Expected probability of congenital heart disease and clinical utility of fetal echocardiography in pregnancies with pre-gestational diabetes

Objectives To establish a prediction model for fetal congenital heart disease (CHD) that could facilitate selective fetal echocardiography screening for diabetic women, and to provide cost analysis of selective versus global screening of these women. Study design A historical cohort study included women with pre-gestational diabetes who received perinatal care at Mayo Clinic, Minnesota from 2007 through 2013. Anatomical scans and fetal echocardiograms were reviewed and charts for postnatal outcomes were abstracted. Logistic regression models were utilized to identify predictors of CHD. Cost of global versus selective screening was estimated using cost per case detected. Results A cohort of 152 women was included. Abnormal anatomy scan was reported in five (3.3%). Among 51 (33.6%) who had fetal echocardiography, eight (15.7%) had abnormal findings. Postnatal echocardiography was conducted in 36 neonates, 12 had abnormalities. Only first trimester HbA1c was independent predictor with observed incidence of CHD of 1.8, 14.3, and 40% and HbA1c of <8, 8-10, and >10% respectively. Global screening strategy missed 22% of fetal cardiac abnormalities, while selective screening missed 33%. Conclusions According to our probability model, selective screening based on HbA1c and/or anatomic surveys seems to be an accurate and cost effective strategy for prenatal diagnosis of CHD among diabetic pregnancies.