Prediction of All-Cause Mortality From Global Longitudinal Speckle StrainCLINICAL PERSPECTIVE
Comparison With Ejection Fraction and Wall Motion Scoring
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 χ2=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 (χ2=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 ≥−12% was found to be equivalent to an EF ≤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.
Received March 9, 2009; accepted July 17, 2009.
Global left ventricular (LV) systolic function, most commonly assessed by echocardiographic ejection fraction (EF), is an important predictor of outcome1 and determines eligibility for interventions such as device implantation2 and valvular surgery.3 However, the measurement of EF presents a number of challenges related to image quality, assumptions of LV geometry, and expertise. Two-dimensional strain (2DS) is an automated and quantitative technique for the measurement of global long-axis function from gray-scale images. Longitudinal tissue deformation is evaluated by frame-by-frame tracking of individual speckles throughout the cardiac cycle, and global longitudinal speckle strain (GLS) is calculated from the mean of 18 cardiac segments. Previous work has shown that EF can be derived from GLS using the regression EF=−4.35*(GLS+3.9).4 2DS is more robust than tissue Doppler-derived strain, does not have angle dependency, and is easier to calculate.
Clinical Perspective on p 356
Although 2DS has been used in a growing number of situations, its prognostic utility has not yet been evaluated. We aimed to study this in a consecutive sample of patients with known or suspected LV impairment and to compare this against other common measures of ventricular function such as EF and wall motion score index (WMSI). We also wished to examine the use of GLS in a number of clinically important subgroups.
We retrospectively studied 546 unselected, consecutive individuals undergoing clinically indicated echocardiography to investigate known or suspected LV impairment. The clinical characteristics of these patients are outlined in Table 1.
Cine loops from 3 standard apical views (4-chamber, 2-chamber, and apical long-axis) were recorded using gray-scale harmonic imaging and saved in raw data format (Vivid 7, General Electric Medical Systems, Horten, Norway). Images were obtained at a frame rate of 50 to 70 per second, and digital loops were saved onto optical disc for off-line analysis (EchoPac 8.0, General Electric Medical Systems). End-diastolic and end-systolic volumes were used to calculate EF by Simpson biplane method from the apical 4- and 2-chamber views.5
Wall motion scores were measured by 2 experienced observers blinded to clinical and outcome data, using a 16-segment model as described by the American Society of Echocardiography. Segments were scored as normal (score=1), hypokinetic (score=2), severely hypokinetic (score=2.5), akinetic (score=3), or dyskinetic (score=4). WMSI was derived as the average of the 16 segments.5
The endocardial borders were traced in the end-systolic frame of the 2D images from the 3 apical views. Speckles were tracked frame-by-frame throughout the LV wall during the cardiac cycle and basal, mid, and apical regions of interest were created. Segments that failed to track were manually adjusted by the operator. Any segments that subsequently failed to track were excluded. Any view in which 2 or more segments could not be tracked was not included in the analysis, and the remaining apical views were averaged to calculate GLS; otherwise, GLS was calculated as the mean strain of all 18 segments. Three patients had 1 view from which 2DS was unable to be reliably measured, and GLS was therefore calculated as the average of the 2DS measures from the 2 remaining apical views. All measurements were made blinded to other results and clinical details. Examples of the technique are shown in Figure 1.
All-cause mortality was the primary end point. Follow-up was obtained after a mean of 5.2±1.5 years by review of the patient’s hospital or family practice chart or telephone interview with the patient or relative.
Analysis was carried out using a standard statistical software program (SPSS version 16, SPSS Inc, Chicago, Ill). Student t test was used to compare differences between 2 groups for continuous variables, and the χ2 test was used to determine significant differences between 2 groups of categorical variables.
Univariate analysis was performed to establish the relationship between baseline clinical features, measures of LV function, and all-cause mortality. Survival was expressed using Kaplan-Meier analysis and log-rank tested for significance both overall and between strata. Cox proportional hazards analysis was used to determine significant predictors of all-cause mortality. Variables with a univariate statistical significance of <0.10 were selected for inclusion into the model as follows. First, significant baseline demographic variables (n=3) known before echocardiography were entered into the model. A series of nested models with the separate addition of EF, WMSI, and GLS were then undertaken. EF was entered into the model as a negative variable to produce a positive hazard ratio to allow comparison with WMSI and GLS. All continuous variables were assessed per unit standard deviation to enhance the comparison of the measures (per change of ±11.9 years for age, ±12.1% for EF, ±0.41 for WMSI, and ±4.3% for GLS). The independence and incremental value of each measure of LV function over baseline was assessed by comparison of model χ2 at each step. Changes in receiver operating characteristic curves between variables for the prediction of death were analyzed using Z scores. Intraobserver and interobserver variations were assessed by intraclass correlation coefficient and compared using Z scores and Bland-Altman plots. Significance was measured as <0.05.
Relationship of GLS, EF, and WMSI
The mean EF was 58±12% (range, 16% to 81%), with a mean WMSI of 1.3±0.4 and mean GLS of −16.6±4.3%. GLS, EF, and WMSI were highly significantly correlated in the population overall and in each subgroup analyzed with the exception of EF and WMSI in those with EF ≤35% (Table 2). GLS and EF were consistently the most strongly correlated and had the strongest correlation in those with wall motion abnormalities.
Survival in the Total Population
There were 91 deaths (16.7%) over 5.2±1.5 years of follow-up. The significant univariate predictors of mortality from baseline variables (overall model, χ2=20.2) were diabetes (hazard ratio [HR], 1.82; 95% CI, 1.15 to 2.86), age (HR, 1.46; 95% CI, 1.19 to 1.88), and hypertension (HR, 159; 95% CI, 1.04 to 2.42) (Table 3). Three separate models were then used to evaluate the additional prognostic information obtained from adding imaging with EF, WMS, or GLS as the second step to the baseline model. Diabetes was the strongest predictor, and hypertension was nonsignificant in each of the 3 models. EF, WMS, and GLS were significant predictors in each of the models constructed, and each produced a significant increase in model χ2 over baseline. GLS imparted the greatest increase in model χ2 (34.9, P<0.001), followed by WMS (28.6, P<0.01) and then EF (25.3, P=0.04) (Table 4 and Figure 2). The area under the receiver operating characteristic curve for GLS (0.63, P<0.01) exceeded that for WMA (0.57, P=0.04) and EF (0.45, P=0.10). Comparison of the areas under the receiver operating characteristic curves for prediction of death showed no significant difference between GLS and EF (P=0.07), but GLS and WMSI (P<0.0001) were significantly different.
The addition of GLS to a model including baseline variables plus EF affected a significant increase in model χ2 (from 25.3 to 34.9, P=0.001), thus proving the incremental value of GLS over EF. In addition, in this model, EF (HR, 0.86; 95% CI, 0.64 to 1.14; P=0.29) was not a significant predictor of outcome in contrast to GLS (HR, 1.65; 95% CI, 1.22 to 2.23; P=0. 001), which was also now the strongest predictor of outcome superior to diabetes (HR, 1.62; 95% CI, 1.02 to 2.57; P=0.04) and age (HR, 1.39; 95% CI, 1.01 to 1.75; P<0.01).
Moderate LV Dysfunction
An EF ≤35% was identified in 29 individuals (5.3%). For those with EF >35%, age was a significant predictor (HR, 1.44; 95% CI, 1.13 to 1.83; P<0.01) at baseline. WMSI was not a significant independent predictor in this group. GLS was a significant predictor (HR, 1.42; 95% CI, 1.12 to 1.8) with age (HR, 1.46; 95% CI, 1.14 to 1.85), and the addition of GLS to the model caused a significant increase in model power (Figure 3). No variables were predictive of mortality in the group with EF ≤35%, most likely due to small sample.
Wall Motion Abnormalities
A resting wall motion abnormality (WMSI >1.0) was reported in 239 patients (43.8%). These individuals were significantly older (62.5±11.2 versus 59.6±12.3 years, P<0.01), had a greater proportion of men (75.3% versus 55.4%, P<0.01), and had lower EF (51.6±12.7 versus 62.4±9.2%, P<0.01) and lower GLS (−14.3±4.3 versus 18.4±3.2, P<0.01) than those with normal WMSI. During follow-up, 45 (14.8%) individuals with normal WMSI died and 46 (19.2%) individuals with a resting wall motion abnormality died. The outcome of these groups was similar (P=0.07, Figure 4A).
For the 307 patients (56.2%) without wall motion abnormalities, GLS (HR, 1.9; 95% CI, 1.28 to 2.81; P<0.01) was a significant predictor of outcome along with diabetes (HR, 2.63; 95% CI, 1.42 to 4.86; P<0.01). The addition of GLS to baseline variables increased model power from 13.2 to 22.1 (P<0.01). The substitution of EF for GLS showed that EF was not a significant predictor and did not increase model power similarly (13.2 to 13.2, P=NS). Both GLS (HR, 1.36; 95% CI, 1.02 to 1.82; P<0.05) and EF (HR, 1.33; 95% CI, 1.02 to 1.75; P<0.05) were significant predictors of death in those with wall motion abnormalities (WMSI >1) and increased model power similarly (11.8 to 15.0, P<0.05).
Use of GLS to Identify Significant LV Dysfunction
An EF ≤35% was present in 29 individuals (5.3%), and 78 (14.3%) had a GLS ≥−12%. During follow-up, 10 (34.5%) individuals in the group with EF ≤35% and 25 (32.1%) individuals in the GLS ≥−12% group died. Eight (27.6%) individuals in the EF ≤35% group and 16 (20.5%) of the GLS ≥−12% were New York Heart Association (NYHA) functional class ≥2. There were no significant differences in demographics between these 2 groups, which differed in relation to WMSI (2.2±0.5 versus 1.9±0.6, P=0.01) and GLS (8.3±2.7 versus 9.2±2.0, P=0.05). Individuals with EF ≤35% had significantly worse survival than those with EF >35% (log rank χ2=8.65, P<0.01). Similarly, individuals with GLS ≥−12% had significantly worse survival than those with GLS ≥−12% (log rank χ2=16.67, P<0.001). The survival curves for the groups with EF ≤35% and GLS ≥−12% almost completely overlaid each other (P=NS), as did the curves for EF >35% and GLS <−12%, indicating comparable prognostic outlooks for these groups (P=NS) (Figure 4B).
Reliability and Feasibility
Intraobserver and interobserver variations for both EF and GLS were good, with the measurement of GLS outperforming that of EF both within and across observers (Figure 5). Interobserver variability intraclass correlation coefficients were 0.803 for EF and 0.916 for GLS (Z score, 2.37; P=0.03); for intraobserver variability, intraclass correlation coefficients were 0.67 for EF and 0.922 for GLS (Z score, 4.11; P<0.01). The average time for the calculation of EF was 75.9 seconds for observer 1 and 75.5 seconds for observer 2. For GLS, this was 95.9 and 93.6 seconds for observers 1 and 2, respectively.
This study is the first to show the prognostic value of global longitudinal strain and compare it against conventional measures of LV systolic function. The measurement of global LV function is the most common indication for echocardiography,6 and LV dysfunction is known to be a strong prognostic marker of adverse outcome,1 most commonly assessed by the calculation of EF by Simpson biplane or wall motion scoring. These methods, however, are heavily dependent on image quality, image orientation, and reader experience. Recent advances in echocardiography and measures of myocardial deformation have enabled clinicians to evaluate LV function using techniques that are not hampered by these limitations. Tissue velocity-derived strain has been assessed for this purpose, but it has technological limitations such as angle dependence, signal noise, and measurement variability.7 Two-dimensional strain is not angle-dependent and has been shown to correlate well with EF measured both by echocardiography8,9 and MRI.4,7 Previous investigators have correlated GLS with measures of LV function, both echocardiographic and MRI derived, in populations with normal EF,8 chronic heart failure, acute ST-elevation myocardial infarction,9 and previous myocardial infarction.4 The calculation of GLS has proven to be both reliable and quick across both experienced and inexperienced observers.8 This study added to previous knowledge by showing the incremental prognostic benefit of GLS over baseline variables and comparing it against conventional measures of LV function.
Incremental Value of GLS
The total population studied was a consecutive clinical group presenting for echocardiography with a wide range of EF. GLS, EF, and WMSI were added to the baseline clinical model, and each provided significant incremental prognostic information, underlining the value of imaging. Of these 3 methods, GLS provided the greatest increase in model power, superior to WMSI, which, in turn, was superior to EF. Although GLS and EF are highly correlated, they measure different aspects of myocardial motion, with EF measuring radial and partly longitudinal function, whereas GLS measures longitudinal function. Simpler measures of longitudinal function, such as mitral annular systolic velocity, have previously been shown to predict mortality, although being shown to be inferior to EF.10 The development of GLS has advanced the measurement of longitudinal function—first because of automation and second because of the ability to measure the longitudinal function of the entire ventricle rather than basal segments alone. It is thought that longitudinal contraction is a particular marker of subendocardial function, which may be disproportionately involved in subclinical disease, including myocardial ischemia.11
EF ≤35% and Diagnosis of Heart Failure
The current American College of Cardiology/American Heart Association Guidelines for the Diagnosis and Management of Chronic Heart Failure12 stipulate that EF be calculated at the initial diagnosis of heart failure (class I recommendation) and used to follow patients serially (class IIa recommendation). The Simpson biplane method is currently recommended by the American Society of Echocardiography for calculation of EF.5 A cutoff value of LVEF ≤35% is often used clinically to classify severe LV dysfunction, and this value serves clinically as a criterion for the prescription of device therapy.2 Only 29 (5.3%) patients in our population had EF ≤35%, and these individuals had a significantly worse outcome than those with EF >35%.
Our group has previously derived a formula for the conversion of EF to GLS.4 Using this formula, a cutoff of GLS ≥−12% is equivalent to EF ≤35%, and 78 (14.3%) individuals of our study group had a GLS ≥−12%. When a Kaplan-Meier curve for individuals with GLS ≥−12% and GLS <−12% were overlain with that of those with EF ≤35% and EF >35%, the curves demonstrated similar outcomes for these groups, with no significant differences between the groups with EF ≤35% and GLS ≥−12% and EF >35% and GLS <−12% (Figure 4B). We speculate that GLS ≥−12% could be recommended as diagnostic cutoff for severe LV dysfunction, possibly improving access to potentially lifesaving treatments such as implantable defibrillators.
Wall Motion Abnormalities
The calculation of WMSI is based on the subjective assessment of regional wall motion and has been shown to predict mortality after myocardial infarction.13 For those without wall motion abnormalities, GLS provided incremental prognostic information over baseline variables, whereas EF did not. Both GLS and EF increased model power similarly in those with wall motion abnormalities (Figure 4). GLS not only measures contraction but is also able to reflect interstitial myocardial changes such as fibrosis, which are often subclinical.14 This may explain the superior predictive power of GLS over EF in those without wall motion abnormalities. In those with wall motion abnormalities when segments are often hypokinetic, scarred, thinned, or calcified, this additive information may be less powerful, with the endocardial border easier to visualize for the calculation of EF.
Relationship and Reliability of Measures of LV Function
GLS had a superior correlation with EF compared with WMSI in each subgroups analyzed. Interobserver and intraobserver variabilities were lower for the measurement of GLS-derived EF compared with EF, although both were excellent (Figure 5). GLS took, on average, 19 seconds longer to calculate than EF.
Despite the strong message from our study about the feasibility and prognostic value of GLS, several limitations must be considered. Despite a large population (n=546) only 29 (5.3%) of our patients had an EF ≤35%. This did not permit analysis of predictors of outcome and model power in this subgroup. We measured longitudinal global strain but not radial or circumferential measures. This measure was chosen because longitudinal strain has been well validated against EF in previous studies. It is reproducible and does not significantly add to the time taken to analyze a study. Whether either radial or circumferential strain would perform differently is unknown. Unfortunately, the images analyzed in this study did not have short-axis images of sufficient quality for the measurement of radial or circumferential strain. Finally, all-cause mortality rather than cardiac mortality was examined because the classification of cardiac death is often problematic.15
The results of this study prove the incremental prognostic benefits of LV function imaging measures over standard baseline variables in a large, clinical population. Global longitudinal strain measurement by 2DS was superior to EF and WMSI for the prediction of outcome and may become the optimal method for assessment of global LV systolic function. Guidelines incorporating measures of LV function may need to be revised to incorporate global longitudinal strain in light of this finding.
Sources of Funding
This study was supported in part by a program grant (519823; Integration of Risk Evaluation in Cardiovascular Disease Management) from the National Health and Medical Research Council, Canberra, Australia.
Drs Stanton and Leano have had no relationships within the last 2 years that are relevant to the topic. Dr Marwick has received research support (equipment and software) from GE Medical Systems for studies of 2D strain. However, the current study was not supported by those grants.
Quiñones MA, Greenberg BH, Kopelen HA, Koilpillai C, Limacher MC, Shindler DM, Shelton BJ, Weiner DH. Echocardiographic predictors of clinical outcome in patients with left ventricular dysfunction enrolled in the SOLVD registry and trials: significance of left ventricular hypertrophy: Studies of Left Ventricular Dysfunction. J Am Coll Cardiol. 2000; 35: 1237–1244.
Epstein AE, DiMarco JP, Ellenbogen KA, Estes NA III, Freedman RA, Gettes LS, Gillinov AM, Gregoratos G, Hammill SC, Hayes DL, Hlatky MA, Newby LK, Page RL, Schoenfeld MH, Silka MJ, Stevenson LW, Sweeney MO. American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the ACC/AHA/NASPE 2002 Guideline Update for Implantation of Cardiac Pacemakers and Antiarrhythmia Devices); American Association for Thoracic Surgery; Society of Thoracic Surgeons. ACC/AHA/HRS 2008 Guidelines for Device-Based Therapy of Cardiac Rhythm Abnormalities: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the ACC/AHA/NASPE 2002 Guideline Update for Implantation of Cardiac Pacemakers and Antiarrhythmia Devices): developed in collaboration with the American Association for Thoracic Surgery and Society of Thoracic Surgeons. J Am Coll Cardiol. 2008; 51: e1–e62.
Bonow RO, Carabello BA, Chatterjee K, de Leon AC Jr, Faxon DP, Freed MD, Gaasch WH, Lytle BW, Nishimura RA, O'Gara PT, O'Rourke RA, Otto CM, Shah PM, Shanewise JS. 2006 Writing Committee Members; American College of Cardiology/American Heart Association Task Force. 2008 Focused update incorporated into the ACC/AHA 2006 guidelines for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 1998 Guidelines for the Management of Patients With Valvular Heart Disease): endorsed by the Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons. Circulation. 2008; 118: e523–e661.
Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, Picard MH, Roman MJ, Seward J, Shanewise JS, Solomon SD, Spencer KT, Sutton MS, Stewart WJ. Chamber Quantification Writing Group; American Society of Echocardiography’s Guidelines and Standards Committee; European Association of Echocardiography. Recommendations for chamber quantification: a report from the American Society of Echocardiography’s Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr. 2005; 18: 1440–1463.
Cheitlin MD, Armstrong WF, Aurigemma GP, Beller GA, Bierman FZ, Davis JL, Douglas PS, Faxon DP, Gillam LD, Kimball TR, Kussmaul WG, Pearlman AS, Philbrick JT, Rakowski H, Thys DM, Antman EM, Smith SC Jr, Alpert JS, Gregoratos G, Anderson JL, Hiratzka LF, Faxon DP, Hunt SA, Fuster V, Jacobs AK, Gibbons RJ, Russell RO. American College of Cardiology; American Heart Association; American Society of Echocardiography. ACC/AHA/ASE 2003 guideline update for the clinical application of echocardiography: summary article: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (ACC/AHA/ASE Committee to Update the 1997 Guidelines for the Clinical Application of Echocardiography). Circulation. 2003; 108: 1146–1162.
Belghitia H, Brette S, Lafitte S, Reant P, Picard F, Serri K, Lafitte M, Courregelongue M, Dos Santos P, Douard H, Roudaut R, DeMaria A. Automated function imaging: a new operator-independent strain method for assessing left ventricular function. Arch Cardiovasc Dis. 2008; 101: 163–169.
Delgado V, Mollema SA, Ypenburg C, Tops LF, van der Wall EE, Schalij MJ, Bax JJ. Relation between global left ventricular longitudinal strain assessed with novel automated function imaging and biplane left ventricular ejection fraction in patients with coronary artery disease. J Am Soc Echocardiogr. 2008; 21: 1244–1250.
Wang M, Yip GW, Wang AY, Zhang Y, Ho PY, Tse MK, Lam PK, Sanderson JE. Peak early diastolic mitral annulus velocity by tissue Doppler imaging adds independent and incremental prognostic value. J Am Coll Cardiol. 2003; 41: 82082–82086.
Jones CJ, Raposo L, Gibson DG. Functional importance of the long axis dynamics of the human left ventricle. Br Heart J. 1990; 63: 215–220.
Hunt SA, Abraham WT, Chin MH, Feldman AM, Francis GS, Ganiats TG, Jessup M, Konstam MA, Mancini DM, Michl K, Oates JA, Rahko PS, Silver MA, Stevenson LW, Yancy CW, Antman EM, Smith SC Jr, Adams CD, Anderson JL, Faxon DP, Fuster V, Halperin JL, Hiratzka LF, Jacobs AK, Nishimura R, Ornato JP, Page RL, Riegel B. American College of Cardiology; American Heart Association Task Force on Practice Guidelines; American College of Chest Physicians; International Society for Heart and Lung Transplantation; Heart Rhythm Society. ACC/AHA 2005 Guideline Update for the Diagnosis and Management of Chronic Heart Failure in the Adult: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure): developed in collaboration with the American College of Chest Physicians and the International Society for Heart and Lung Transplantation: endorsed by the Heart Rhythm Society. Circulation. 2005; 112: e154–e235.
Park TH, Nagueh SF, Khoury DS, Kopelen HA, Akrivakis S, Nasser K, Ren G, Frangogiannis NG. Impact of myocardial structure and function postinfarction on diastolic strain measurements: implications for assessment of myocardial viability. Am J Physiol Heart Circ Physiol. 2006; 290: H724–H731.
Global left ventricular systolic function is an important determinant of mortality, and ejection fraction (EF) is the most widely applied parameter for its measurement. Unfortunately, EF has a number of important technical limitations, and, despite its use as an arbiter of therapy (eg, defibrillator implantation), it is often qualitatively assessed. There is a need for a simple and automated quantitative alternative to EF. Global longitudinal speckle strain (GLS) is such an automated technique for measurement of long-axis function, which has been shown to correlate with EF, and we sought in this study to identify whether it could be used for the prediction of mortality. In this study of 546 consecutive individuals undergoing echocardiography for assessment of resting left ventricular function, Simpson biplane EF was measured and GLS was calculated from 3 standard apical views using 2D speckle tracking. The incremental value of EF and GLS to significant clinical variables was assessed in nested Cox models. Although addition of EF (hazard ratio, 1.23; P=0.03) added to the predictive power of clinical variables, the addition of GLS (hazard ratio, 1.45; P<0.001) caused the greatest increment in model power (χ2=34.9, P<0.001). A GLS ≥−12% was found to be equivalent to an EF ≤35% for the prediction of prognosis.