Validation of a Prediction Model for Need of Neonatal Cardiac Surgery or Catheter-Based Therapy in Fetuses with Congenital Heart Defects

Abstract

INTRODUCTION: Prenatal diagnosis of congenital heart diseases allows timely in-utero referral to specialized centers and reduces the risk of morbidity and mortality. A model was developed to predict the need of neonatal cardiac surgery or catheter-based therapy in fetuses with congenital heart defects. With this model, based on fetal cardiac ultrasonographic findings, it would be possible to select patients in need of prenatal referral. OBJECTIVES: To validate such predictive model. METHODS: A prospective multicenter cohort validation study was conducted. Univariate and multivariate logistic regression analysis was performed, evaluating calibration by Hosmer-Lemeshow test and discrimination by area under the curve (AUC) of the receiver operating characteristic (ROC). RESULTS: In 58 (51.8%) of the 112 patients, neonatal cardiac surgery or catheter-based therapy was needed. There was no significant difference between observed and predicted event rates (p 0.232), and discrimination was satisfactory (AUC of the ROC 0.833; CI95%: 0.757-0.909). For a cut-off probability of 0.3 (that showed significant risk of need for neonatal cardiac invasive therapy in the original model), sensitivity was 96.6%, specificity 55.6%, positive predictive value 70% and negative predictive value 93.8%. CONCLUSIONS: The application of this predictive model in fetuses with congenital heart defects is feasible, shows a good diagnostic ability and can be a useful tool for perinatal management.