Table 3

Incident diabetes per 5% increase in arsenic metabolism biomarkers iAs%, MMA%, and DMA%

Arsenic metabolismModel 1Model 2Model 3Model 4
Conventional approach (per 5% increase)
 iAs%0.89 (0.80–1.00)0.92 (0.82–1.03)0.93 (0.83–1.05)1.00 (0.89–1.12)
 MMA%0.77 (0.69–0.85)0.78 (0.70–0.86)0.78 (0.70–0.87)0.84 (0.76–0.94)
 DMA%1.14 (1.07–1.22)1.14 (1.06–1.21)1.13(1.06–1.21)1.07 (1.00–1.15)
Leave-one-out approach
 iAs% (per 5% increase) corresponds to:
  MMA% (per 5% decrease)1.32 (1.10–1.59)1.34 (1.11–1.61)1.35 (1.12–1.62)1.31 (1.08–1.58)
  DMA% (per 5% decrease)1.01 (0.90–1.13)1.03 (0.92–1.15)1.04 (0.92–1.16)1.08 (0.96–1.21)
 MMA% (per 5% increase) corresponds to:
  iAs% (per 5% decrease)0.76 (0.63–0.91)0.75 (0.62–0.90)0.74 (0.62–0.89)0.77 (0.63–0.93)
  DMA% (per 5% decrease)0.77 (0.69–0.86)0.77 (0.69–0.86)0.77 (0.69–0.86)0.82 (0.73–0.92)
 DMA% (per 5% increase) corresponds to:
  iAs% (per 5% decrease)0.99 (0.88–1.11)0.97 (0.87–1.09)0.96 (0.86–1.08)0.93 (0.82–1.05)
  MMA% (per 5% decrease)1.31 (1.17–1.46)1.30 (1.16–1.45)1.30 (1.16–1.45)1.21 (1.08–1.36)
  • Data are hazard ratio (95% CI). Because the three biomarkers equal 100%, models entered two biomarkers at a time. All models were adjusted for the sum of iAs, MMA, and DMA (μg/g creatinine). In the conventional approach, each arsenic metabolism biomarker (iAs%, MMA%, and DMA%) is entered alone in the model. In the leave-one-out approach, two arsenic metabolism biomarkers are entered in the model. In that model, an increase in each arsenic metabolism biomarker corresponds to a decrease in the biomarker that is left out of the model. For instance, an increase in iAs% corresponds to a decrease in MMA% when DMA% is included in the model and MMA% is left out. The magnitude of the association for iAs% when MMA% is left out will be the same but in the opposite direction of MMA% when iAs% is left out. Both in the conventional approach and in the leave-one-out approach we adjusted for the sum of inorganic and methylated arsenic concentrations in urine to hold arsenic exposure levels constant. Model 1: stratified by study center, adjusted for age (age as time metric and age at baseline were treated as staggered entries), and adjusted for the sum of iAS and methylated arsenic concentrations. Model 2: further adjusted for sex and education. Model 3: further adjusted for smoking and alcohol drinking. Model 4: further adjusted for BMI and waist-to-hip ratio.