Fetal head and abdominal circumferences: I. Evaluation of measurement errors.

Several sources of error associated with the measurement of fetal head and abdominal circumferences (HC, AC) were examined using dynamic image ultrasound. Statistically significant systematic intraobserver errors were not seen unless technical differences in the images were present (HC only) and the range for random errors was 4-5% (2 SD) for both measurements. Although differences in the procedures used to acquire images for measurement had some effect on accuracy (HC only), the primary source of variations in measurements made by different investigators was the measurement methods. Different measuring devices gave different results and different investigators used the same devices differently. Under optimal measuring conditions, the interobserver error for experienced investigators was 1.2 (+/-4.3 SD)% for HC and 2.4 (+/-1.6 SD)% for AC. Comparison of prenatal and postnatal measurements indicated that there were no systematic differences for HC but for AC, the prenatal measurements were 6% greater. Random differences were +/-6% (2 SD) for HC and +/-13% (2 SD) for AC.