A comparison of single versus multiple growth ultrasonographic examinations in predicting birth weight

Our purpose was to determine whether two or more ultrasonographic fetal growth assessments provide a superior estimate of birth weight than does a single examination.

Five hundred and eighty five ultrasonographic procedures were performed in 263 patients, divided into single (n = 249) and multiple (n = 247) examination groups. Ultrasonographically estimated fetal weight percentiles and abdominal circumference percentiles were compared with gestationally corrected birth weight percentiles. After the gestational age range with the fewest errors in birth weight percentile prediction (32 to 36 weeks) was determined, patients with a single examination in this range were assigned to the single examination group. In the group with multiple examinations averaged ultrasonographic percentiles were used to predict birth weight percentile. Mean absolute and percentage errors were compared for predictive accuracy by means of analysis of variance and Student t test.

There was a linear correlation between the estimated fetal weight and abdominal circumference percentiles and the birth weight percentile, (r = 0.72, p < 0.0001). The accuracy of birth weight percentile predictions was similar whether one or multiple examinations were performed in the third trimester. Both the abdominal circumference percentile and estimated fetal weight percentile underpredicted birth weight, although the abdominal circumference percentile errors (1% to 2%) were statistically smaller than those derived from estimated fetal weights percentile (9% to 11%, p < 0.0001). Both abdominal circumference percentile and estimated fetal weights percentile consistently overidentified fetuses < 10th percentile (small for gestational age) and underidentified fetuses > 90th percentile (large for gestational age). However, multiple abdominal circumference percentile measurements resulted in improved predictions for small for gestational age (sensitivity 100%, specificity 88%) and large for gestational age (sensitivity 84%, specificity 100%).

With either the single or multiple examination approach birth weight percentile estimates were within 10% of the actual birth weight percentile approximately 50% of the time. Multiple ultrasonographic examinations provided little improvement in prediction of birth weight compared with a single observation. Multiple measurements of the abdominal circumference percentile may provide improved accuracy in identifying large for gestational age and small for gestational age fetuses.