Objective: To compare clinical and sonographic estimates of birth weights with five new estimation techniques that involve measurements of soft tissue, for identifying newborns with birth weights of at least 4000 g.
Methods: Over 1 year, each woman at or after 36 weeks' gestation and suspected of having a macrosomic fetus had clinical and sonographic estimates of fetal weight (EFW) based on femur length (FL) and head and abdominal circumference, followed by five additional ways to identify excessive growth: cheek-to-cheek diameter, thigh soft tissue, ratio of thigh soft tissue to FL, upper arm subcutaneous tissue, and EFW derived from it. Areas (+/- standard error) of receiver operating characteristic (ROC) curves were calculated and compared with the area under the nondiagnostic line. P <.05 was considered statistically significant.
Results: Among 100 women recruited, 28 newborns weighed 4000 g or more. The areas under the ROC curves with clinical (0.72 +/- 0.06) and sonographic predictions using biometric characteristics (0.73 +/- 0.06) had the highest but similar accuracies (P.05). Three of the five newer methods (upper arm or thigh subcutaneous tissue and ratio of thigh subcutaneous tissue to FL) were poor diagnostic tests (range of areas under ROC 0.52 +/- 0.06 to 0.58 +/- 0.07). Estimated fetal weight based on upper arm soft tissue thickness and cheek-to-cheek diameter (areas 0.70 +/- 0.06 and 0.67 +/- 0.06, respectively) were not significantly better than clinical predictions (P.05) for detecting macrosomic fetuses. About 110 macrosomic and nonmacrosomic infants combined would be needed to have 80% power to detect a difference between ROC curves with areas of 0.58 (thigh subcutaneous tissue) and 0.72 (clinical estimate).
Conclusion: ROC curves indicated that measurements of soft tissue are not superior to clinical or sonographic predictions in identifying fetuses with weights of at least 4000 g.