Table 2.

Mean differences in gestational age and term birth weight for a +10-percentage point increase in the probability that private well arsenic concentrations exceed the respective threshold

	Gestational Age (weeks)			Term Birth Weight (grams)
No. of Births	3,580,755			3,305,090
No. of Counties	3,105			3,105
Private well arsenic level	Unadjusted^a β (95% CI)	Adjusted^b β (95% CI)	Adjusted 2^c β (95% CI)	Unadjusted^a β (95% CI)	Adjusted^b β (95% CI)	Adjusted 2^c β (95% CI)	Adjusted 3^d β (95% CI)
Boosted regression tree: 1 μg/L
≤1 μg/L	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)
>1 μg/L	−0.01 (−0.01, −0.01)*	−0.01 (−0.01, −0.01)*	−0.01 (−0.01, −0.01)*	−3.13 (−4.30, −1.95)*	−3.12 (−4.05, −2.18)*	−3.10 (−4.04, −2.17)*	−2.84 (−3.74, −1.94)*
Boosted regression tree: 5 μg/L
≤5 μg/L	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)
>5 μg/L	−0.01 (−0.01, 0.00)	−0.01 (−0.01, 0.00)	−0.01 (−0.01, 0.00)	−2.24 (−4.10, −0.38)*	−2.73 (−4.19, −1.26)*	−2.75 (−4.21, −1.29)*	−2.20 (−3.61, −0.78)*
Boosted regression tree: 10 μg/L
≤10 μg/L	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)
>10 μg/L	−0.01 (−0.02, 0.00)	−0.01 (−0.02, 0.00)	−0.01 (−0.02, 0.00)	−3.08 (−5.91, −0.26)*	−3.85 (−6.07, −1.64)*	−3.89 (−6.10, −1.68)*	−2.99 (−5.13, −0.86)*
Random forest classification: 5 and 10 μg/L
≤5 μg/L	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)	0.00 (reference)
>5 to ≤10 μg/L	0.00 (−0.01, 0.02)	0.00 (−0.02, 0.01)	0.00 (−0.02, 0.01)	−3.62 (−7.97, 0.73)	−4.39 (−7.91, −0.86)*	−4.37 (−7.89, −0.85)*	−1.62 (−5.06, 1.82)
>10 μg/L	0.00 (−0.02, 0.01)	0.00 (−0.01, 0.01)	0.00 (−0.01, 0.01)	−1.49 (−4.79, 1.82)	−1.54 (−4.16, 1.06)	−1.56 (−4.16, 1.05)	−1.95 (−4.47, 0.40)

Note: CI, confidence interval. Asterisks denote statistical significance.

Unadjusted model includes the percentage of private wells in the county predicted to have arsenic concentrations within the respective category with nested random intercepts for maternal county and state of residence.

Adjusted model additionally includes maternal age (modeled as a restricted cubic spline), race/ethnicity (non-Hispanic white, non-Hispanic black, non-Hispanic other, or Hispanic), marital status (unmarried or married), educational attainment (less than high school, high school diploma or equivalent, some college, or college degree or greater), smoking during pregnancy (yes or no), pre-pregnancy BMI (modeled as a restricted cubic spline), rurality/urbanicity (metropolitan, non-metropolitan, less urbanized, or thinly populated), and the average annual concentration of particulate matter with aerodynamic diameter less than 2.5 μm in the county (μg/m3).

Adjusted model 2 additionally includes the proportion of private well users in each county.

Adjusted model 3 additionally includes the average elevation of the county (m).