Figure 4.

Comparison of hazard ratio (HR) estimates and confidence interval (CI) for the four causes of deaths between the present study, IER predictions and meta-analysis for the outdoor air pollution (OAP) study, based on the same exposure contrast between the fifth ($15.5{\mathrm{\mu }\mathrm{g}/\mathrm{m}}^3$) and 95th ($77.1{\mathrm{\mu }\mathrm{g}/\mathrm{m}}^3$) percentile (as was used for IER) for the age range of 60–64 y. IER predictions: estimates based on IER functions as reported in Cohen et al, (2017) between the 5th ($15.5{\mathrm{\mu }\mathrm{g}/\mathrm{m}}^3$) and 95th ($77.1{\mathrm{\mu }\mathrm{g}/\mathrm{m}}^3$) percentile for the age range of 60–64 years. Meta-analysis: HRs (95% CI) presented was calculated from the meta-analysis summary estimate and standard error based on the same exposure contrast as was used for the IER ($15.5–77.1{\mathrm{\mu }\mathrm{g}/\mathrm{m}}^3$), using the cohorts [a list of the cohort was in Table 1 in Cohen et al. (2017)] examining the association between outdoor concentrations of ${\mathrm{PM}}_{2.5}$ and cause-specific mortality.