Fetal echocardiography for congenital heart disease diagnosis: a meta-analysis, power analysis and missing data analysis

Background Prenatal ultrasonography is the most widely available diagnostic test for fetal congenital heart disease (CHD), but the factors influencing its diagnostic accuracy remain uncertain despite extensive research. The aim of the present study was to evaluate the potential role of demographic, clinical and ultrasonographic characteristics on diagnostic yields for detecting CHD. Methods A systematic search of PubMed, ISI Web of Science, SinoMed, and the Cochrane Library was performed to identify studies assessing the accuracy of prenatal ultrasound in the detection of CHD. A random effects model was used to generate pooled sensitivity and specificity in addition to summary receiver operating characteristic (SROC) curves. Results Overall, prenatal ultrasound in the detection of CHD had a moderate sensitivity of 68.1% (95% CI 59.6-75.5) and a favorable specificity of 99.9% (99.7-99.9). Risk level and gestation age were independent predictors of diagnostic performance for detecting CHD (p=0.004 vs. p=0.002, respectively). The pooled sensitivities significantly increased to varying extents with the following echocardiographic views: 48.7% (34.8-67.2) for four-chamber view (4CV); 58.0% (40.3-73.9) for a combination of 4CV and outflow tract views (OTV); 73.5% (59.2-84.1) for combination of 4CV, OTV and three vessels and trachea view (3VTV); 77.1% (62.0-87.5) for extensive cardiac echocardiography examination (ECEE); and 89.6% (81.0-94.6) for spatiotemporal image correlation (STIC). Conclusions Prenatal ultrasound is a powerful tool for the diagnosis of CHD; however, a single ultrasonographic regime is not definitive on its own and must be interpreted in the context of demographic and clinical characteristics.