Prediction of gestational diabetes at early pregnancy by using a clinical prediction model

Objectives:

The oral glucose tolerance test (OGTT) is recommended to rule out gestational diabetes mellitus (GDM) at 24+0 to 27+6 weeks of gestation. However, it is an ongoing debate how to reduce the number of invasive and expensive OGTT examinations in the clinical setting of pregnancy care. This study aims to assess the predictive accuracy of a previously defined risk estimation model (Göbl 2012) at a very early stage of pregnancy.

Methods:

A total of 112 pregnant women were consecutively recruited from our outpatient department (Department of Obstetrics and Gynecology, Division of Obstetrics and Feto-maternal Medicine, Medical University of Vienna) and received a broad risk evaluation before 15+6 weeks of gestation, including maternal age, ethnicity, preconceptional disease (e.g. dyslipidaemia), body mass index (BMI), fasting plasma glucose, amount of glucosuria, and family history of diabetes. GDM was diagnosed by using the standard OGTT (24+0 to 27+6 weeks of gestation) or if insulin treatment was necessary.

Results:

Of 112 pregnant women included in this study 29 (26%) developed GDM. Our previously defined risk estimation model including fasting glucose as well as simple anamnestic and clinical variables showed a fair accuracy to detect the progression to GDM already at early gestation with an area under the receiver operating characteristic curve (ROC-AUC) of 0.72 (95% CI: 0.60 – 0.84). The test accuracy of our model was superior as compared to the ROC-AUC of actual (0.69, 95% CI: 0.57 – 0.81) or preconceptional BMI (0.64, 95% CI: 0.52 – 0.76). Our previously defined cut-off (score level ≥0.2) showed a high sensitivity (93%), and was thus able to rule out GDM progression with a negative predictive value of 83%, however, lacked in specificity as well.

Conclusion:

Our previously defined risk estimation model, which was originally developed at second and third trimester of pregnancy provides a fair predictive accuracy when applied at early gestation. The recruitment of our study is in progress to identify other candidate predictors (including laboratory measurements and dietary patterns) to provide an optimized diagnostic algorithm.

No conflict of interest has been declared by the author(s).