Diagnostic accuracy of transthoracic three-dimensional left ventricular function analysis compared with cardiac magnetic resonance imaging: a bivariate meta-analysis

Studies evaluating the diagnostic concordance of three-dimensional echocardiography (3DE) and cardiac magnetic resonance imaging (CMR) for quantification of left ventricular ejection fraction (LVEF) have yielded inconsistent findings. We searched PubMed, Embase and Cochrane Library databases for prospective studies that compared the diagnostic accuracy of 3DE for quantification of systolic left ventricular function in adults with that of CMR (the gold standard method). We constructed 2 × 2 tables at the study-level based on an LVEF threshold of 55%. We performed a bivariate random-effects meta-analysis to calculate summary estimates of diagnostic accuracy. Twelve studies involving 581 patients were included. 3DE showed overall sensitivity and specificity of 76% (95% confidence interval [CI] 59–88%) and 91% (95% CI 84–95%), respectively. The diagnostic accuracy of 3DE was high with an area under the hierarchical summary receiver-operating characteristic curve (HSROC) of 0.92 (95% CI 0.90–0.94) for appropriate quantification of LVEF. The diagnostic performance was high irrespective of the applied method (semi-automated or automated algorithm) of 3D left ventricular quantification. Between-study heterogeneity was high (p <0.001 for Q statistic). The high diagnostic accuracy of 3DE as compared with CMR allows safe use of the method, especially for followup of patients requiring serial quantification of left ventricular function.