Use of glucose, insulin, and C-reactive protein to determine need for glucose tolerance testing.

Abstract

Glucose intolerance has been shown to be a better predictor of morbidity and mortality than impaired fasting glucose. However, glucose tolerance tests are inconvenient and expensive. This study evaluated the relative frequencies of glucose intolerance and impaired fasting glucose and sought to determine if 2-hour glucose could be predicted from simple demographic and laboratory data in an obese population. Eighty-nine obese subjects (median BMI 35 kg/m(2), range 30 to 40 kg/m(2)) underwent glucose tolerance testing. Using step-wise linear and logistic regression analysis, fasting glucose, high-sensitivity C-reactive protein (hsCRP), fasting insulin, high-density lipoprotein cholesterol, triglycerides, weight, height, BMI, waist circumference, hip circumference, waist-to-hip ratio, sex, and age were assessed as predictors of glucose intolerance. Impaired glucose tolerance was more prevalent (27%) than impaired fasting glucose (5.6%). Only fasting glucose and hsCRP were significant (p < 0.05) independent predictors of impaired 2-hour glucose (>140 mg/dL). A fasting glucose > or =100 mg/dL or an hsCRP > 0.32 mg/dL (upper quartile of the normal range) detected 81% (sensitivity) of obese subjects with impaired glucose tolerance; however, specificity was poor (46%). Fasting insulin > or =6 micro U/mL had better sensitivity (92%) but poorer specificity (30%). Impaired glucose tolerance is more common than impaired fasting glucose in an obese population. Possible strategies to avoid doing glucose tolerance tests in all obese patients would be to do glucose tolerance testing only in those whose fasting glucose is > or =100 mg/dL or whose hsCRP exceeds 0.32 mg/dL or those whose fasting insulin is > or =6 micro U/mL.