Comparison of the accuracy of fetal weight estimation using clinical and sonographic methods

1. To compare the accuracy of clinical and sonographic estimations of fetal weight in pregnant women. 2. To determine the contributing factors that may affect the accuracy of these two methods.

Prospective study.

297 pregnant women who were admitted at labor room, Siriraj Hospital during the period of January 6 to February 26, 2004 were enrolled. The fetal weight was estimated clinically by the physicians, then blindly followed by sonographic estimation within 24 hours before delivery. The mean absolute error was calculated from the absolute of the difference between the estimated fetal weight and the actual birth weight of each method. Main outcome measurements were simple error, absolute error, absolute percentage error, and accuracy within 10% of actual birth weight.

The accuracy of clinical estimation of fetal weight was similar to sonographic estimation. The accuracy within 10% of both methods were 66.7 (95% CI 61.3, 72.0) and 65.3 (95% CI 60.1, 71.0), respectively. The estimation by both methods tend to be underestimated with the mean of absolute error 264.7 +/- 299.6 and 265.0 +/- 236.3 grams, respectively, and the mean of percentage error 9.0 +/- 9.7 and 8.6 +/- 6.9% of actual birth weight. The accuracy amongst possible contributing factors were compared and analyzed. The only one factor effect the accuracy significantly was actual birth weight < 2,500 grams in clinical estimation (P < 0.05). Sensitivity and specificity for prediction of birth weight lower than 2,500 grams was 82.6, 94.2% by clinical and 64.4, 97.6% by sonographic estimation. The positive predictive value and negative predictive value of both methods were 54.3, 98.5% and 82.9, 93.9%, respectively, while the efficacy was 93.3 and 92.6%.

Intrapartum clinical estimation of fetal weight was accurate as sonographic estimation, while the mean of error in grams or in percentage of birth weight were indifferent. The low-birth weight influenced the accuracy of clinical estimation significantly. However, clinical estimation is good enough for screening of the low-birth weight because of its high sensitivity and negative predictive value.