Table 4.

Cox proportional-hazards modeling results for the secondary analysis of the association between CHD and categorical TWA inorganic arsenic exposure.

Variable	Univariate model HR (95% CI)	Full model HR (95% CI)	Final model HR (95% CI)
Arsenic exposure TWA
1–20 μg/L-yeara	1.0	1.0	1.0
20–30 μg/L-yearb	1.24 (0.70, 2.31)	1.25 (0.60, 2.61)	1.23 (0.56, 2.18)
30–45 μg/L-yearc	2.14 (1.22, 3.98)	2.08 (1.11, 3.92)	2.18 (1.23, 4.02)
45–88 μg/L-yeard	3.12 (1.11, 9.02)	3.34 (1.15, 9.30)	3.10 (1.10, 9.11)
Female sex		0.41 (0.24, 0.69)	0.35 (0.19, 0.48)
Hispanic ethnicity		1.01 (0.62, 1.70)
Primary family member diagnosed with CHD		1.55 (0.88, 2.70)	1.63 (0.94, 2.82)
Low income		1.21 (0.71, 2.20)
Diabetic (diagnosed at baseline visit)		2.05 (0.51, 9.21)
BMI (per 5.5 kg/m2)		0.85 (0.59, 1.22)
Sedentary physical activity		1.12 (0.65, 1.80)
Current smoker		1.03 (0.63, 1.68)
High alcohol consumption		1.70 (0.72, 4.10)
Low-density cholesterol (53 μg/dL)e		1.46 (1.03, 2.08)	1.40 (1.02, 1.99)
Triglycerides (90 μg/dL)e		0.94 (0.65, 1.34)
High-density cholesterol (17 μg/dL)e		0.67 (0.43, 1.02)
Folate (57 μg)e		1.00 (0.99, 1.01)
Selenium (185 μg)e		1.00 (0.99, 1.03)
Univariate model: proportional-hazards model with TWA arsenic exposure (main risk factor) as independent variable. Full model: proportional-hazards model with TWA arsenic exposure (main risk factor) and all listed variables as time-dependent independent variables. Final model: proportional-hazards model with TWA arsenic exposure (main risk factor) and statistically significant covariates (independent variables). Time-dependent covariates are BMI, physical activity, smoking status, alcohol consumption, serum lipid levels, and micronutrient intake. aPerson-years of follow up = 4,806. bPerson-years of follow up = 1,335. cPerson-years of follow up = 534. dPerson-years of follow up = 98. eIQR of subcohort at baseline.