Does the availability of maternal HbA1c results improve the accuracy of sonographic diagnosis of macrosomia?

Abstract

To determine whether measuring maternal glycosylated hemoglobin (HbA1c) can improve the accuracy of sonographic estimation of fetal macrosomia. Sonographic estimation of fetal weight (EFW) and maternal HbA1c were obtained in term, non-diabetic patients within 1 week before delivery. Neonatal birth weights were recorded at delivery and compared with both sonographic estimations and HbA1c. Macrosomia was defined as birth weight of >or=4000 g. The absolute error of the sonographic EFW was calculated. Receiver-operating characteristics (ROC) curve analysis was used to evaluate sonographic EFW and HbA1c as predictors of birth weight >or=4000 g. Variables were tested using regression analysis and student's t-test. One hundred and sixty two patients were evaluated between July and December 2002. Twenty-eight patients (17.3%) delivered macrosomic infants. Sonographic EFW >or=4000 g predicted macrosomia with sensitivity, specificity and positive and negative predictive values of 66.6%, 88.8%, 54.5% and 93.0%, respectively. Its overall accuracy was 85.5%. The area under the ROC curve of sonographic EFW in the prediction of macrosomia was 0.9 (P < 0.001). HbA1c levels in women delivering macrosomic and non-macrosomic neonates were 5.3 +/- 0.7% and 5.2 +/- 0.5%, respectively (P = 0.27). The area under the ROC curve of HbA1c in the prediction of macrosomia was 0.53 (P = 0.27). Maternal HbA1c is not a useful test in the prediction of birth weight. It therefore cannot be used to improve the accuracy of sonographic EFW.