Fig. 4
Illustration of temporally-decomposed and related analyses. Hazard ratios (and 95% CIs) for temporally-decomposed cohort analyses estimated using the complex proportional hazards regression model adjusting for age, sex, race-ethnicity, marital status, inflation-adjusted household income, education, smoking status, BMI, U.S. Census region, urban versus rural designation, and survey year. Cardiopulmonary mortality is based on ICD-10 codes and includes: cardiovascular disease (I00-I09, I11, I13, I20-I51), cerebrovascular disease (I60-I69), chronic lower respiratory disease (J40-J47), and influenza and pneumonia (J09-J18). Data used to generate plot are listed in Additional file 1 Table S3.
aTime-independent estimate using 17-yr (1999–2015) mean PM2.5 in the complex proportional hazards regression model [25].
bTime-independent estimate using 28-yr (1988–2015) mean PM2.5 in the basic proportional hazards regression model, with back-casted PM2.5 data for 1988 through 1998 [25].
cCohort results using two- and five-year lagged PM2.5 were pooled using fixed-effect (FE) meta-analysis.