Abstract
Background. The incidence of fetal macrosomia (infant birth weight ≥4000g) has been increasing during recent decades and is associated with maternal and fetal complications. Accurate prenatal prediction of macrosomia therefore would be very useful for planning the strategy for delivery. Unfortunately, the accuracy level of birth weight estimation even by modern ultrasound equipment is still relatively low. It may cause potential harm when macrosomia is missed. On the other hand it may increase the number of unnecessary interventions when it is wrongly suspected. A vast variety of investigators have been trying to combine additional clinical and maternal data, considered as risk factors, in the hope for a better estimation of fetal macrosomia. None of these methods has been useful in the routine clinical practice.
Objective. This study was conducted to find out, whether the sonographic assessment of the fetal weight at term or near term has had influence on the mode of delivery. Concomitantly, the predictive quality of the recently suggested combined diagnostic methods has been analyzed in comparison to sonographic weight estimation alone in an unselected population.
Methods. In the first part of the study 3435 pregnant women, who delivered term, singleton, live born infants between 2004 – 2007 at the University Hospital of Zurich were analyzed. The study population was divided into 4 groups (true positive, true negative, false positive, false negative) according to the estimated fetal weight (EFW) and regarding the birth weight (BW).
In the second part of the study the retrospective cohort data was obtained from 1062 pregnancies of an unselected population. The estimated fetal sonographic weight was obtained within the last week prior to delivery. The combination models published by Mazouni et al. and Nahum and Stanislaw were employed to predict the presence of macrosomia at birth in these infants. Receiver-operating characteristics (ROC) curves were generated to compare the prediction of macrosomia when using different observation methods. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy were calculated. The mean values of two groups were compared using two sample t test and X² test for comparison of proportions. Additionally, X² test was completed by Post Hoc Cell analysis with the Wilson test.
Results. The first study population included 364 (10.6%) macrosomic infants within a mean BW of 3404 ± 471g and a mean of EFW 3279 ± 475g. Cesarean sections were performed for 58.2% of the pregnant women, where fetal macrosomia was truly ruled in (true positive) and for 34.6% of the women who delivered normal weight infants (true negative). Overestimation of fetal weight (false positive) has led to the 46.3% rate of cesarean sections. However, only 27.0% of false negative defined infants were delivered during caesarean section where fetal weight was underestimated. The difference of the cesarean section rate between true positive and false negative groups was equal 0.31 (CI 95%, 0.21 - 0.4) and greater than that between false positive and true negative groups 0.12 (CI 95% 0.01 - 0.23) (P< 0.0001).
In the second study population macrosomia was present in 135 of 1062 (12.7%) newborns.
The prediction of a probable macrosomia using ROC curve analysis revealed ultrasound alone to be significantly superior to the Mazouni et al. combined method (AUC 0.922 (CI 95%, 0.902-0.943) vs. 0.747 (CI 95%, 0.700-0.794), (P<0.0001), respectively, whereas the Nahum and Stanislaw equations were similar but not superior to ultrasound alone (AUC 0.895 (CI 95% 0.839-0.950) [3], (AUC 0.887 (CI 95% 0.834-0.941) [4], (AUC 0.885 (CI 95% 0.831-0.940) [11], vs. 0.912 (CI 95% 0.867-0.958) respectively, (P<0.0001).
The accuracy of macrosomia prediction was similar for ultrasound alone and the Nahum and Stanislaw equations (~ 90%) whereas the Mazouni et al. nomogram reached only 51.7% (probability cut-off level of 50%).
Interpretation. This study proves the hypothesis that an inappropriate prediction of the fetal weight influences the method of delivery. Underestimation of the fetal weight in the group of macrosomic infants has led to a lesser amount of cesarean sections in comparison with the group where fetal macrosomia was correctly diagnosed. On the other hand, unexpected fetal and maternal complications could occur because of inadequate care during labor. Furthermore, parents would have had a lack of appropriate information and could not participate in taking a decision regarding the mode of delivery.
When applying the model proposed by Mazouni et al. to our unselected population, accuracy of the prediction of macrosomia was distinctly lower than that of ultrasound alone. The best equations of Nahum and Stanislaw for the prediction of macrosomia did not appear to be superior to the ultrasound estimation alone neither. It is true that the error associated with the sonographic estimations of the fetal weight at term reduces its value; nevertheless, currently sonography is still the best and most objective method for birth weight estimation available and a combination with pregnancy specific data does not improve the predictive quality of macrosomia detection at delivery at least in an unselected population. The future aim would be to introduce an automated individual measurement error evaluation to detect the changes associated with an unacceptable rate of unfavourable outcomes and therefore improve the sonographic accuracy.
Balsyte, Dalia