Associations of Air Pollution and Pediatric Asthma in Cleveland, Ohio

Table 3

Poisson regression models for each of the fine and coarse PM components individually, adjusting for daily minimum temperature, daily median mixing height, and NO₂ LUR quartiles.

Component	Average of Metro Health & GT Craig sites
	MR^a (95% CI) per doubling	value^b
Fine biomass/industrial	1.028 (0.968–1.091)	0.37
Fine steel Zn and motor vehicle	0.992 (0.954–1.032)	0.69
Fine crustal	1.084 (1.020–1.153)	0.009
Fine steel Fe/fabrication	1.114 (1.072–1.153)	<0.001
Fine coal combustion	0.995 (0.952–1.041)	0.84
Coarse road salt	1.008 (0.991–1.025)	0.35
Coarse motor vehicle	1.072 (01.018–1.129)	0.008
Coarse slag and fly-ash	0.999 (0.973–1.024)	0.91
Coarse crustal	1.051 (0.995–1.111)	0.08
Coarse iron/steel production and fabrication	1.064 (1.029–1.100)	<0.001
Results over the 10 individual PM component models
Minimum temperature^c		0.043 to 0.26
Mixing height^d	0.861 (0.782–0.949) to 0.884 (0.801–0.977)	≤0.016
Seasonality^e fall	1.13 (0.94–1.36) to 1.26 (1.05–1.51)	0.013 to 0.18
Winter	0.93 (0.71–1.23) to 1.07 (0.83–1.38)	≥0.62
LUR NO2 quartile^f		<0.001
1^st	1
2^nd	1.316 (1.133–1.528)
3^rd	1.922 (1.654–2.234)
4^th	1.893 (1.589–2.256)
LUR coarse PM quartile^g	1	<0.001
1^st	0.898 (0.777–1.038)
2^nd	1.314 (1.140–1.514)
3^rd	1.789 (1.517–2.109)
4^th

^aMR: means ratio = proportional change in the mean number of population-adjusted ED visits per day. ^bp values are summarized across the models for the different pollution components. ^cMinimum temperature: no MR estimates are calculated because temperature is modeled using smoothing splines. ^dMixing height: results are summarized as a range across the models for the different pollution components. ^eSeasonality: MR corresponds to a comparison to spring/summer. ^fLUR quartiles: MR corresponds to a comparison to the 1^st quartile. Results are summarized for the component models with the most available data. ^gResults reported for model constructed with coarse PM quartiles in place of NO₂ quartiles. The bold p values reflect statistical significance in our model and meet the rigor of Bonferroni-corrected 0.005 significance level.