Purpose: Fetal weight is an important predictive parameter of neonatal morbidity and mortality. Precise estimation of fetal weight is therefore a valuable item of information for further prenatal and obstetric management. Many regression formulae for sonographic fetal weight estimation have been published during the last 30 years, which, unfortunately, generally show poor rates of accuracy. The aim of this study was to compare commonly used formulae in different birth weight groups in order to assess whether any of the formulae are more or less favourable. The second purpose was to estimate the role of examiners on the accuracy of fetal weight estimation.
Materials and methods: 11 different formulae were evaluated in a group of 1941 pregnancies. Each fetus underwent ultrasound examination with complete biometric parameters within seven days before delivery. The assessment was carried out by either experienced or inexperienced sonographers.
Results: Over the whole weight range and in the subgroup of newborns with a birth weight less than 2500 g (n = 160), two Hadlock regression formulae (including abdominal circumference, femur length, head circumference, biparietal diameter, and abdominal circumference, femur length, biparietal diameter, respectively) showed the best levels of accuracy. Infants with a birth weight between 2500 and 3999 g (n = 1570) were best estimated using the gender-specific Schild formula. Macrosomic newborns (n = 211) were best evaluated using Merz's regression formula. The more skilled sonographers were able to achieve much higher intraclass correlation coefficients for all formulae.
Conclusion: Some regression formulae appear to be favourable within defined weight ranges. Accuracy of the formulae, however, is still unsatisfactory, and new formulae focusing on specific weight ranges (e. g., macrosomic fetuses) are needed. In addition, experience in obstetric ultrasound improves accuracy of fetal weight estimation.