Ultrasound prediction of large fetuses. Epidemiological and clinical investigations

Sammanfattning: The proportions of newborns with a birth weight (BW) >4,000 g and of macrosomic newborns (BW >4,500 g) have increased during the last two decades, parallel with an increasing maternal pre-pregnancy body mass index (BMI) and age at the time for pregnancy. Delivery of a large fetus, especially >4,500 g might cause perinatal complications for both the mother and her child. The aims of this work were to investigate the detection rate of large-for-gestational age (LGA) term newborns by using a routine fetal two-dimensional (2D) ultrasound examination for fetal growth in the third trimester of pregnancy, and to examine whether the detection rate could be further improved by including maternal pre-pregnancy and pregnancy-related variables to the estimated fetal weight (FW) by developing a prediction formula for risk calculation of LGA newborn. Furthermore, to investigate if the accuracy of BW prediction in prolonged pregnancies and in pregnancies with suspected large fetuses could be further improved by using three-dimensional (3D) ultrasound technique with volumetry of fetal structures. Material and methods. A population-based perinatal register, Perinatal Revision South, was used to identify term singleton pregnancies with a routine ultrasound examination in the third trimester of pregnancy from 1995 through 2009. The difference between the BW z-score and the FW z-score at the ultrasound examination in the third trimester, divided by the time elapsed between ultrasound examination and birth was assessed for each fetus. Maternal variables were evaluated for a possible impact on the third trimester fetal growth using multivariate linear and polynomial regression analyses. In order to develop a prediction model for risk calculation of LGA term newborn the dataset (n=48,809) was divided into a development sample and a validation sample. The development sample was used to identify maternal characteristics associated to LGA using multiple logistic regression analyses. The obtained odds ratios were converted to likelihood ratios and included in a prediction model based on Bayesian theorem for risk calculation of LGA newborn. The prediction model was tested on the validation sample. For the prospective comparative studies, pregnant women >286 days of gestation (n=176) and pregnant women with a fetus estimated to be LGA at the third trimester routine ultrasound examination (n=114) were included. 2D and 3D FW estimation formulas known from the literature were used. Mean percentage error (MPE), absolute MPE, receiver operating characteristic (ROC) curves and the area under the curve were used for comparison of the accuracy in BW prediction by the various formulas close to birth. Results. The results showed that maternal pre-pregnancy variables affected third trimester fetal growth. Increasing maternal BMI and body length, and pre-existing diabetes mellitus influenced fetal growth positively, while heavy smoking affected fetal growth negatively. A good detection rate of LGA term newborns was found when using a routine fetal ultrasound examination for fetal growth control in the third trimester of pregnancy. The detection rate could be further improved by adding maternal variables associated with LGA term newborns to the ultrasonically estimated FW using a prediction model based on the Bayesian theorem. The most critical subgroup of infants with BW >4,500 g was more accurately predicted using 3D ultrasound technique including volumetry of fetal thigh and abdomen, compared to the conventional 2D ultrasound technique, with or without maternal body weight included. At an estimated FW >4,300 g, using Lindell and Maršál formula, the detection rate was 93 %, while the false positive rate was 36 %, which was close to the most optimal and clinically acceptable relation between the detection rate and false positive rate illustrated by the ROC curve. In prolonged pregnancies with a wide range of BWs, no significant differences in BW prediction close to birth were found between the 3D and 2D formulas. Conclusions. An antenatal detection of LGA/macrosomic term newborns might minimize maternal and fetal perinatal complications due to delivery of a large fetus. The prediction can be improved by using a model utilizing the Bayesian theorem including the estimated FW at a routine 2D ultrasound examination in the third trimester of pregnancy and maternal variables associated with a large fetus. For further improvement of BW prediction in the clinically most critical subgroup of infants with BW >4,500 g, a 3D ultrasound examination including volumetry of fetal thigh and abdomen might be offered.