Prediction of gestational diabetes mellitus using early oral glucose tolerance test

IntroductionGestational Diabetes Mellitus (GDM) is defined as diabetes first detected at the second or third trimester of pregnancy, excluding preexisting diabetes. We aimed to build a predictive model of GDM using booking oral glucose tolerance test (OGTT) values. Materials and MethodsSeventy-five healthy mothers who underwent 75g OGTT at 12-14 weeks and at 24-28 weeks were recruited. GDM was diagnosed at 28 weeks by cutoffs proposed by the Hyperglycemia and Adverse Pregnancy Outcomes study. Sensitivities and specificities for diagnosing GDM using different cut-offs for each of the three booking OGTT variables were measured. A series of multivariate binary logistic regression models were fitted using different combinations of the three booking OGTT variables. In-sample sensitivities and specificities for different cutoff probabilities of the models were calculated and Receiver Operating Characteristic (ROC) curves were constructed. The Area Under the Curve (AUC) of the ROC curve and the best cutoff value which maximized the sum of sensitivity and specificity of each model were computed. ResultsAUC of ROC curves for isolated fasting, 1 hour and 2 hour booking OGTT values for the prediction of GDM were 69.8%, 67.1% and 61.0% respectively. However, the logistic regression model with fasting and 1 hour booking OGTT values as predictors out-performed all other models with an AUC of 76.3%, in-sample sensitivity of 87.5% and a negative predictive value of 95.12%. ConclusionsThe future occurrence of GDM can be predicted utilizing a logistic model with fasting and 1 hour booking OGTT variables, which enables early identification and intervention.