Table 8– Logistic regression dose–response model in nonsmokers, with the primary exposure variable of urinary arsenic excretion and the outcome variable of interest of the presence or absence of dyspnoea

Independent variables	OR (95% CI)	p-value
Urinary arsenic quintile^#
1: reference	1.00
2	1.37 (0.97–1.92)	0.073
3	1.92 (1.38–2.65)	<0.001
4	1.94 (1.41–2.68)	<0.001
5	1.87 (1.36–2.58)	<0.001**

All data were adjusted for age, education, body mass index, systolic blood pressure and sex. ^#: equal quintiles of urinary arsenic excretion (in μg·L⁻¹) divided by the urinary concentration of creatinine (in g·L⁻¹), resulting in an adjusted concentration of arsenic (in μg·g⁻¹ creatinine). **: p<0.01 by Chi-squared test for trend.