Serum 25-hydroxyvitamin D concentration does not independently predict incident diabetes in older women

Aims To investigate whether 25-hydroxyvitamin D concentration was associated with incident diabetes in a large cohort of older women. Methods Data were analysed from women included in the Study of Osteoporotic Fractures, a cohort of community-dwelling women aged >= 65years at enrolment. Serum 25-hydroxyvitamin D concentration was assessed at the year 6 visit, as were BMI and other factors associated with vitamin D and/or diabetes. Diabetes status was determined at each subsequent visit by self-report and medication use. Only those without prevalent diabetes at the year 6 visit were included in the present analysis (N=5463, mean age 76.5years). Results During a mean +/- sd follow-up of 8.6 +/- 4.4years, incident diabetes was reported in 320 participants. The mean BMI was higher in those with a 25-hydroxyvitamin D concentration <20ng/ml (<50nmol/l) than in those with concentrations 20-30 or >= 30ng/ml [50-74 or >= 75nmol/l (P<0.0001)]. A higher 25-hydroxyvitamin D concentration was associated with a 13% lower risk of incident diabetes after adjustment for age and clinic site [hazard ratio 0.87, 95% CI 0.76-0.99, per sd increase in 25-hydroxyvitamin D]; however, the addition of BMI to the model attenuated the estimated effect (hazard ratio 0.97, 95% CI 0.86-1.11). Adjustment for additional potential confounders yielded similar results. Conclusions Serum 25-hydroxyvitamin D does not independently predict incident diabetes in older women. Although those with higher 25-hydroxyvitamin D concentrations are less likely to develop diabetes, this is mainly explained by their lower BMI.