Accuracy and modifying factors of the sonographic estimation of fetal weight in a high-risk population

There have been a number of reviews assessing the accuracy of different met hods of sonographic estimation of fetal weight, without identifying any cle arly superior equation. In order to optimise accuracy in a high-risk popula tion, we decided to compare some of the most popular early equations with t he newer volume-based equations, and to try and identify factors that affec t the ability of these equations to estimate fetal weight accurately. We collected the scan and delivery details of 192 fetuses born within one w eek of a sonographic estimation of fetal weight. We then applied three of t he most popular equations and two newer volume-based equations to the recor ded fetal biometric parameters to assess the performance of each equation o verall, and under varying maternal, fetal, and scan conditions. The equations of Shepard, Hadlock A, Hadlock B and Combs produced similar r esults with systematic (mean) errors in the range 1.2-1.9% and random error characterised by one standard deviation in the range of 8.6-9.5%. Dudley's volume-based equation produced a significant systematic error in the form of a mean error of 7.4%, which corresponds to a mean birthweight (BW) which is 7.4% above the mean estimated fetal weight (EFW). When we stratified the study group by birthweight, Combs' equation produced significant differences in the mean error, (p < 0.00001), that ranged from a mean overestimation in fetal weight of 8.5% for babies with BW < 1000 g to a mean underestimation in fetal weight of 6.2% for babies with BW > 3000 g. Oligohydramnios resulted in a trend towards an increased mean error for all equations which was only statistically significant for Hadlock B. The equations Shepard and Hadlock A performed best in our high-risk populat ion. They produced the smallest systematic errors across the entire study g roup and were not adversely affected by variations in birthweight, liquor v olume, or fetal presentation. The newer, volume-based equations were disapp ointing, producing large systematic errors. Large random errors in all equa tions continue to be the Achilles' heel that limit the value of sonographic EFW.