Diagnostic performance of handheld echocardiography for the assessment of basic cardiac morphology and function: a...

To investigate the intra- and interrater variability of expert users in the interpretation of handheld echocardiographic studies (HAND).

We scanned 320 consecutive patients with both HAND and high-end (HIGH) scanners. Images were interpreted independently by two blinded level III echocardiographers. Readings from the HIGH scanner served as the gold standard. Segmental endocardial-border delineation was scored to describe image quality. Assessment of left ventricular (LV) dimensions and regional/global LV function, and grading of valve disease were compared.

We obtained correlations of r > 0.8 (P < 0.01) for intrarater variability for both expert readers when they analyzed HAND and HIGH images in relation to image quality, wall-motion abnormalities, and LV measurements. For intrarater variability of LVEF assessment, the correlations were at least moderate (r > 0.6, P < 0.01). Interrater variability for HIGH images was r = 0.9 (P < 0.01) for all parameters. Interrater variability for HAND images was less favorable for all parameters, but was at least moderate (r > 0.6, P < 0.01). All cases of pericardial effusion were detected. The agreement for the detection and grading of mitral and aortic regurgitation was at least moderate (κ > 0.6, P < 0.01). Detection of tricuspid regurgitation was less favorable, but only cases of mild regurgitation were missed. All cases of aortic stenosis were detected by both echocardiographers.

In relation to the basic assessment of cardiac morphology and function, the interpretation by experienced echocardiographers of images obtained using handheld echocardiographic devices showed a moderate to very good correlation with standard echocardiography.