Estimation of fetal weight by means of ultrasound: a comparison of methods

An accurate prediction of birth weight during gestation can provide useful information for assessing fetal and newborn health status, enabling the clinician to better predict infant morbidity and mortality. Two previously reported standard methods for birth-weight estimation used data collected in utero to derive formulas by least-squares linear regression. The rationale for the inclusion of particular variables in these equations, however, has not been clearly defined. This study was undertaken to examine the efficacy of some previously used variables as well as some new variables in estimating fetal birth weight. The authors used measurements of femur length, biparietal diameter, and abdominal circumference from 107 fetuses (2500 to 4000 g) as variables to compare the two standard methods of birth-weight estimation. A new formula is presented that is derived from a simple model based on the known relationship between volume and weight. The head is represented as a sphere and the body as a cylinder. This study presents a more systematic approach to formula development in which statistical biases are minimized by examining the underlying distributions of the variables used to predict birth weight.