Menstrual age-dependent systematic error in sonographic fetal weight estimation: a mathematical model

PURPOSE: We used computer modeling techniques to evaluate the accuracy of different types of sonographic formulas for estimating fetal weight across the full range of clinically important menstrual ages. METHODS: Input data for the computer modeling techniques were derived from published British standards for normal distributions of sonographic biometric growth parameters and their correlation coefficients; these standards had been derived from fetal populations whose ages were determined using sonography. The accuracy of each of 10 formulas for estimating fetal weight was calculated by comparing the weight estimates obtained with these formulas in simulated populations with the weight estimates expected from birth weight data, from 24 weeks' menstrual age to term. Preterm weights were estimated by interpolation from term birth weights using sonographic growth curves. With an ideal formula, the median weight estimates at term should not differ from the population birth weight median. RESULTS: The simulated output sonographic values closely matched those of the original population. The accuracy of the fetal weight estimation differed by menstrual age and between various formulas. Most methods tended to overestimate fetal weight at term. Shepard's formula progressively overestimated weights from about 2% at 32 weeks to more than 15% at term. The accuracy of Combs's and Shinozuka's volumetric formulas varied least by menstrual age. Hadlock's formula underestimated preterm fetal weight by up to 7% and overestimated fetal weight at term by up to 5%. CONCLUSIONS: The accuracy of sonographic fetal weight estimation based on volumetric formulas is more consistent across menstrual ages than are other methods.