Six published fetal weight estimating regression models proposed for clinical use were evaluated in 259 pregnant women who delivered within 72 h of an ultrasound evaluation performed with sector scanner. The patient sample included 89 (33.2%) fetal weights that were below the 10th or above the 90th percentile for menstrual age. The actual mean percent error (systematic error), standard deviation (random error), and the number of large errors of prediction for all equations were greatest in fetuses that were small- and large-for-gestational age. Whereas there were no significant differences between equations for the patient sample as a whole, equation AC,BPD (Shepard) had the smallest systematic error in intrauterine growth retarded, premature, and normal-term fetuses less than 4000 g. Conversely, the systematic error of the models that included femur length was smallest at the upper end of the weight scale and in macrosomic fetuses in general. In that regard, the accuracy of fetal weight prediction could be increased by selecting the appropriate model for the proper clinical indications. Although these findings can be explained by the limitations of the current regression models in estimating fetal soft tissue mass, a subtle effect of the use of the sector scanner on the results of this study cannot be completely excluded and requires further investigation.

No abstract