Predictive value of a single early fetal weight estimate in normal pregnancies

To evaluate the accuracy of ultrasound-based fetal weight estimates made at 28-34 weeks of gestation in predicting small- and large-for-gestational-age infants (SGA, LGA) at term.

Two-hundred and fifty-nine patients with a healthy, singleton pregnancy in whom fetal biometry measurements were routinely performed between 28 and 34 weeks' gestation, were recruited at term delivery. The sonographic estimated fetal weight (EFW) and the birth weight were converted to percentiles on the basis of locally developed growth charts and compared. Multivariate linear stepwise regression analysis was used to predict the birth weight and birth weight percentile. The resulting equation (projectile formula) was used to determine the calculated birth weight, and that value was compared with the actual birth weight. The Bland and Altman plot and Passing and Bablok regression were used to compare between the calculated birth weight and the actual birth weight.

Mean gestational age at ultrasound examination was 32+/-1.6 weeks (28-34), and mean age at delivery was 39+/-1.7 weeks (37-42). The multivariate correlation between the calculated birth weight and the birth weight (R2 = 0.524) was higher than the correlation between the sonographic EFW and the birth weight (R2 = 0.083). Both the sonographic EFW and the calculated birth weight are characterized by low positive predictive values in predicting SGA or LGA infants. The calculated birth weight was more accurate in excluding SGA and LGA infants (negative predictive values of 99.5% and 100%, respectively). On method comparison tests, the calculated birth weight was not significantly different than the actual birth weight.

Fetal weight estimation at the early third trimester poorly predicts the birth weight centile at term. It remains uncertain, however, if it would be useful to use the calculated birth weight in pregnancies with clinically suspected SGA or LGA fetuses.