Table 3.
Multivariate linear regression model results on blood cadmium, lead, manganese, total mercury, and selenium levels (N=1,637 study participants who had non-missing values for all covariates). Data shown are regression-adjusted geometric mean ratios (i.e., ratio of the current level geometric mean to that of the reference level, adjusted for all other variables in the model) and 95% CI.
| Variable | Level | Cadmium1 | Lead1 | Total Mercury1,2 | Manganese1 | Selenium1 |
|---|---|---|---|---|---|---|
| Sex (Ref: Male) | Female | 1.25 (1.18–1.33) | 0.71 (0.67–0.75) | - | 1.03 (0.99–1.07) | - |
| Age group (Ref: 70+) | 50–59 years | 0.93 (0.86–1.00) | - | - | - | - |
| 60–69 years | 0.91 (0.85–0.98) | - | - | - | - | |
| Center (Ref: Arizona) | North Dakota/ South Dakota | 2.04 (1.86–2.24) | 1.45 (1.33–1.59) | - | 1.25 (1.18–1.32) | 1.12 (1.09–1.15) |
| Oklahoma | 1.23 (1.12–1.35) | 1.19 (1.09–1.30) | - | 1.10 (1.04–1.17) | 1.06 (1.03–1.09) | |
| Body Mass Index (Ref: Normal, 18.5–24.9 kg/m2) | <18.5 kg/m2 | - | 1.32 (1.01–1.73) | - | - | 0.80 (0.73–0.87) |
| 25.0–29.9 kg/m2 | - | 0.90 (0.83–0.97) | - | - | 1.04 (1.02–1.07) | |
| ≥30 kg/m2 | - | 0.83 (0.77–0.89) | - | - | 1.02 (0.99–1.04) | |
| Waist circumference (Ref: Normal) | At risk | 0.93 (0.88–0.99) | - | 0.89 (0.81–0.98) | 1.07 (1.03–1.11) | - |
| Household income (Ref: <15,000/year) | $15,000+/year | - | 0.93 (0.87–1.00) | 1.37 (1.21–1.54) | - | - |
| Income meets needs (Ref: Needs met) | Needs not met | - | - | 0.75 (0.66–0.85) | - | - |
| Smoker (Ref: Never) | Former | 1.04 (0.98–1.11) | 1.03 (0.97–1.09) | - | - | - |
| Current | 1.99 (1.85–2.14) | 1.20 (1.13–1.29) | - | - | - | |
| Second-hand smoke (Ref: 0h/d) | 1–7 hours/day | 0.99 (0.93–1.06) | - | - | - | - |
| 8+ hours/day | 1.14 (1.03–1.25) | - | - | - | - | |
| Alcohol consumption (Ref: No drinks) | 1–5 drinks/week | 1.06 (0.96–1.17) | 1.13 (1.03–1.24) | 1.21 (1.03–1.43) | 1.00 (0.95–1.07) | - |
| >5 drinks/week | 1.12 (1.02–1.23) | 1.24 (1.13–1.35) | 0.98 (0.84–1.15) | 1.06 (1.00–1.13) | - | |
| Education (Ref: 12+ years) | <12 years | 1.07 (1.01–1.13) | 1.07 (1.01–1.13) | 0.77 (0.70–0.85) | 1.05 (1.01–1.09) | - |
| Model Fit (R2) | 0.44 | 0.26 | 0.10 | 0.07 | 0.07 | |
Variables with blank cells were not included in the final model.
For blood mercury, quantile regression was also used to model the median mercury level for the left-censoring due to non-detects. Both quantile regression and multi-variate linear regression models indicated the same set of regressors, so only the latter is reported in this table to produce the regression-adjusted geometric mean ratios