Nuchal translucency and major congenital heart defects in fetuses with normal karyotype: a meta‐analysis
Journal
Ultrasound in Obstetrics and Gynecology
Date Issued
April 18, 2013
DOI
10.1002/uog.12488
Abstract
Objective: To pool published data regarding the sensitivity and specificity of nuchal translucency (NT) in the diagnosis of major congenital heart defects (CHDs) in fetuses with normal karyotype. Methods: MEDLINE and Scopus searches using combinations of the terms 'nuchal' and 'cardiac*' were complemented by perusal of references of the retrieved articles and an additional automated search using the 'search for related articles' function on PubMed. Only fetuses with normal karyotype and major CHDs were analyzed. Weighted estimates were made and summary receiver-operating characteristics curves were constructed. Results: The analysis included 20 studies (205 232 fetuses; 537 cases with major CHDs). The pooled sensitivity and specificity of NT > 95th centile for diagnosis of major CHDs was 44.4% (95% CI, 39.5-49.5) and 94.5% (95% CI, 94.4-94.6), respectively. The pooled sensitivity and specificity of NT > 99th centile was 19.5% (95% CI, 15.9-23.5) and 99.1% (95% CI, 99.1-99.2), respectively. For the subgroup of studies in which NT was measured by Fetal Medicine Foundation-certified operators, the pooled sensitivity and specificity of NT > 95th centile was 45.6% (95% CI, 39.6-51.7) and 94.7% (95% CI, 94.6-94.9), respectively. The corresponding estimates for NT > 99th centile were 21.0% (95%CI, 16.5-26.1) and 99.2% (95% CI, 99.2-99.3). The pooled positive likelihood ratio for NT > 99th centile was 30.5 (95% CI, 24.3-38.6). There was high across-studies heterogeneity for most estimates. Conclusion Approximately 44% of chromosomally normal fetuses with CHDs have NT > 95th centile and 20% have NT > 99th centile. However, there is high heterogeneity across studies, which largely remains even in subgroup analyses of studies of apparently similar design, potentially indicating the presence of some residual unidentified bias.