ABSTRACT
We previously reported a logistic regression model for prediction of GDM from maternal characteristics and medical history in 75,161 singleton pregnancies. In this study of 1376 twin and 13,760 singleton pregnancies recruited at 11-13 weeks' gestation, we extend the model to include terms for twin pregnancies. We found the respective odds of GDM in dichorionic and monochorionic twin pregnancies to be 1.36 (95% CI: 1.02-1.81) and 2.78 (95% CI: 1.72-4.48) times higher than in singleton pregnancies. In both singleton and twin pregnancies, the risk for GDM increased with maternal age and weight and birth weight z-score of a baby in a previous pregnancy and is higher in women with a previous pregnancy complicated by GDM; in those with a first- or second-degree relative with diabetes mellitus; in women of Black, East Asian, and South Asian racial origin; and in pregnancies conceived through the use of ovulation-induction drugs. In singleton pregnancies, at 10% and 20% false-positive rate, the detection rate was 43% and 58%, respectively. In twin pregnancies, using risk cut-offs corresponding to 10% and 20% false-positive rates in singletons, the respective false-positive rates were 27% and 47%, and the detection rates were 63% and 81%.