Evaluation of the Accuracy of Fetal Weight Prediction Technology in Predicting Fetal Macrosomia

Background: Pregnancies complicated by the presence of fetal macrosomia, or birth weight > 4000g, are associated with increased risk of adverse neonatal and maternal outcomes. The ability to identify macrosomic fetuses prior to birth can help guide counseling regarding these complications and options for delivery. Current guidelines recommend shared decision-making regarding options for delivery planning in these pregnancies, including expectant management, induction of labor, and cesarean delivery. In June of 2024, the obstetrical ultrasound software used at the University of Louisville adopted the use of a fetal birth weight prediction technology, using parameters used in standard growth ultrasounds and extrapolating data to predict what a fetus’s weight would be at 39 weeks gestation. The ability to predict a fetuses weight at 39 weeks earlier on in pregnancy would be a crucial tool for providers to start conversations with patients regarding delivery planning earlier. Prior research has focused on evaluating parameters such as estimated fetal weight or abdominal circumference on the day of the ultrasound, and evaluating whether or not this finding is associated with adverse outcomes such as fetal macrosomia. There is a lack of research examining the accuracy of fetal weight predictions compared with actual delivery weights or the association of these predicted weights with adverse outcomes. Objective: To investigate the ability of sonographic fetal weight prediction technology to predict the presence of fetal macrosomia at birth. Methods/Design: This was a retrospective chart review examining patients who had growth ultrasounds with predicted fetal weight at 39 weeks >4000g between 6/1/2024 and 12/31/2024. The primary outcome of the study was the discrepancy between predicted fetal weight at 39 weeks and actual birth weight. Secondary outcomes included gestational age at delivery, latency between ultrasound and delivery, maternal BMI, presence of maternal diabetes or delivery complications such as shoulder dystocia, postpartum hemorrhage, or OASIS injury. Patients were also stratified by predicted birth weight (4000g– 4499g; 4500g-4999g; 5000g or greater) to assess differences in the ability of ultrasound to predict fetal macrosomia with different predicted fetal weights. Results: Out of 155 patients included in the study, the average maternal age was 32.1 years (+/- 5.3 years). 35% of patients carried a diagnosis of diabetes, and average maternal BMI was 35.6 (+/- 7.1). The average gestational age at delivery was 38.1 weeks (+/- 1.3 weeks). Average latency between ultrasound date and delivery date was 22.8 days (+/- 13.5 days). 63% of patients in the entire cohort had Cesarean deliveries, with repeat Cesareans accounting for 35% of the cohort. 37% of the cohort had vaginal deliveries. The rate of shoulder dystocia among patients who had a vaginal delivery was 4%. The rate of OASIS injury among patients who had a vaginal delivery was 2%. The rate of PPH was 15%. Out of the full cohort of patients, the average predicted EFW at 39 weeks was 4.35 kg (+/- 0.32 kg). The average actual birth weight was 3.79 kg (+/- 0.48 kg). The average over-prediction error for the full cohort was therefore 562g (+/- 421g). 32% of babies in the full cohort were born at or above 4000g. When examining only patients who delivered between 38+0 and 40+0 (within one week of 39 weeks), out of 76 patients, the over-prediction error was smaller at 379g (+/- 331g). The average predicted birth weight at 39 weeks in this subgroup was 4.29kg (+/- 0.25kg) and the average actual birth weight in this subgroup was 3.91kg (+/- 0.37kg). 36% of babies in this sub-group were born macrosomic. When stratifying this subgroup further into predicted weights at 39 weeks of 4000-4499, 4500-4999, or over 5000g, the over-prediction error increased as the predicted weight category increased. The over-prediction error was 346g (+/- 317g) in the 4000-4499g subgroup; 465g (+/- 318g) in the 4500-4999g subgroup, and 1223g in the 5000+g subgroup, with a p-value of 0.021. No significant differences in over-prediction error were seen when stratifying by maternal BMI, maternal age, or latency between ultrasound date and delivery date. Conclusion: Among the entire cohort of patients, only 32% of babies that had a predicted EFW at 39 weeks above 4000g were born macrosomic. Among the subgroup who delivered within one week of 39 weeks, 36% of babies were born macrosomic. This is consistent with findings from another study examining third trimester ultrasounds showing EFW > 90th percentile which demonstrated a positive predictive value of 41% for the presence of fetal macrosomia at birth. This indicates that fetal birth weight prediction technology may be of similar accuracy to third trimester growth ultrasounds, and therefore that predictions from earlier ultrasounds may be used to guide counseling. There was a statistically significant increase in the over-prediction error as the predicted birth weight at 39 weeks increased. There was no significant difference in over-prediction error seen when stratifying for maternal BMI, maternal age, or latency between ultrasound date and delivery date.