Ultrasonographic weight estimation in large for gestational age fetuses: a comparison of 17 sonographic formulas and four models algorithms.

Objective: To evaluate the accuracy of different formulas and role of fetal parameters (cephalic, abdominal, femur) used for estimation fetal weight (EFW) in large for gestational age (LGA) fetus in diabetic and non-diabetic mothers.

Methods: Seventeen formulas were assessed individually and clustered in four algorithms (X, Y, Z, W) on the basis of fetal biometric parameters using the mean absolute % error, standard deviation (SD), prediction within +/-5%, +/-10%, +/-15% of error and introducing new variable hypotenuse test (HPT) that can sum up precision and accuracy of formulas employed. For predicting fetal macrosomia (BW > or = 4200 g) a receiver-operating characteristic curve was constructed.

Results: Warsof2 formula showed the lowest mean % error, SD and HPT (p < 0.01) with overall prediction +/-5, +/-10%, +/-15% of birth weight in 68, 94 and 98%. The formulas that were only based on abdominal measurement (Warsof2, Hadlock1, Campbell) showed the best ability to identify fetal macrosomia. The X algorithm confirming primary role of abdominal circumference for EFW in diabetic mothers.

Conclusions: Accuracy of EFW in LGA fetuses is attributable to the biometric parameters used. Our findings show that the best formulas for EFW are those which only consider the abdominal measurements, especially in diabetic mothers. The new variable that we propose (HPT) confirms this result.