI, −0.387–1.098). These scenario contrasts are not estimates of preventable COPD. Sufficient LTPA was associated with lower COPD prevalence odds after five-category age adjustment. The repeated cross-sectional design cannot establish temporal or causal direction, and reverse causation remains plausible. The hypothetical exposure scenarios are descriptive model-based contrasts and not intervention effects.