[Developing a risk appraisal model for gestational diabetes mellitus]

Abstract

Objective: To identify risk factors associated with gestational diabetes mellitus (GDM) and establish a risk appraisal model for GDM.

Methods: A case-control study of 156 women with GDM and 362 pregnant women without GDM was undertaken. Non-conditional logistic regression analysis was performed to identify risk factors associated with GDM and establish a prediction model. The diagnostic efficiency of this model was estimated by receiver operator characteristic curve (ROC) and Hosmer-lemeshow goodness-of-fit.

Results: Age, high blood pressure, history of poor pregnant outcome, obesity before pregnancy, excessive consumption of fruits, history of DM of immediate family members and taste preference were identified as risk factors of GDM. High consumption of white meat was identified as a protective factor. When the threshold value was set at 3.6, maximum sensitivity (73.7%) and specificity (68.5%) was obtained. The area under curve (AUC) of ROC was 0.79, with Hosmer-lemeshow P = 0.97.

Conclusion: The developed model has high diagnostic efficiency, which provides a good instrument for screening women with high-risk GDM.