Context: Previous studies report cross-sectional associations between circulating cortisol levels and type 2 diabetes in older adults; none assessed the association prospectively. Objective: This study’s objective was to examine the association of serum cortisol with prevalent and incident diabetes in community-dwelling older adults. Methods: Between 1984-87, 885 men and 729 women aged 50 or older had serum fasting morning cortisol measured and an oral glucose tolerance test administered at a baseline research clinic visit and were followed for up to 25 years. Logistic and cox proportional hazards regression models were used to assess cortisol associations with prevalent and incident diabetes. Results: Median age of participants was 75 years, median body mass index was 25kg/m2, and median cortisol was 91mg/ml. At baseline, 249 participants had prevalent diabetes; 118 developed diabetes during follow-up. In unadjusted analyses, the odds of prevalent diabetes for the highest cortisol quartile compared to the lowest was 1.95 (95% CI: 1.32, 2.87) and remained significant after adjusting for age, sex and diabetes risk factors (OR: 1.85, 95% CI: 1.21, 2.83); associations were not significant for the second and third quartiles. Serum cortisol was not associated with incident diabetes before or after adjusting for covariates (fully-adjusted HR for highest versus lowest quartile=1.02, 95% CI: 0.58, 1.78). Conclusions: Results suggest that in older adults, elevated cortisol levels may be a result of diabetes and its complications, rather than a cause. Further studies using cohorts with repeated cortisol measures are needed to confirm the direction of the hypercortisolemia-diabetes link.
