To detect possible sources of bias in ultrasound prediction of gestational age. Using the Swedish Medical Birth Registry, 571 617 women were identified who were delivered between 1990 and 1997 and who had obtainable information on last menstrual period and expected date of delivery according to ultrasound. Male fetuses were more likely than females to be judged older than the last menstrual period date suggested at early fetometry. The estimated magnitude of the systematic error by infant gender corresponded to 1.5 days. Similarly, the fetuses of young women, multiparous women, smokers and women with low educational level were at increased risk of being smaller than expected at ultrasound examination in early pregnancy. A strong association was seen between adjustments of expected date of delivery -7 days or more and small-for-gestational age according to ultrasound at birth. Compared to singleton pregnancies, twin pregnancies were more likely to be judged more progressed at ultrasound fetometry than the last menstrual period date suggested. Compared to last menstrual period estimates, routine ultrasound measurements to predict date of delivery are comparatively reliable but systematic errors are inherent in the method. The erroneous adjusted dates may be due to incorrect measurements or systematic bias (e.g. gender), but they are also likely to reflect early growth restriction (e.g. in the case of maternal smoking and small-for-gestational age). Further studies are needed to investigate whether the systematic errors in ultrasound prediction of gestational age could lead to suboptimal obstetric management in adjusted pregnancies.
Read full abstract