Table 2.
Estimates of excess deaths attributable to changes in PM2.5 concentration over time by form of hazard function (linear or non-linear), cohort (CanCHEC and CPS II), and country (Canada and USA)
| Country (population weighted change in PM2.5) | Hazard model form—cohort | Number of excess deathsa (95 % confidence interval) | Percent change in baseline mortality rate |
|---|---|---|---|
| Canada (2.0 μg/m3) | Linear—CanCHEC | 3480 (2940–4020) | 1.55 |
| Linear—CPS II | 3090 (2430–3750) | 1.38 | |
| Non-linear optimal—CanCHEC | 3146 (2700–3610) | 1.41 | |
| Non-linear ensemble—CanCHEC | 3320 (2720–4060) | 1.48 | |
| Non-linear optimal—CPS II | 4300 (3420–5200) | 1.92 | |
| Non-linear ensemble—CPS II | 4240 (3100–5560) | 1.90 | |
| Combined non-linear ensemblea | 3640 (2780–4500) | 1.62 | |
| USA (3.7 μg/m3) | Linear—CanCHEC | 68,700 (58,000–79,300) | 2.82 |
| Linear—CPS II | 60,900 (47,500–74,100) | 2.50 | |
| Non-linear optimal—CanCHEC | 46,600 (39,700–53,400) | 1.92 | |
| Non-linear ensemble—CanCHEC | 49,000 (40,700–57,100) | 2.01 | |
| Non-linear optimal—CPS II | 77,700 (62,200–93,100) | 3.20 | |
| Non-linear ensemble—CPS II | 76,700 (60,600–93,000) | 3.15 | |
| Combined non-linear ensemblea | 61,900 (34,700–89,100) | 2.54 |
aMeta-analytic combination of CanCHEC and CPS II ensemble models