Poster session II Diabetes, labor, medical-surgical-disease, obstetric quality & safety, prematurity, ultrasound-imaging: Abstracts 237 – 386| Volume 206, ISSUE 1, SUPPLEMENT , S164, January 01, 2012

348: The accuracy of prenatal ultrasound in the diagnosis of true microcephaly


      To determine the accuracy of prenatal ultrasound in detection of microcephaly, defined as birth head circumference (HCbirth) <10%ile.

      Study Design

      This was a retrospective chart review of fetuses identified prenatally with a head circumference (HCfetal) of <3%ile. HCbirth were abstracted from the newborn medical chart, and Z-scores calculated for both. Sensitivity, specificity, false positive (FP) and false negative (FN) rates were calculated for HCfetal. An ROC curve was calculated to determine the accuracy of HCfetal in the detection of microcephaly.


      An ultrasound database search from Jan 2005 to July 2011 for HC <3% identified 730 ultrasounds of 455 fetuses in 433 patients. There were 375 live births in this group. Median Z-scores were similar between fetal and neonatal groups (−1.16 [−1.6, −0.9] and −1.2 [−1.8, −0.7], respectively). The overall prevalence of microcephaly was similar in both groups using an HCfetal Z-score cutoff of ≤ −1.3 (40.1% and 49.6% respectively). A Z-score of ≤ −1.3 had 44.6% sensitivity, 35.1% specificity, 44.9% FP rate, and 45.9% FN rate for detection of microcephaly (p=0.08). Using a HCfetal Z-score cutoff of ≤ −1.7 yielded 28.8% sensitivity, 21% specificity, 62.6% FP and 28.2% FN rate (p=0.09). The area under the ROC curve was 0.6, indicating that HCfetal is an inaccurate test for microcephaly. Gestational age at time of ultrasound was not significantly associated with microcephaly (95% CI −0.27-0.01).


      Prenatal ultrasound was able to detect a population at risk for microcephaly as evidenced by the high prevalence in this group at birth; however, the ability of prenatal ultrasound to predict a specific HCbirth for an individual is poor. Sensitivity and specificity for prenatal ultrasound are low, and FP and FN rates are high. Factors which contribute to the inaccuracy of prenatal ultrasound include the variability of the microcephaly phenotype, heterogeneous etiologies, lack of consensus regarding the definition of microcephaly, high rate of concurrent growth restriction, and lack of gender- or race-specific fetal growth curves.