Ultrasonographic estimation of fetal weight in the clinically macrosomic fetus.

The purpose of this study is to evaluate models for the prediction of birth weight in fetuses suspected of being macrosomic. A total of 141 patients with standard measurements of the head, abdomen, and femur were studied. Linear regressions were performed with single parameters, squares, and all possible cross products in the generation of models with log (birth weight) and birth weight as dependent variables. These models were then compared with a group of previously published equations. Clinically, all models performed poorly. However, two models were significantly less accurate in the prediction of birth weight (p less than or equal to 0.05). The best results were obtained by equations that used abdominal circumference and femur length measurements. There was no improvement in models that contained log (birth weight) or birth weight as dependent variables or models with complex variables such as squares or cross products of measured parameters. In conclusion, when evaluating patients at risk of macrosomia, the best estimates of fetal weight can be obtained by models that contain abdominal circumference and femur length.