Accuracy of Antenatal Ultrasound in Predicting Large for Gestational Age Babies: Population Based Cohort Study

Background Pregnancies with large for gestational age fetuses are at increased risk of adverse maternal and neonatal outcomes. There is uncertainty about how to manage birth in such pregnancies. Current guidelines recommend a discussion with women of the pros and cons of options including expectant management, induction of labour and caesarean birth. For women to be able to make an informed decision about birth, antenatal detection of large for gestational age is essential. Objective To investigate the ability of antenatal ultrasound scans to predict large for gestational age at birth. Study Design In this retrospective cohort study, we analysed data from a routinely collected database from the West Midlands, United Kingdom. We included pregnancies that had an antenatal ultrasound estimated fetal weight between 35+0 and 38+0 weeks gestation for any indication, as well as a subgroup where the reason for scan was that the fetus was suspected to be big. Large for gestational age was defined as >90th customized GROW centile, for estimated fetal weight as well as for neonatal weight. We also tested performance of an uncustomized standard, with Hadlock fetal weight >90th centile and neonatal weight >4kg. We calculated diagnostic characteristics for the whole population as well as groups with different maternal body mass indices. Results The study cohort consisted of 26,527 pregnancies which on average had a scan at 256 days gestation and delivered 20 days later at a median of 276 days (interquartile range 15). In total 2241 (8.4%) of neonates were large for gestational age by customized centiles, of which 1459 (65.1%) had a scan estimated fetal weight >90th centile, with a false positive rate of 8.6% and a positive predictive value of 41.0%. In the subgroup of 912 (3.4%) pregnancies scanned for a suspected large fetus, 293 (32.1%) babies were large for gestational age at birth, giving a positive predictive value of 50.3%, with sensitivity of 77.1% and false positive rate 36.0%. When comparing subgroups from low (<18.5kg/m2) to high body mass index (>30kg/m2), sensitivity increased from 55.6 to 67.8%, false positive rate from 5.2 to 11.5% and positive predictive value from 32.1 to 42.3%. A total of 2585 (9.7%) babies were macrosomic (birthweight >4kg), and of these 1058 (40.9%) were large for gestational age (>90th centile) antenatally by Hadlock’s growth standard, with a false positive rate of 4.9% and a positive predictive value 41.0%. Analysis within subgroups showed better performance by customized compared to uncustomized standards for low body mass index (<18.5; diagnostic odds ratio 23.0 vs 6.4) and high body mass index (>30; diagnostic odds ration 16.2 vs 8.8). Conclusion Late third trimester ultrasound estimation of fetal weight for any indication has good ability to identify and predict large for gestational age at birth, and improves with the use of a customised standard. Detection rate is better when ultrasound is performed for a suspected large fetus, however at the risk of higher false positive diagnosis. Our results provide information for women and clinicians to aid antenatal decision making about birth of a fetus suspected of being large for gestational age.