Objectives: To derive a formula for sonographic estimated fetal weight
(EFW) based on a two-compartment model of fetal volume and to test it
against two widely used formulas, especially at the extremes of fetal
weight for which existing formulas are generally inaccurate. Methods:
We analyzed 865 consecutive sonograms that met the following inclusio
n criteria: singleton pregnancy, normal anatomy, delivery within 3 day
s of sonography, and measurements of biparietal diameter (BPD), head c
ircumference (HC), abdominal circumference (AC), and femur length (FL)
. The weight of the fetal head was modeled to be proportional to HC3,
and the weight of the trunk proportional to AC2 x FL. The proportional
ity constants were found by multiple linear regression on 380 sonogram
s performed in 1990 (the ''derivation set''). The new formula was test
ed for accuracy of prediction of actual birth weight against the formu
las of Hadlock et al and Shepard et al using 485 sonograms from 1991-1
992 (the ''validation set''). Results: In the derivation set, the form
ula EFW = (0.23718 x AC2 x FL) + (0.03312 x HC3) was fit; the correlat
ion with actual birth weight had an r value of 0.996. In the validatio
n set, the new formula produced smaller systematic errors and smaller
absolute errors than either the Hadlock or Shepard formula both overal
l and in fetal weight strata from less than 1000 g to over 4000 g. Con
clusion: The new formula makes geometric sense and provides accurate e
stimates of fetal weight across a broad range of weights.