Accuracy of ultrasound estimation of fetal weight in twin pregnancy

Objective

To determine the accuracy of formulas for estimation of fetal weight in twin pregnancy.

Methods

Inclusion criteria from the ESPRiT twin cohort were twin pregnancies that resulted in live-born twins without congenital anomalies or twin-twin transfusion syndrome, ultrasound examination within 3 days of delivery, birth weight (BW) > 500 g and gestational age > 24 weeks. A total of 63 formulas using various combinations of abdominal circumference (AC), femur length (FL), biparietal diameter (BPD) and head circumference (HC) for the estimation of fetal weight (EFW), were compared to BW for accuracy in 226 twin pregnancies/452 fetuses.

Results

Using median percentage error (MPE), the most accurate formulas were the Hadlock formulas (1984, 1985 with MPE < 1%) incorporating AC, HC, and FL. The INTERGROWTH-21st formula, which incorporates AC and HC, marginally underestimated BW (MPE −3.7%). The Hadlock formulas were more likely to overestimate BW than underestimate it. Using small-for-gestational (SGA, EFW < 5th centile), the sensitivity for detection of low BW (BW < 5th centile) was 92%, 81% and 89% for the Hadlock 1984, Hadlock 1985 and INTERGROWTH-21st formulas, respectively. SGA prevalence was noted to be higher for the INTERGROWTH-21st formula for fetal factors.

Conclusion

Among 63 proposed formulas for EFW, the Hadlock formulas incorporating HC, AC and FL were the most accurate in twins. However, the INTERGROWTH-21st standard, which incorporates AC and HC, did not show lower detection rates for LBW. The frequency and intensity of twin ultrasound surveillance is such that restriction of ultrasound examination to two parameters in ultrasound could be recommended without diminution of accuracy.