Reliability of sonographic fetal weight estimation in triplet pregnancies: a retrospective cohort study.

PURPOSE: To review our experience in ultrasound fetal weight estimation in our large population of triplet pregnancies.

METHODS: Ninety-seven triplet pregnancies were retrospectively included between January 2003 and January 2017. Sonographic fetal weight estimation using Hadlock's and Schild's formulas was compared to actual birth weight in a tertiary-care center in Vienna, Austria. Statistical analyses were performed using a stepwise linear regression model and crosstabs.

RESULTS: The median discrepancy between the sonographically estimated fetal weight by Hadlock's formula and the actual birth weight was 106 g (IQR 56-190). The percentage error and its standard deviation were - 2.5 ± 12.1%, and the median percentage error was - 3.6%. Concerning the use of Hadlock's formula, estimated fetal weight was the most important factor predictive of actual birth weight with an estimate of 0.920 (p < 0.001). Female neonates had been overestimated by a mean of 50.473 g per fetus. The sonographic prediction of small-for-gestational-age neonates was significantly reliable (p < 0.001), with positive and negative predictive values ranging from 81.3 to 100.0%. Similar results were obtained for Schild's formula.

CONCLUSION: Even if sonographically estimated fetal weight in triplet pregnancies has a high overall accuracy of fetal weight estimation, there are some limitations in prediction of intrauterine growth restrictions, especially in female fetuses.