OBJECTIVE - Little is known about the covariates of hyperglycemia and
hyperinsulinemia. We examined candidate factors in postmenopausal wome
n. RESEARCH DESIGN AND METHODS - We determined the cross-sectional ass
ociations of sociodemographic, body size, lifestyle, reproductive, and
menopausal factors with pre-trial fasting and postchallenge glucose a
nd insulin levels in 869 postmenopausal women aged 45-65 years. Women
were participants in the Postmenopausal Estrogen/Progestin Interventio
ns study who were not taking estrogen or insulin. RESULTS - Plasma glu
cose levels increased significantly with age; serum insulin levels did
not. BMI and waist-to-hip ratio (WHR) each showed graded positive and
independent associations with glucose and insulin levels. Alcohol int
ake, cigarette smoking, physical activity, parity, education, and inco
me were also associated with insulin or glucose in age-adjusted models
. In multivariable models, BMI and WHR explained 18% of the variabilit
y in fasting glucose, 16% in postchallenge glucose, 28% in fasting ins
ulin, and 17% in postchallenge insulin. Age and all other factors comb
ined accounted for <6% of the variance in glucose or insulin. In multi
ply adjusted models, African-American and Hispanic women had higher fa
sting and 2-h insulin levels than non-Hispanic white women. CONCLUSION
S - Most of the variance in glycemia and insulin is unexplained. Measu
res of obesity and fat distribution account for nearly all the explain
ed variance.