Table 3

HR for incident diabetes with twofold increase in urinary metal concentrations

MetalsInitial model*Full model†
HR (95% CI)P valueHR (95% CI)P value‡
Arsenic1.06 (0.98 to 1.15)0.131.19 (1.10 to 1.30)<0.0001
Barium0.98 (0.88 to 1.09)0.700.96 (0.85 to 1.09)0.53
Cadmium1.00 (0.90 to 1.10)0.920.96 (0.86 to 1.07)0.42
Cobalt0.97 (0.85 to 1.10)0.601.01 (0.88 to 1.15)0.90
Cesium1.12 (0.92 to 1.37)0.251.23 (0.98 to 1.50)0.06
Copper1.06 (0.91 to 1.23)0.470.96 (0.82 to 1.13)0.65
Mercury0.83 (0.75 to 0.92)0.030.92 (0.82 to 1.03)0.12
Manganese1.14 (0.94 to 1.37)0.181.10 (0.90 to 1.35)0.33
Molybdenum0.93 (0.81 to 1.07)0.291.04 (0.90 to 1.21)0.58
Nickel1.08 (0.93 to 1.25)0.331.15 (0.98 to 1.35)0.08
Lead1.12 (0.99 to 1.27)0.071.20 (1.05 to 1.37)0.006
Antimony1.03 (0.91 to 1.17)0.611.07 (0.93 to 1.22)0.36
Tin1.10 (1.01 to 1.20)0.031.11 (1.01 to 1.22)0.04
Thallium1.05 (0.95 to 1.16)0.381.04 (0.93 to 1.16)0.52
Zinc1.48 (1.27 to 1.74)<0.00011.31 (1.11 to 1.53)0.001

  • All models were constructed by Cox proportional hazards model.

  • *Initial model: adjustment for age, race/ethnicity, study site, and specific gravity (log-transformed).

  • †Full model: initial model with additional adjustment for education, household income, body mass index (baseline level), waist circumference (baseline level), smoking status, alcohol consumption, physical activity score, total energy intake, menopausal status, and use of hormone. In full model, seafood and rice intake was additionally adjusted for arsenic, cadmium, and mercury models; zinc intake from diets and supplements was additionally adjusted for zinc model.

  • ‡Significance level at α=0.006 corresponding to a false discovery rate of 0.05 using the Benjamini-Hochberg method.