Comparison of eleven commonly used formulae for sonographic estimation of fetal weight in prediction of actual birth weight

Abstract

Objectives: The aim of the study is to compare the accuracy of 11 formulas in predicting fetal weight. Material and methods: The study includes 1072 pregnant women of gestational age from 28 to 42 weeks, who gave birth between January and June 2017. Pregnant women were divided into five groups; group 1, where actual birth weight (ABW) was less than 2500 g, group 2, where ABW was between 2500-4000 g, group 3, where ABW was above 4000 g. Group 4 - newborns with birth weight under 10 percentile and group 5 - newborns with birth weight above 90 percentile. The accuracy of the estimated fetal weight (EFW) was assessed by calculating absolute percentage error (APE) and 'limits-of-agreement'. R Spearman correlation was utilized between EFW and ABW. Results: The most accurate formula for group 1 is Hadlock3 (MAPE = 7.04%), the narrowest limits of agreement has Combs - [mean (SD): 99.41 g (269.57 g)]. For group 2, the lowest MAPE (5.43%) has Ott, the narrowest limits of agreement belongs to Combs - [mean (SD): -101.36 g (275.88 g)] . For group 3 is Hadlock3 (MAPE = 5.79%), the narrowest limits of agreement has Hadlock5 [mean (SD): -637.32 g (209.59 g)]. For group 4 is Combs (MAPE = 7.72%), the narrowest limits of agreement has Combs [mean (SD): 195.77 g (264.97 g)]. For gr oup 5 is Warsof2 (MAPE = 7.06%), the narrowest limits of agreement has Campbell [mean (SD): 227.81 g (299.26 g)]. Conclusions: Median of absolute percentage error is the most useful parameter to predict birth weight. Each group of fetuses needs different formula to predict the most accurate weight.