217: Validation of the Delphi Procedure Consensus criteria for defining Fetal Growth Restriction (FGR)

Abstract

Recently a Delphi procedure was used to establish new criteria for defining FGR (Ultrasound Obstet Gynecol. 2016;48:333-9). These criteria still require clinical validation. We sought to validate the recommended criteria by comparing the Delphi criteria (Delphi FGR) with our current standard (an estimated fetal weight (EFW) <10th percentile by Hadlock’s standard (Current FGR); in predicting adverse neonatal outcomes (ANO). This is a prospective cohort study of women presenting for fetal growth assessment between 26 and 36 weeks gestation (GA). Delphi FGR criteria include: two solitary parameters (abdominal circumference (AC) or EFW < 3rd %) and four contributory parameters (EFW or AC < 10th) centile; AC or EFW crossing centiles by > two quartiles on growth charts and cerebroplacental ratio < 5th % or UA-PI > 95th %. The primary outcome was the prediction of a composite of ANO including one or more of: neonatal intensive care unit admissions, cord pH less than 7.1, 5 minute Apgar <7, respiratory distress syndrome, intraventricular hemorrhage, neonatal seizures or death. The discriminatory ability of each criteria were compared using area under receiver operating curves (AUC). The sensitivity, specificity and predictive values for the definitions of FGR were also compared. From 1055 women included in the study, ANO was seen in 139 (13.2%). The cohort had only 2 cases of early FGR (FGR <32 weeks), therefore the study focused on late FGR; delivering after 33 weeks and had EFW performed within 3 weeks of delivery. Current FGR had an association with ANO that was not statistically significant: OR and 95% CI of 1.1, 0.5-2.0; compared with the Delphi FGR: OR 2.1, 95% CI: 1.0-4.1. The area under the ROC curve (AUC) for the current criteria was 0.50, 95% CI (0.48-0.53) and using Delphi FGR was 0.66, 95% CI (0.54-0.76). The Table shows screening characteristics using these definitions of FGR. If the false positive rate is fixed at 10%, the sensitivity for Delphi FGR is: 40%. The specificity and positive predictive values of the new consensus definitions were higher. Example of ROC curve using Delphi FGR is shown (Figure). Although the two definitions have poor sensitivity, the new criteria for defining FGR have stronger association with and are more specific for predicting adverse neonatal outcomes. Our findings support adopting the Consensus criteria, but larger studies are needed to validate the definition for early FGR and long-term infant outcomes.View Large Image Figure ViewerDownload Hi-res image Download (PPT)