Effect modification of long-term air pollution exposure on the association of physical activity with COPD hospitalization: a prospective cohort study of 0.5 million Chinese adults

Figures

• 

  Fig. 1. Exposure-response relationships of physical activity level with COPD incidence stratified by the median of average annual concentration of (A) PM2.5 and (B) NO2, and (C) warm-season O3. Solid line represents hazard ratios, and the ribbon represents its 95% confidence interval. All models were stratified by age-at-risk (in 5-year scale), ten study areas, and sex, and were adjusted for education, occupation, household income, smoking status, alcohol drinking, cooking fuel type, heating fuel type, self-rated health status, body mass index, temperature and relative humidity. p-values shown were derived from likelihood ratio tests comparing models with and without interaction terms between physical activity and air pollution, which were FDR-adjusted. Abbreviation: MET-h/d, metabolic equivalent task hours per day; FDR, false discovery rate.

• 

  Fig. 2. Associations of physical activity level categories with COPD incidence stratified by the median of average annual concentration of (A) PM2.5 and (B) NO2, and (C) warm-season O3. All models were stratified by age-at-risk (in 5-year scale), ten study areas, and sex, and were adjusted for education, occupation, household income, smoking status, alcohol drinking, cooking fuel type, heating fuel type, self-rated health status, body mass index, temperature and relative humidity. Abbreviation: MET-h/d, metabolic equivalent task hours per day.

• 

  Fig. 3. Exposure-response relationships of physical activity level with COPD incidence stratified by the upper tertile of average annual concentration of (A) PM2.5 and (B) NO2, and (C) warm-season O3. Solid line represents hazard ratio and the ribbon represents its 95% confidence interval. All models were stratified by age-at-risk (in 5-year scale), ten study areas, and sex, and were adjusted for education, occupation, household income, smoking status, alcohol drinking, cooking fuel type, heating fuel type, self-rated health status, body mass index, temperature and relative humidity. Abbreviation: MET-h/d, metabolic equivalent task hours per day.

• 

  Fig. 4. Associations of physical activity level categories with COPD incidence stratified by the upper tertile of average annual concentration of (A) PM2.5 and (B) NO2, and (C) warm-season O3. All models were stratified by age-at-risk (in 5-year scale), ten study areas, and sex, and were adjusted for education, occupation, household income, smoking status, alcohol drinking, cooking fuel type, heating fuel type, self-rated health status, body mass index, temperature and relative humidity. Abbreviation: MET-h/d, metabolic equivalent task hours per day.

• 

  Fig. 5. Exposure-response relationships of physical activity level with COPD incidence stratified by the median of average annual concentration of PM2.5 (A & D) and NO2 (B & E), and warm-season O3 (C & F) and smoking status. Solid line represents hazard ratio and the ribbon represents its 95% confidence interval. Never-regular smoker: never or occasional smoker; Ever-regular smoker: ex-regular or current smoker. All models were stratified by age-at-risk (in 5-year scale), ten study areas, and sex, and were adjusted for education, occupation, household income, alcohol drinking, cooking fuel type, heating fuel type, self-rated health status, body mass index, temperature and relative humidity. p-values shown were derived from likelihood ratio tests comparing models with and without interaction terms between physical activity and air pollution, which were FDR-adjusted. Abbreviation: MET-h/d, metabolic equivalent task hours per day; FDR, false discovery rate.