Individualized Diabetes Risk Prediction in Women with a History of Gestational Diabetes

Abstract

Without treatment, prediabetic women with a history of gestational diabetes (GDM) are 71% more likely to progress to type 2 diabetes (DM) compared to parous women without prior GDM. In women with prior GDM, both intensive lifestyle intervention (ILI) and metformin are effective in reducing the progression to DM in the Diabetes Prevention Program (DPP). Evidence suggests heterogeneity in treatment response. We performed a secondary analysis of DPP data, including 1555 parous women with prediabetes; 19% had prior GDM, 31% were < 45 years old, 44% were of non-Caucasian race/ethnicity and 71% were obese. At three years, 323 women developed diabetes. We examined 16 baseline clinical variables using Cox proportional hazards regression with backward elimination (excluding variables with p>0.05) to predict DM risk. Our final multivariable model includes treatment arm (ILI vs. metformin vs. placebo), BMI category, height, waist-hip ratio, fasting glucose and triglycerides, and hemoglobin A1c (Table). Interactions with treatment arms were explored. This model differs from others recently developed for the general population at risk for DM. There remains a need for external validation to determine predictive performance. In the future, use of a clinical prediction model to calculate individual 3-year DM risk specifically for women with prior GDM may improve decision making for prediabetes treatment in this high-risk group. Disclosure B. Man: None. A. Schwartz: None. Y. Xia: None. B.S. Gerber: None.