Predicting impaired glucose tolerance using common clinical information: data from the Third National Health and Nutrition Examination Survey.

OBJECTIVE To develop a score to predict impaired glucose tolerance (IGT) using common clinical data. RESEARCH DESIGN AND METHODS We analyzed data from the Third National Health and Nutrition Examination Survey (NHANES III) for 2,746 individuals aged 40-74 years who completed an oral glucose tolerance test. IGT was defined as a 2-h postchallenge glucose > or =140 mg/dl (7.7 mmol/l). We performed bivariate and multivariate analyses to describe the association of IGT with commonly available clinical information. A numerical score to predict IGT was derived from the results of the multivariate logistic regression models. RESULTS Fasting glucose levels between 101 and 109 mg/dl (5.6 and 6.0 mmol/l) or between 110 and 125 mg/dl (6.1 and 6.9 mmol/l) were associated with IGT (odds ratio 1.8 and 6.2, respectively; P < 0.05). BMI > or =25 kg/m(2), Mexican-American ethnicity, age between 60 and 74 years, hypertension, and triglyceride level > or =150 mg/dl (1.69 mmol/l) were also associated with IGT. The area under the receiver operating characteristic curve for an 8-point scale derived from the multivariate analysis was 0.74 (95% CI 0.72-0.76). Setting a low cut point of 2 on this scale resulted in high sensitivity (86%), whereas a high cut point of 6 yielded high specificity (97%) for the detection of IGT. CONCLUSIONS A numerical score based on common clinical data can identify individuals with a low or high likelihood of having IGT.