Ultrasound dating and birth weight at term

EDITOR, - Tine Brink Henriksen and Allen Wilcox1 comment on our finding that the birthweight curve in our study was steeper at term than that observed in other studies.2 We ascribed this to our use of dates based on routine ultrasound scans instead of menstrual dates to derive our weight for gestational age standard.2 They suggest that a constitutionally large fetu's gestational age is overestimated by the dating scan and that therefore bigger babies, with presumably larger birth weights at term, would selectively be plotted against later gestations, thus creating bias.

While it is true that the study they refer to showed variation in biparietal diameter between small (<=10th centile) and large (>=90th centile) babies,3 the difference between these extremes amounted to only about 1.5 mm at 18 weeks (figure 2 in reference 3). This difference is equivalent to less than one day's variation in gestational age 4 and would translate at term to a difference in birth weight of under 20 g. In contrast, our mean birth weight at 40 weeks (3522 g) is about 120 g higher, at 41 weeks (3652 g) 167 g higher, and at 42 weeks (3750 g) 260 g higher than previous British birthweight standards, which are based on menstrual dates and show considerable flattening at term.2

We have since examined our accuracy of gestational dating by ultrasound measurement of the fetal biparietal diameter using Campbell and Newman's dating tables.4 We studied 19 consecutive singleton, normal term pregnancies resulting from assisted conception by in vitro fertilisation or intrauterine insemination booked at our hospital. All had the usual detailed ultrasound examination at 18-19 weeks by ultrasonographers who were not aware of the precise gestational age at the time of measurement. Gestational age derived from the biparietal diameter4 was compared with the true gestational age as determined from the date of fertilisation. The mean error (age estimated on ultrasonography minus true gestational age) was -0.57 days, with a standard deviation of 2.96 days and normal distribution.

We then looked at the possible effect of constitutional variation shown by the measurement obtained from the dating scan on the subsequent birth weight. There was no significant correlation between biparietal diameter centiles, adjusted for gestational age, and birth weight for gestation centiles (r=0.256; P=0.37). Furthermore, there was no significant difference in average birth weight for gestation centile between the two groups of babies that had the largest and smallest biparietal diameters (average of centiles, 47 v 40).

Thus the 95% confidence interval for ultrasound dating by biparietal diameter is -5.9 to 5.9 days in our unit. In contrast, as we previously reported,5 “certain” menstrual dates (routine ultrasound dates being used as a reference) have a heavily skewed distribution of error of -9 to 27 days (95% confidence interval). This inclination for menstrual dates to overestimate the true dates can led to birth weights being plotted against gestational ages that are too advanced. We believe this to be the reason for an artificial flattening of the birthweight curve, emphasing the need for dates of determined on the basis of ultrasound scans to be used to derive accurate birthweight standards.