Abstract 11325: Prediction of Diabetes in the Framingham Offspring Study
Introduction: Diabetes mellitus, defined as a fasting serum glucose of > 125 mg/dL, is a leading cause of heart disease, kidney failure, blindness, and neuropathy. Lifestyle modification with caloric restriction and exercise has been shown to decrease the risk of developing diabetes by more than 50%. Our goal in this research was to develop an optimal model for predicting diabetes risk over approximately 10 years in middle aged subjects in order to be able individuals who would be candidates for lifestyle modification.
Methods: Using plasma or serum samples obtained from subjects after an overnight fast who were participants in cycle 6 of the Framingham Offspring Study, the following parameters were measured using standardized automated analyses with coefficients of variation of < 5%: adiponectin, C reactive protein, glycated albumin, high density lipoprotein cholesterol (HDL-C), insulin, triglycerides, and total cholesterol. In addition blood pressure, body mass index, waist circumference, parental history of diabetes, and use of cholesterol lowering medications was assessed. Complete follow-up information including coronary heart disease status (CHD) over a median of 9.3 years was available for 2,416 men and women, mean age 58 years and non-diabetic at onset, of whom 166 developed diabetes (6.9%, fasting glucose > 125 mg/dL or on treatment for diabetes). A model was developed to predict new onset diabetes using only those variables that provided significant information in a stepwise logistic regression analysis to calculate the area under the curve (AROC) and C statistic.
Results: The following variables entered the model and provided an overall C statistic of 0.924: 1. fasting serum glucose, 2. body mass index, 3. % glycated albumin, 4. adiponectin, 5. parental history of diabetes, 6. triglycerides, and 7. use of cholesterol lowering medications. This model is far superior to previous published models which have maximal C statistic values of 0.85. High risk is > 20% over 10 years, moderate risk is 10-20%, and low risk is < 10%.
Conclusions: An optimal model for predicting new onset diabetes over about 10 years has been developed and can be used to identify high risk subjects who can then be targeted for aggressive lifestyle management.
- © 2013 by American Heart Association, Inc.