Fetal growth restriction and small for gestational age as predictors of neonatal morbidity: which growth nomogram to use?
Daisy Leon-Martinez,Lisbet S Lundsberg,Jennifer Culhane,Jun Zhang,Moeun Son,Uma M Reddy
DOI: https://doi.org/10.1016/j.ajog.2023.06.035
Abstract:Background: Fetal growth nomograms were developed to screen for fetal growth restriction and guide clinical care to improve perinatal outcomes; however, existing literature remains inconclusive regarding which nomogram is the gold standard. Objective: This study aimed to compare the ability of 4 commonly used nomograms (Hadlock, International Fetal and Newborn Growth Consortium for the 21st Century, Eunice Kennedy Shriver National Institute of Child Health and Human Development-unified standard, and World Health Organization fetal growth charts) and 1 institution-specific reference to predict small for gestational age and poor neonatal outcomes. Study design: This was a retrospective cohort study of all nonanomalous singleton pregnancies undergoing ultrasound at ≥20 weeks of gestation between 2013 and 2020 and delivering at a single academic center. Using random selection methods, the study sample was restricted to 1 pregnancy per patient and 1 ultrasound per pregnancy completed at ≥22 weeks of gestation. Fetal biometry data were used to calculate estimated fetal weight and percentiles according to the aforementioned 5 nomograms. Maternal and neonatal data were extracted from electronic medical records. Logistic regression was used to estimate the association between estimated fetal weight of <10th and <3rd percentiles compared with estimated fetal weight of 10th to 90th percentile as the reference group for small for gestational age and the neonatal composite outcomes (perinatal mortality, hypoxic-ischemic encephalopathy or seizures, respiratory morbidity, intraventricular hemorrhage, necrotizing enterocolitis, hyperbilirubinemia or hypoglycemia requiring neonatal intensive care unit admission, and retinopathy of prematurity). Receiver operating characteristic curve contrast estimation (primary analysis) and test characteristics were calculated for all nomograms and the prediction of small for gestational age and the neonatal composite outcomes. We restricted the sample to ultrasounds performed within 28 days of delivery; moreover, similar analyses were completed to assess the prediction of small for gestational age and neonatal composite outcomes. Results: Among 10,045 participants, the proportion of fetuses classified as <10th percentile varied across nomograms from 4.9% to 9.7%. Fetuses with an estimated fetal weight of <10th percentile had an increased risk of small for gestational age (odds ratio, 9.9 [95% confidence interval, 8.5-11.5] to 12.8 [95% confidence interval, 10.9-15.0]). In addition, the estimated fetal weight of <10th and <3rd percentile was associated with increased risk of the neonatal composite outcome (odds ratio, 2.4 [95% confidence interval, 2.0-2.8] to 3.5 [95% confidence interval, 2.9-4.3] and 5.7 [95% confidence interval, 4.5-7.2] to 8.8 [95% confidence interval, 6.6-11.8], respectively). The prediction of small for gestational age with an estimated fetal weight of <10th percentile had a positive likelihood ratio of 6.3 to 8.5 and an area under the curve of 0.62 to 0.67. Similarly, the prediction of the neonatal composite outcome with an estimated fetal weight of <10th percentile had a positive likelihood ratio of 2.1 to 3.1 and an area under the curve of 0.55 to 0.57. When analyses were restricted to ultrasound within 4 weeks of delivery, among fetuses with an estimated fetal weight of <10th percentile, the risk of small for gestational age increased across all nomograms (odds ratio, 16.7 [95% confidence interval, 12.6-22.3] to 25.1 [95% confidence interval, 17.0-37.0]), and prediction improved (positive likelihood ratio, 8.3-15.0; area under the curve, 0.69-0.75). Similarly, the risk of neonatal composite outcome increased (odds ratio, 3.2 [95% confidence interval, 2.4-4.2] to 5.2 [95% confidence interval, 3.8-7.2]), and prediction marginally improved (positive likelihood ratio, 2.4-4.1; area under the curve, 0.60-0.62). Importantly, the risk of both being small for gestational age and having the neonatal composite outcome further increased (odds ratio, 21.4 [95% confidence interval, 13.6-33.6] to 28.7 (95% confidence interval, 18.6-44.3]), and the prediction of concurrent small for gestational age and neonatal composite outcome greatly improved (positive likelihood ratio, 6.0-10.0; area under the curve, 0.80-0.83). Conclusion: In this large cohort, Hadlock, recent fetal growth nomograms, and a local population-derived fetal growth reference performed comparably in the prediction of small for gestational age and neonatal composite outcomes.