Validity of sonographic prediction of birth weight: A study of three algorithms in a cohort of healthy pregnant women of Yoruba descent in a suburb of Lagos state, Southwest Nigeria

Abstract

Background: Accurate estimation of fetal birth weight is critical in determining the delivery route and management of the neonate. Purpose of Study: The purpose is to determine the accuracy of Hadlock IV, Campbell, and Shepard's algorithm as predictors of birth weight in a cohort of fetuses of Yoruba descent. Materials and Methods: Fetal weight (FW) was estimated in a sample of 384 fetuses using Hadlock IV, Campbell, and Shepard's algorithm while actual birth weight (ABW) was measured. Receiver operating characteristic curves were plotted and used to determine the accuracy and sensitivity of each algorithm. Results: Most babies (84.6%) had normal estimated fetal weight (EFW) and ABW; mean FW = 3.2 ± 0.5 kg); 10% had low weight while 5.5% were macrosomic. While EFW correlated positively and strongly with ABW, the Hadlock IV algorithm had the strongest correlation (r = 0.978). The Hadlock IV, Campbell, and Shepard's algorithms had 92%, 72%, and 56% accuracy within the tenth centile, respectively. At 95% confidence interval, Hadlock IV was the most accurate predictor of normal birth and low birth weight (area under the curve [AUC] =0.91 and 0.94, respectively). Campbell was the most accurate predictor of macrosomia (AUC = 0.89). Conclusion: While Hadlock IV and Campbell algorithm are valid predictors, the Shepard model is a doubtful birth weight predictor among fetuses of Yoruba origin. When there is a need for absolute birth weight values, the Hadlock IV algorithm is preferred for suspected normal and low-weight babies while the Campbell model is preferred for fetuses weighing >4 kg among Yoruba fetuses.