The rapid progress in fetal echocardiography has lead to early detection of congenital heart diseases. Increasing evidences have shown that prenatal diagnosis could be life saving in certain cases. However, there is no agreement on which protocol is most adaptive diagnostic one. Thus, we use meta-analysis to conduct a pooled performance test on 5 diagnostic protocols.
We searched PUBMED, EMBASE, the Cochrane Central Register of Controlled Trials and WHO clinical trails registry center to identify relevant studies up to August, 2012. We performed meta-analysis in a fixed/random-effect model using Meta-disc 1.4. We used STATA 11.0 to estimate the publication bias and SPSS 17.0 to evaluate variance.
We use results from 81 studies in 63 articles to analyze the pooled accuracy. The overall performance of pooled sensitivities of spatiotemporal image correlation (STIC), extend cardiac echography examination (ECEE) and 4 chambers view + outflow tract view + 3 vessels and trachea view (4 CV+OTV+3 VTV) were around 0.90, which was significant higher than that of 4 chambers view + outflow tract view or 3 vessels and trachea view (4 CV+OTV/3 VTV) and 4 chambers view (4 CV). Unfortunately the pooled specificity of STIC was 0.92, which was significant lower than that of other 4 protocols which reached at 1.00. The area under the summary receiver operating characteristic curves value of STIC, ECEE, 4 CV+OTV+3 VTV, 4 CV+OTV/3 VTV and 4 CV were 0.9700, 0.9971, 0.9983, 0.9929 and 0.9928 respectively.
These results suggest a great diagnostic potential for fetal echocardiography detection as a reliable method of fetal congenital heart disease. But at least 3 sections view (4 CV, OTV and 3 VTV) should be included in scan protocol, while the STIC can be used to provide more information for local details of defects, and can not be used to make a definite diagnosis alone with its low specificity.