Skip to main content
. 2020 Aug 26;191:110052. doi: 10.1016/j.envres.2020.110052

Table 3.

Summary of sensitivity analyses of association of PM2.5 and COVID-19 incidence.

Model IRR 95% CI % cases % population
Best fitting multivariate model (province as fixed effect) 1.07 0.97–1.18 99.9a 99.6a
Exclude Montréal 1.07 0.96–1.18 72.4 94.5
Exclude 3 provinces with aggregated health regions 1.15 1.00–1.32b 94.6 81
Exclude health regions with population density < median 1.16 1.00–1.34b 94.8 82
Ontario and Quebec only 1.21 1.03–1.40 84.9 61.5
April 19 data (35,986 cases) 1.10 1.01–1.19 99.9 99.6
Best fitting multivariate model (province as random effect) 1.09 0.98–1.21c 99.9a 99.6a
a

Five health regions excluded due to missing data.

b

Lower bound >1 with additional decimal places.

c

Corrected for overdispersion by multiplying standard error by square root of dispersion factor (Bolker, 2020).