Risk assessment for preeclampsia in women with gestational diabetes mellitus

Aim: To develop a clinical model to assess the risk of preeclampsia in women with gestational diabetes mellitus (GDM). Methods: We studied clinical characteristics of 813 consecutive women who had GDM between January 2003 and February 2008 at our institution. The clinical features which were significantly associated with preeclampsia by multiple logistic regression analysis were integrated into a risk model. The diagnostic performance of this model was then evaluated from a receiver-operating characteristic (ROC) curve analysis. Results: A total of 78 women with GDM (9.6%) developed preeclampsia. In a multivariable analysis, first-trimester BMI >= 27 kg/m(2) (P<0.001), GDM diagnosed within 20 weeks of gestation (P<0.001), and poor glycemic control (P<0.001) were associated with preeclampsia. These three factors were incorporated into a risk-scoring model which ranged from 0 to 3 points. At the optimal cut-off score of >= 2, the sensitivity, specificity, and area under the curve (AUC) for preeclampsia were high at 76.9% [95% confidence interval (CI) 69.0-85.2], 92.8% (95% CI 85.9-98.1), and 0.849 (95% CI 0.792-0.905), respectively. Conclusion: A model based on clinical data yielded predicted the development of preeclampsia in women with GDM.