Background: Although global left ventricular systolic function is an important determinant of mortality, standard measures such as ejection fraction (EF) and wall motion score index (WMSI) have important technical limitations. The aim of this study was to compare global longitudinal speckle strain (GLS), an automated technique for measurement of long-axis function, with EF and WMSI for the prediction of mortality.
Methods and results: Of 546 consecutive individuals undergoing echocardiography for assessment of resting left ventricular function, 91 died over a period of 5.2+/-1.5 years. In addition to Simpson biplane EF, WMSI was determined by 2 experienced readers and GLS was calculated from 3 standard apical views using 2D speckle tracking. The incremental value of EF, WMSI, and GLS to significant clinical variables was assessed in nested Cox models. Clinical factors associated with outcome (model chi(2)=20.2) were age (hazard ratio [HR], 1.46; P<0.01), diabetes (HR, 1.88; P=0.01), and hypertension (HR, 1.59; P<0.05). Although addition of EF (HR, 1.23; P=0.03) or WMSI (HR, 1.28; P<0.01) added to the predictive power of clinical variables, the addition of GLS (HR, 1.45; P<0.001) caused the greatest increment in model power (chi(2)=34.9, P<0.001). GLS also provided incremental value in subgroups with EF >35% and those with and without wall motion abnormalities. A GLS >or=-12% was found to be equivalent to an EF <or=35% for the prediction of prognosis. Intraobserver and interobserver variations for EF and GLS were similar.
Conclusions: GLS is a superior predictor of outcome to either EF or WMSI and may become the optimal method for assessment of global left ventricular systolic function.