Meta-Analysis of Accuracy of Left Ventricular Mass Measurement by Three-Dimensional Echocardiography

• Published in: The American journal of cardiology Vol. 110; no. 3; pp. 445 - 452
• Main Authors: Shimada, Yuichi J., MD, Shiota, Takahiro, MD, PhD
• Format: Journal Article
• Language: English
• Published: New York, NY Elsevier Inc 01.08.2012; Elsevier; Elsevier Limited
• Subjects: Biological and medical sciences; Cardiology. Vascular system; Cardiovascular; Cardiovascular disease; Cardiovascular system; Echocardiography, Three-Dimensional; Humans; Hypertrophy, Left Ventricular - diagnosis; Hypertrophy, Left Ventricular - diagnostic imaging; Investigative techniques, diagnostic techniques (general aspects); Magnetic Resonance Imaging; Measurement; Medical sciences; Meta-analysis; Regression analysis; Reproducibility of Results; Ultrasonic imaging; Ultrasonic investigative techniques; Sonography; Mass; Measurement; Accuracy; Echocardiography; Analysis; Tumor; Tridimensional image; Circulatory system; Cardiology; Left ventricle
• ISSN: 0002-9149; 1879-1913
• DOI: 10.1016/j.amjcard.2012.03.046

Left ventricular (LV) hypertrophy is a fundamental prognostic factor in a variety of cardiac diseases. Three-dimensional echocardiography (3DE) has achieved better estimation of LV mass than 2-dimensional echocardiography. However, significant underestimation has often been reported, and no previous study has synthesized these data. The aim of this meta-analysis was to investigate if there has been improvement in the accuracy in LV mass measurement by 3DE over time. Studies comparing LV mass between 3DE and magnetic resonance imaging were eligible. A cumulative meta-analysis was performed to investigate improvement in accuracy, followed by subgroup and meta-regression analysis to reveal factors affecting the bias. A total of 25 studies including 671 comparisons were analyzed. Studies published in or before 2004 showed high heterogeneity (I2 = 69%) and significant underestimation of LV mass by 3DE (−5.7 g, 95% confidence interval −11.3 to −0.2, p = 0.04). Studies published from 2005 to 2007 were still heterogenous (I2 = 60%) but showed less systematic bias (−0.5 g, 95% confidence interval −2.5 to 1.5, p = 0.63). In contrast, studies published in or after 2008 were highly homogenous (I2 = 3%) and showed excellent accuracy (−0.1 g, 95% confidence interval −2.2 to 1.9, p = 0.90). Investigation of factors affecting the bias revealed that evaluation of cardiac patients compared to healthy volunteers led to larger bias (p <0.05). In conclusion, this meta-analysis elucidates the underestimation of LV mass by 3DE, its improvement over the past decade, and factors affecting the bias. These data provide a more detailed basis for improving the accuracy of 3DE, an indispensable step toward further clinical application.