How to improve the accurancy of diagnostic ultrasound to identify small for gestational age fetuses?

Small for gestational age (SGA) fetuses are at higher risk of poor outcome at birth. Since a significant number is misdiagnosed, in the present study we evaluated whether fetal growth may be assessed more precisely than currently, by using a multivariate approach. Methods We retrospectively evaluated 9447 singleton pregnancies, between 22 and 42 weeks of gestation. Women without 1st trimester dating, with fetal malformations and preterm deliveries were excluded. A fetal routinary biometry was recorded and estimated fetal weight (EFW) was calculated. Nomograms for each biometric parameter and for EFW were constructed. Each measurement of male and female fetal parameters was evaluated at each gestational age (GA) and converted to Multiples of Median (MoM). Statistical analysis was done using the unpaired t-test or Mann-Whitney U test for non parametric values. In the group of 5942 women that delivered from 0 to 6 days from US, abdominal circumference (AC) values were compared to Local Reference Curve, whilst all EFWs were compared to Festini nomogram, constructed and validated on Tuscany population. Sensitivity, specificity, predictive values and likelihood ratios of AC and EFW were assessed by using Receiver Operating Curve (ROC) analysis. Results 7359 pregnancies met the inclusion criteria. At birth 890 neonates were SGA (prevalence of 14.9%). Statistically significant gender difference in fetal AC-MoM and EFW-MoM values was observed at each GA, with lower results in female than male fetuses (p < 0.0001). According to Local Reference Curve (for AC) and to Festini nomogram (for EFW), at the best cut-off chosen by ROC curve analysis, in the identification of SGA fetuses AC and EFW achieved a sensitivity of 90.0% and 94.2% respectively; specificity 86.1% and 84.5%; LR+ 6.47 and 6.07; LR- 0.12 and 0.07; PPV 29.4% and 28.1%; NPV 99.3% and 99.6%. The area under the ROC curve was 0.915 for AC and 0.932 for EFW, respectively (p < 0.0001). Conclusion The present data suggest that taking care of gender variables, i.e. using sex specific nomograms and comparing EFW with a local population specific chart, it is possible to enhance the accuracy of the US assessment of fetal growth and weight. These devices correctly categorize each fetus during gestation and better estimate the risk of misdiagnosing fetuses at higher risk of poor outcome at birth