Optimal risk assessment of small-for-gestational-age fetuses using 31-34-week biometry in a low-risk population.

OBJECTIVE

To compare the performance of traditional growth charts for estimated fetal weight (EFW) and a validated pragmatic probabilistic approach using biometry at 31-34 weeks' gestation to screen for late pregnancy small-for-gestational age (SGA) fetuses in a low-risk population.

METHODS

Records of ultrasound biometry at 31-34 weeks were reviewed in 7755 consecutive low-risk women between 2002 and 2011. Fetal malformations, Doppler anomalies and preterm delivery before 37 weeks were excluded. SGA was defined by various percentile cut-offs of birth weight. The probability of SGA was modeled as a function of Z-scores of femur length, abdominal circumference and head circumference. The model was validated on a second independent dataset of 1725 pregnancies from a different screening unit. The screening performance of this probabilistic approach was compared with those of traditional EFW growth charts. The additional value of factoring in maternal characteristics was also ascertained.

RESULTS

Using national birth-weight charts, the proportions of newborns at 37-42 weeks with birth weight<3(rd) , <5(th) and<10(th) centiles were 3%, 6% and 12%, respectively, and there was a 2% rate of birth weight<2500 g. For a 10% false-positive rate, a direct probabilistic approach yielded a 51% detection rate of neonates with birth weight<10(th) centile, compared to the 32% and 48% detection rates given by the 10(th) centile cut-off of two reference charts for EFW. Adding maternal characteristics significantly improved detection rate by 2% to 53%.

CONCLUSIONS

The suggested validated approach to screening for late SGA fetuses outperforms traditional approaches using growth charts. By adding maternal characteristics, this screening method offers a favorable alternative to customized charts.