Spinnato J A, Allen R D, Mendenhall H W
Department of Obstetrics and Gynecology, University of South Alabama Medical Center, Mobile.
Obstet Gynecol. 1988 Jun;71(6 Pt 1):893-8.
From 259 cases studied, 245 were selected in which a live-born infant was delivered within 35 days of a complete fetal ultrasound evaluation. Multiple linear regression using the least-squares method enable us to generate an equation that incorporated lapse time (examination-to-birth interval) with the natural logarithm of head circumference, femur length, and abdominal circumference to estimate birth weight. With a lapse time mean of 16 +/- 11 days and a range of zero to 35 days, the generated equation accurately predicted birth weight (R2 = 0.84; P less than or equal to .0001). For all birth weights, the mean error was -15 +/- 306 g, the percent mean error was 0.51 +/- 10.2%, and the mean absolute error of the estimate was 82 g/kg birth weight. This accuracy was maintained across the full range of lapse time observed. For examinations performed within one week of delivery (N = 71), this formula more accurately predicted birth weight than five existing static formulas tested. The accuracy observed during model development was confirmed during testing upon 167 non-model cases. The accurate prediction of birth weight from remote ultrasound data is possible when lapse time is included in the predicting equation. The clinical value of this model is suggested when ultrasound is unavailable or unreliable.
在259例研究病例中，选取了245例在完整的胎儿超声评估后35天内分娩出活产婴儿的病例。使用最小二乘法进行多元线性回归，使我们能够生成一个方程，该方程将时间间隔（检查至出生的间隔）与头围、股骨长度和腹围的自然对数相结合，以估计出生体重。时间间隔的平均值为16±11天，范围为0至35天，生成的方程准确地预测了出生体重（R2 = 0.84；P≤0.0001）。对于所有出生体重，平均误差为-15±306克，平均误差百分比为0.51±10.2%，估计的平均绝对误差为82克/千克出生体重。在观察到的整个时间间隔范围内，这种准确性都得以保持。对于在分娩后一周内进行的检查（N = 71），该公式比测试的五个现有的静态公式更准确地预测了出生体重。在模型开发过程中观察到的准确性在对167例非模型病例进行测试时得到了证实。当预测方程中包含时间间隔时，从远程超声数据准确预测出生体重是可能的。当超声检查不可用或不可靠时，该模型的临床价值得以体现。
