Association between Six Environmental Chemicals and Lung Cancer Incidence in the United States

Table 2

Association of lung cancer incidence with six selected chemicals using linear regression*.

	Number of counties with non zero release	Univariate model		Multivariate model
Carcinogens		Coefficient	𝑃 value <	Coefficient	𝑃 value <
Arsenic (yes : no)	2	−0.08	0.99	−2.74	0.76
Arsenic (in continuous)		−0.06	0.97	−0.50	0.73
1,3-butadiene (yes : no)	12	3.36	0.51	3.88	0.34
1,3-butadiene (in continuous)		0.63	0.30	0.65	0.17
Cadmium (yes : no)	11	5.11	0.33	2.94	0.46
Cadmium (in continuous)		1.10	0.34	0.49	0.58
Chromium (yes : no)	46	8.59	0.002	5.33	0.02
Chromium (in continuous)		1.43	0.0009	0.99	0.008
Formaldehyde (yes : no)	34	12.40	0.0001	9.08	0.0009
Formaldehyde (in continuous)		1.40	0.0001	1.02	0.001
Nickel (yes : no)	43	9.57	0.0008	5.98	0.01
Nickel (in continuous)		1.39	0.002	0.95	0.01
No release of selected chemicals (reference)	68
Low total amount of release^†	38	5.44	0.07	4.47	0.05
High total amount of release^†	30	14.07	0.0001	12.34	0.0001
Total amount (in continuous)		1.28	0.0001	1.10	0.0001

*The amount of each selected chemical in continuous and total amount of all selected chemicals were log transformed. ^†The cut point for low or high release of total amount of selected chemicals was set at 2000 pounds annually (low: <2000 pounds, high: ≥2000 pounds). In the multivariate model for each selected chemical, adjusted variables included proportion of nonwhite population in 1990, prevalence of smoking in 2003–2006, proportion of male population, proportion with college or higher education in 1990, proportion of families below poverty in 1990, and metro or nonmetro counties (metro counties: Rural-Urban Continuum Code < 4; nonmetro counties: Rural-Urban Continuum Code ≥ 4).