Reliability of left ventricular volumes and function measurements using three-dimensional speckle tracking echocardiography

Aims Although the accuracy of three-dimensional speckle tracking echocardiography (STE) to quantify left ventricular (LV) volumes and function has been demonstrated, its reliability has not been studied in great detail. The aim of the study was to determine inter-observer, intra-observer, and test-retest reliability of echocardiographic measurements of LV volumes and function using three-dimensional STE. Methods and results A total of 140 consecutive patients presenting for routine echocardiographic examination underwent three-dimensional STE. Twenty-three of the 140 patients (16%) were excluded from the analysis due to atrial fibrillation or insufficient image quality. In the remaining 117 patients [69 males, age 59 +/- 16 years, ejection fraction (EF) 51+13%], intra-observer and inter-observer reliability of LV volumes, EF, and global and segmental strain measurements was determined, whereas test-retest reliability was assessed in a subgroup of 50 patients. LV volumes and EF measurements demonstrated good reliability [intraclass correlation coefficient (ICC): 0.85-0.99; standard error of measurement (SEM): 3.1-9.2 mL and 1.7-4.0%, respectively]. Reliability of global circumferential strain measurements (ICC: 0.85-0.97; SEM: 1.4-2.6%) was superior to longitudinal (ICC: 0.66-0.92; SEM: 1.0-2.1%) and radial strain measurements (ICC: 0.52-0.88; SEM: 4.4-8.1%), with similar results found for segmental strain measurements (P<0.001 for all). Reliability was not significantly affected by the image quality or temporal resolution of 3D data sets for any parameter. Conclusion Good intra-observer, inter-observer, and test-retest reliability support the use of three-dimensional STE for routine evaluation of LV volumes and EF. Global and segmental circumferential strain measurements also demonstrate high reliability, whereas analysis by a single observer is currently recommended for longitudinal and radial strain due to limited inter-observer and test-retest reliability.