A risk-prediction model using parameters of maternal body composition to identify gestational diabetes mellitus in early pregnancy

Accurate early risk-prediction for gestational diabetes mellitus (GDM) would target intervention and prevention in women at the highest risk. We evaluated maternal risk-factors and parameters of body-composition to develop a prediction model for GDM in early gestation. A prospective observational study was undertaken. Pregnant women aged between 18 and 50 y of age with gestational age between 10 and 16 weeks were included in the study. Women aged ≤18 y, twin-pregnancies, known foetal anomaly or pre-existing condition affecting oedema status were excluded. 8-point-skinfold thickness (SFT), mid-upper-arm-circumference (MUAC), waist, hip, weight and ultrasound measurements of subcutaneous (SAT) and visceral abdominal-adipose (VAT) were measured. Oral-glucose-tolerance-test (OGTT) for GDM diagnosis was undertaken at 28 weeks gestation. Binomial logistic-regression models were used to predict GDM. ROC-analysis determined discrimination and concordance of model and individual variables. 188 women underwent OGTT at ~28 weeks gestation. 20 women developed GDM. BMI (24.7 kg m−2 (±6.1), 29.9 kg m−2 (±7.8), p = 0.022), abdominal SAT(1.32 cm (CI 1.31, 1.53), 1.99 cm (CI 1.64, 2.31), p = 0.027), abdominal VAT(0.78 cm (CI 0.8, 0.96), 1.41 cm (CI 1.11, 1.65), p = 0.002), truncal SFT (84.8 mm (CI 88.2, 101.6), 130.4 mm (CI 105.1, 140.1), p = 0.010), waist (79.8 cm (CI 80.3, 84.1), 90.3 cm (CI 85.9, 96.2), p = 0.006) and gluteal hip (94.3 cm (CI 93.9, 98.0), 108.6 cm (CI 99.9, 111.6), p = 0.023) were higher in GDM vs. non-GDM. After screening variables for inclusion into the multivariate model, family history of diabetes, previous perinatal death, overall insulin resistant condition, abdominal SAT and VAT, 8-point SFT, MUAC and weight were included. The combined multivariate prediction model achieved an excellent level of discrimination, with an AUC of 0.860 (CI 0.774, 0.945) for GDM. An early gestation risk prediction model, incorporating known risk-factors, and parameters of body-composition, accurately identify pregnant women in their first-trimester who developed GDM later on in gestation. This methodology could be used clinically to identify at-risk pregnancies, and target specific treatment through referred services to those mothers who would most benefit.