- Email: [email protected]
Prognostic implications of ejection fraction from linear echocardiographic dimensions: the strong heart study
Prognostic implications of ejection fraction from linear echocardiographic dimensions: The Strong Heart Study Richard B. Devereux, MD,a Mary J. Roman, MD,a Vittorio Palmieri, MD,a Jennifer E. Liu, MD,a Elisa T. Lee, PhD,b Lyle G. Best, MD,c Richard R. Fabsitz, MA,d Richard J. Rodeheffer, MD,e and Barbara V. Howard, PhDf New York, NY, Timber Lake, SD, Bethesda, Md, Washington, DC, and Rochester, Minn
Background
Although echocardiography is commonly used to assess left ventricular (LV) systolic function, few data are available concerning the prognostic significance of LV ejection fraction (EF) calculated from linear echocardiographic measurements or 2-dimensional (2-D) wall motion scores in population-based samples.
Methods Echocardiography was used in the second Strong Heart Study (SHS) examination to calculate LV EF in 2948 American Indians without prevalent coronary heart disease; 2923 had 2-D wall motion scores. Results
Mildly and severely reduced LV EF occurred in 10% and 2% of participants, was associated with older age, male sex, higher systolic pressure, heart rate and markers of renal disease and inflammation. During 37 ⫾ 9 months follow-up, cardiovascular death occurred in 2%, 5% and 12% of participants with normal, mildly reduced and severely reduced EF; all cause mortality rates were 6%, 10% and 32% (both P ⬍ .001). In Cox proportional hazards analyses, adjusting for covariates, cardiovascular death was higher with mildly reduced EF (risk ratio [RR] 2.9, 95% CI 1.6 –5.4, P ⫽ .0007) and especially with severely reduced EF (RR 6.9, 95% CI 3.0 –15.9, P ⬍ .0001); all-cause mortality was increased with severe LV dysfunction (RR 4.8, 95% CI 2.8 – 8.1, P ⬍ .001) and marginally with mildly reduced EF (odds ratio 1.4, 95% CI 0.95–2.15, P ⫽ .08). Segmental LV dysfunction and mildly and severely reduced EF from 2-D wall motion scores were associated with 3.3-fold (95% CI 1.1–9.4, P ⫽ .02), 3.5-fold (95% CI 2.1–5.8) and 3.8-fold (95% CI 1.9 –7.6) (all P ⬍ .001) increased rates of cardiovascular death.
Conclusions LV EF from linear echocardiographic measurements as well as segmental LV dysfunction and EF from 2-D wall motion scores strongly and independently predict cardiovascular mortality. Reduced EF by simple echocardiographic method has estimated population-attributable risks of about 35% for cardiovascular death and 12% for all-cause mortality in a population-based sample of middle-aged to elderly adults. (Am Heart J 2003;146:527–34.) It is well recognized that left ventricular (LV) ejection fraction (EF) bears a strong negative relation to rates of cardiovascular morbidity and mortality. The initial large-scale studies documenting this relationship utilized LV angiograms performed at cardiac catheterFrom aCornell Medical Center, New York, NY, bUniversity of Oklahoma Health Sciences Center, Oklahoma City, Okla, cMissouri Breaks Industries Research, Timber Lake, SD, dNational Heart Lung and Blood Institute, Bethesda, Md, eMedStar Research Institute, Washington, DC, andfMayo Clinic, Rochester, Minn. Supported in part by grants U01-HL41642, U01-HL41652 and U01-HL41654 from the National Heart, Lung and Blood Institute and M10RR0047-34 (GCRC) from the National Institutes of Health, Bethesda, Md. This manuscript presents views of the authors and not necessarily those of the Indian Health Service. Submitted October 4, 2002; accepted February 5, 2003. Reprint requests: Richard B. Devereux, MD, New York Presbyterian Hospital, 525 East 68th Street, NY, NY 10021. E-mail: [email protected] © 2003, Mosby, Inc. All rights reserved. 0002-8703/2003/$30.00 ⫹ 0 doi:10.1016/S0002-8703(03)00229-1
ization and coronary arteriography.1,2 Subsequent studies have shown that LV EF measured by nuclear angiography or 2-dimensional (2-D) echocardiography also strongly predicts cardiovascular events.3,4 However, large-scale application of these techniques in clinical practice or population-based research is limited by the need for radiation exposure and variably complex and time-consuming methods of measurement. In clinical practice, estimates of LV systolic function are commonly derived from simple echocardiographic measurements of linear LV dimensions or from 2-D regional wall motion scores, which have been validated by comparison with contrast or nuclear angiographic reference standards.5– 8 Numerous studies have shown that measurements of LV mass from linear echocardiographic LV dimensions predict cardiovascular events and death,9 –12 but no data exist on the prognostic significance of LV EF calculated from simple LV linear dimension measurements or from 2-D echocar-
American Heart Journal September 2003
528 Devereux et al
diographic wall motion scores. Accordingly, the present study was undertaken to relate baseline measurements of EF by these echocardiographic methods to event rates in the population-based cohort of American Indian participants in the Strong Heart Study (SHS).
Methods The SHS is a population-based survey of cardiovascular risk factors and prevalent and incident cardiovascular disease in American Indians. As previously described,13,14 members, aged 45 to 74 years, of 13 tribes in Arizona, Oklahoma and North and South Dakota were recruited (overall participation rate 62%) for a first examination from 1989 to 1992. The second SHS examination was conducted from 1993 to 1995 to assess change over time in body habitus, blood pressure and other baseline measures and to add echocardiography among 3630 participants (89% of survivors). Standardized measurements of seated blood pressure; aspects of body habitus including body mass index, waist/hip ratio and percent body fat by bioelectric impedance; fasting glucose, insulin, lipid and lipoprotein concentrations; and 2-hour glucose tolerance test and glycosylated hemoglobin levels were obtained. Diabetes mellitus was diagnosed by World Health Organization criteria15; individuals with normal fasting sugars include those with normal or impaired glucose tolerance. Prevalent definite coronary heart disease was identified by published criteria.16 Deaths were identified through regular review of hospital, governmental and Indian Health Service records and regular contact with participants and their informants, with vital status known for 99.8% of cohort members. Causes of death were classified as cardiovascular or due to other causes by an expert physician panel on the basis of death certificates, medical records and informant interviews, as previously described.17,18
Echocardiographic methods Imaging and Doppler echocardiograms were performed as previously described.19,20 A standardized protocol was followed under which the parasternal acoustic window was used to record ⱖ10 consecutive beats of 2-D and M-mode recordings of LV internal diameter and wall thicknesses just below the mitral leaflet tips in long and short-axis views, long-axis views of the mitral valve and color flow recordings to search for mitral and aortic regurgitation, and 2-D longaxis views of the aortic root. The apical window was used to record ⱖ10 cycles of 2 and 4-chamber images and color Doppler recordings to assess LV wall motion and identify valvular regurgitation.
Echocardiographic measurements Correct orientation of planes for imaging and Doppler recordings was verified as previously described.21 Measurements of LV internal dimension and septal and posterior wall thicknesses were made at end-diastole and end-systole by American Society of Echocardiography recommendations22,23 on up to 3 cardiac cycles. Segmental wall motion was graded on a scale from normal to dyskinetic in 5 segments at the
base of the left ventricle, 5 at mid-cavity level and 4 at the apex.24
Calculation of derived variables End-diastolic LV dimensions were used to calculate LV mass by a formula shown to predict necropsy LV weight accurately (r ⫽ 0.90, P ⬍ .001)25 and to show good reproducibility ( ⫽ 0.93, P ⬍ .001) in a separate series of 183 hypertensive patients.26 LV mass from 2-D echocardiographic linear measurements has also been shown to have strong predictive value for cardiovascular death.27 Relative wall thickness was calculated as posterior wall thickness/internal radius. Systolic fractional shortening in percent of the LV internal dimension and end-systolic wall stress were calculated by standard methods.21 End-diastolic and end-systolic LV volumes, calculated by the Teichholz method,5,6 were used to calculate LV ejection fraction (EF). Partition values were used to separate SHS participants with normal LV EF from those with mild LV dysfunction (EF ⬍54%, the lower limit of the 95% CI in 362 apparently normal members of a reference population28), and those with mild LV dysfunction from those with severe dysfunction (EF ⬍40%).29,30 To estimate EF from segmental wall motion, normal segments were assigned a value of 4.5, 3.5 to mildly hypokinetic, 2.5 to moderately hypokinetic, 1.5 to severely hypokinetic and 0 to akinetic segments.31 Values were summed, with adjustment if needed for unscorable segments, resulting in an EF of 63% if all segments were normal, 49% if all were mildly hypokinetic, 35% if all were moderately hypokinetic and 21% if all were severely hypokinetic. The lower limit of the 95% CI for EF from wall motion scores in normal-weight, normotensive, nondiabetic SHS participants was 51%; EFs of 40% to 50% were considered mildly reduced, whereas 2-D EFs ⬍40% were considered severely reduced.
Measures of myocardial performance The primary approach to assess myocardial contractile efficiency was examination of LV midwall shortening in relation to circumferential end-systolic stress measured at the level of the LV midwall.32–34 Midwall shortening from echocardiographic measurements was expressed as a percent of the value predicted from circumferential end-systolic stress using equations derived from previously-studied normal subjects32; this variable is termed stress-corrected midwall shortening.35
Statistical analyses Data are expressed as mean ⫾ SD. Differences between groups were assessed by ANOVA with the Scheffe´ test. Because of differences between SHS participants in Arizona, Oklahoma and North and South Dakota for clinical and echocardiographic variables,14,36 indicator variables comparing Arizona and Oklahoma participants to those in North and South Dakota were entered as covariates. Independence of differences in echocardiographic variables from effects of clinical covariates was assessed using general factorial ANOVA and the Sidak test with consideration of age, sex, systolic pressure, antihypertensive treatment, diabetes, lowdensity lipoprotein [LDL] cholesterol, cigarette smoking, center and (for variables not already adjusted for body size) height and body mass index. Relations of LV EF from linear
American Heart Journal Volume 146, Number 3
Devereux et al 529
Table I. Characteristics of participants by level of ejection fraction
Age (y) Sex (% female) Smoking (%) Systolic BP (mm Hg) Diastolic BP (mm Hg) Hypertension medication (%) BMI (kg/m2) Height (cm) Fasting glucose (mg/dL) 2-Hour glucose (mg/dL) Plasma creatinine (mg/dL) Albumin/creatinine (log10) LDL cholesterol (mg/dL) HDL cholesterol (mg/dL) C-Reactive protein (mg/dL) Fibrinogen (mg/dL)
Normal EF (n ⴝ 2598)
Mild dysfunction (n ⴝ 293)
Severe dysfunction (n ⴝ 57)
59 ⫾ 8 68 31 129 ⫾ 20 75 ⫾ 10 29 31.1 ⫾ 6.2 164 ⫾ 9 153 ⫾ 79 170 ⫾ 96 0.97 ⫾ 0.73 3.16 ⫾ 1.92 119 ⫾ 34 41.6 ⫾ 13.2 6.49 ⫾ 9.19 360 ⫾ 83
60 ⫾ 8 42‡ 36 134 ⫾ 23‡ 77 ⫾ 11* 36* 31.3 ⫾ 6.7 168 ⫾ 9‡ 153 ⫾ 74 170 ⫾ 91 1.31 ⫾ 1.60‡ 3.67 ⫾ 2.27‡ 116 ⫾ 34 40.9 ⫾ 13.7 8.31 ⫾ 13.76* 370 ⫾ 84
62 ⫾ 9 54* 27 142 ⫾ 26‡§ 77 ⫾ 10‡ 45* 27.9 ⫾ 4.2†㛳 165 ⫾ 9§ 150 ⫾ 70 159 ⫾ 76 2.74 ⫾ 3.10‡¶ 5.23 ⫾ 2.60‡¶ 105 ⫾ 37* 46.4 ⫾ 19.5*§ 9.03 ⫾ 11.30 391 ⫾ 83*
BMI, Body mass index; BP, blood pressure. *Statistical significance P versus group with normal EF, ⬍.05. †Statistical significance P versus group with normal EF, ⬍.01. ‡Statistical significance P versus group with normal EF, ⬍.001. §Statistical significance P versus group with mild dysfunction, ⬍.05. 㛳Statistical significance P versus group with mild dysfunction, ⬍.01. ¶Statistical significance P versus group with mild dysfunction, ⬍.001.
dimensions or wall motion scores and of segmental or global LV dysfunction to adverse outcomes were assessed by Cox proportional hazards analysis with the same covariates. Survival curves were plotted to display outcome in groups defined by LV function. Two-tailed P ⬍ .05 was considered significant.
Results Characteristics of participants Of 3501 participants who underwent echocardiography, 269 had prevalent coronary heart disease (myocardial infarction or imaging or angiographic evidence of coronary obstruction verified by medical record review by a physician panel) and were excluded; 2948 (91%) of the remainder had LV EF measurements needed for the present study. Excluded individuals were older (mean 62 vs 60 years), heavier (body mass index 32.0 ⫾ 7.7 vs 31.0 ⫾ 6.2 kg/m2) and had worse lung function (forced expiratory volume in 1 second 74% vs 76% of predicted) (all P ⬍ .05). Women comprised 1910 (65%) participants: 1045 lived in Arizona, 984 in Oklahoma and 920 in North and South Dakota; 1482 (50%) had diabetes and 866 (30%) were on antihypertensive treatment. Doppler echocardiography revealed 1⫹ mitral regurgitation in 553 (18.8%), 2⫹ regurgitation in 46 (1.6%), 3⫹ regurgitation in 8 (0.3%) and 4⫹ regurgitation in 5 (0.2%); 50 (1.7%) participants had experienced congestive heart failure.
LV systolic function EF calculated from linear echocardiographic measurements was in the normal range for our laboratory (⬎54.0%) in 2598 participants (88%), mildly reduced (40%–54%) in 293 participants (10%), and severely reduced (⬍40%) in 2%; 21 (1.2%), 7 (2.4%), and 12 (21.1%) of these participants, respectively, had experienced heart failure (P ⬍ .001). Visual scoring of wall motion from 2-D images was normal in 2704 (92%) participants and revealed segmental wall motion abnormalities in 162 (6%) and global LV dysfunction in 57 (2%). Segmental and global LV dysfunction from wall motion scores were present in 2% and 0.1% of participants with normal EF from linear measurements, 25% and 15% of those with mildly subnormal linear Efs, and 70% and 21% of those with severely reduced linear EFs. EF from wall motion scores was mildly reduced in 78 participants (2.6%) and severely reduced in 33 (1.1%). Stress-corrected LV midwall shortening, a measure of myocardial function, was depressed in 295 individuals (10%).
Characteristics of participants with different levels of EF Participants with mildly and severely reduced EFs were older, disproportionately male, and had higher systolic pressure than those with normal EF (Table I). Compared to individuals with normal EF, heart rate was higher with mildly or severely reduced EF, and
American Heart Journal September 2003
530 Devereux et al
Table II. Echocardiographic findings by level of ejection fraction
Septal thickness (cm) Internal diameter (cm) Posterior wall thickness (cm) Relative wall thickness LV mass (g) LV mass/height2.7 (g/m2.7) Cardiac index (L/min/m2) Total peripheral resistance (dyne-s-cm⫺5) Pulse pressure/stroke index (mm Hg/mL/m2)
Normal EF (n ⴝ 2598)
Mild dysfunction (n ⴝ 293)
Severe dysfunction (n ⴝ 57)
0.92 ⫾ 0.002 4.89 ⫾ 0.008 0.86 ⫾ 0.002 0.35 ⫾ 0.001 153 ⫾ 0.63 40.4 ⫾ 0.169 2.52 ⫾ 10.86 1695 ⫾ 8.45 1.50 ⫾ 0.09
0.93 ⫾ 0.007 5.32 ⫾ 0.025‡ 0.86 ⫾ 0.006 0.33 ⫾ 0.003‡ 178 ⫾ 1.95‡ 46.0 ⫾ 0.523‡ 2.39 ⫾ 33.17† 1752 ⫾ 26.06 1.59 ⫾ 0.027†
0.96 ⫾ 0.016* 6.09 ⫾ 0.062‡¶ 0.91 ⫾ 0.014†㛳 0.30 ⫾ 0.007‡§ 235 ⫾ 4.77‡¶ 62.6 ⫾ 1.28‡¶ 2.23 ⫾ 81.47† 1845 ⫾ 65.43 1.74 ⫾ 0.067†
BMI, Body mass index; BP, blood pressure. *Statistical significance P versus group with normal EF, ⬍.05. †Statistical significance P versus group with normal EF, ⬍.01. ‡Statistical significance P versus group with normal EF, ⬍.001. §Statistical significance P versus group with mild dysfunction, ⬍.05. 㛳Statistical significance P versus group with mild dysfunction, ⬍.01. ¶Statistical significance P versus group with mild dysfunction, ⬍.001.
body mass index was lower with severe LV dysfunction. There was no difference among groups in fasting or 2-hour postload glucose, but there were stepwise increases from the normal EF to the severe dysfunction group in plasma creatinine and urine albumin/creatinine ratios. LDL cholesterol was lower and high-density lipoprotein cholesterol higher in participants with severe LV dysfunction, whereas C-reactive protein and fibrinogen levels tended to be elevated with reduced EFs.
Echocardiographic findings in relation to EF LV wall thicknesses and, especially, internal diameter increased progressively from participants with normal LV EF to those with mildly and severely reduced EFs (Table II). As a result, LV relative wall thickness fell parallel to lower EF, whereas LV mass and mass/ height2.7 increased stepwise with poorer LV systolic function. Cardiac index was lower and peripheral resistance tended to be higher with severe LV dysfunction; pulse pressure/stroke index, an indirect measure of arterial stiffness, was higher in participants with worse LV systolic function.
Relation of LV EF from linear dimensions to prognosis During the mean follow-up of 37 ⫾ 9 months there were a total of 204 deaths, including 63 (32%) attributed to cardiovascular causes; a total of 206 participants suffered fatal or nonfatal cardiovascular events. In univariate analyses, there were stepwise increases from the normal EF group to that with mildly or severely reduced EF in the incidences of cardiovascular death (2% to 5% and 12%, P ⬍ .001), all-cause mortality (6% to 10% and 32%, P ⬍ .001) or cardiovascular events (6% to 13% and 33%, P ⬍ .001). Of note, non-
cardiovascular death rate was substantially more common with severely reduced EF than with either normal or mildly reduced linear EF (20% vs 5% and 4%, P ⬍ .001). The small group of participants who had experienced heart failure had, compared to those without heart failure, higher rates of cardiovascular death (16% vs 1.9%) and all-cause mortality (34% vs 6.5%, both P ⬍ .001). In Cox proportional hazard analyses that adjusted for age, sex, body mass index, height, systolic pressure, antihypertensive treatment, diabetes, LDL cholesterol and cigarette smoking, the likelihoods of cardiovascular death were higher with mildly reduced LV EF (risk ratio [RR] 2.9, 95% CI 1.6 –5.4, P ⫽ .0007) and especially with severely reduced EF (RR 6.9, 95% CI 3.0 – 15.9, P ⬍ .0001), independent of associations with older age (P ⫽ .004), lower body mass index (P ⬍ .05), antihypertensive treatment (P ⫽ .01) and diabetes (P ⫽ .008) (Figure 1). In analyses with the same covariates, all-cause mortality was markedly elevated with severely reduced EF (RR 4.8, 95% CI 2.8 – 8.1, P ⬍ .0001) and marginally so with mildly reduced EF (RR 1.4, 95% CI 0.95–2.15, P ⫽ .08), independent of associations with older age (P ⬍ .0001), male sex (P ⬍ .007), lower body mass index (P ⫽ .007), antihypertensive treatment (P ⫽ .01), diabetes (P ⫽ .004) and LDL cholesterol (P ⫽ .02) (Figure 2). Considered as a continuous variable, EF from linear LV dimensions was the strongest predictor of cardiovascular mortality (Wald statistic 17.6, P ⬍ .0001); when EF from 2-D wall motion scores was substituted for that from linear dimensions, it was also the strongest predictor of cardiovascular mortality (Wald statistic 30.2, P ⬍ .0001).
American Heart Journal Volume 146, Number 3
Figure 1
Freedom from cardiovascular death (vertical axis), adjusted for covariates described in the text, is significantly lower in SHS participants with mildly reduced EF (40%–54%) and especially with severely reduced EF (⬍40%) compared to those with normal EF from linear echocardiographic LV dimensions.
Relation of LV dysfunction from wall motion scores to prognosis The proportion of SHS participants suffering cardiovascular death was higher in groups with segmental or global LV dysfunction by 2-D scores (8.5% and 8.1%) than in that with visually normal LV systolic function (1.8%, P ⬍ .001). All-cause mortality increased stepwise from the group with visually normal LV function (6.4%) to those with segmental (15.3%) or global dysfunction (22.6%) (P ⬍ .001). In an initial survival analysis that considered segmental and global LV dysfunction and the covariates enumerated above, segmental dysfunction (RR 3.3, 95% CI 1.1–9.4, P ⫽ .02), but not global dysfunction (RR 2.9, 95% CI 0.4 –21.1, P ⫽ .33), independently predicted cardiovascular death. In a similar model, all-cause mortality was associated with both segmental LV dysfunction (RR 1.7, 95% CI 1.1– 2.7, P ⫽ .023) and global LV dysfunction by visual scoring (RR 4.1, 95% CI 2.3–7.2, P ⬍ .001), independent of effects of age, sex, body mass index, antihypertensive medication, diabetes and LDL cholesterol. A final Cox regression analysis considered mildly and severely decreased EF from 2-D wall motion scores in addition to the above covariates; the likelihood of cardiovascular death was similarly increased with mildly reduced 2-D EF (RR 3.5, 95% CI 2.1–5.8) and severely reduced 2-D EF (RR 3.8, 95% CI 1.9 –7.6) (both P ⬍ .001) (Figure 3).
Devereux et al 531
Figure 2
Freedom from all-cause death (vertical axis), adjusted for covariates described in the text, is significantly lower in SHS participants with severely reduced EF (⬍40%) compared to those with normal EF or mildly reduced EF (40%–54%) from linear echocardiographic LV dimensions.
Discussion The present study provides the first population-based evidence that measurement of EF from LV linear dimensions measured from 2-D or 2-D guided M-mode echocardiography strongly predicts the subsequent likelihood of both cardiovascular and all-cause mortality independent of associations of these adverse outcomes with older age, male sex, diabetes, and other established risk factors. In a sample of ambulatory adults, free of established coronary heart disease at baseline and sufficiently healthy to participate in an extensive evaluation, there were moderate prevalences of both mildly and severely reduced EF (10% and 2%, respectively). As previously reported,6,7 reduced LV function in SHS participants was associated with older age, male sex, higher systolic blood pressure, diabetes and renal dysfunction. In analyses adjusting for these covariates, relative risks of cardiovascular and all-cause mortality were 2.9 and 1.4-fold higher in the 10% of SHS participants with mildly reduced EF and 6.9 and 4.8-fold higher in the 2% of participants with severely reduced EF. These findings suggest that the population-attributable risk of cardiovascular death associated with reduced LV EF is approximately 35%, and the population-attributable risk of all-cause mortality associated with reduced LV EF is approximately 12%. The ability to assess LV systolic function by echocardiography in a population-based sample is facilitated by the use of simple LV linear dimensions that can be
American Heart Journal September 2003
532 Devereux et al
Figure 3
Freedom from cardiovascular death (vertical axis), adjusted for covariates described in the text, is similarly lower in SHS participants with mildly reduced EF (40%–54%) or severely reduced EF (⬍40%) compared to those with normal EF from 2-D echocardiographic wall motion scores.
obtained in a high proportion of individuals. Linear measurements of end-diastolic and end-systolic LV internal diameter were obtained from 2-D echocardiographic long-axis images— or M-mode recordings when the beam was correctly oriented and yielded better interface definition—in 91% of SHS participants in the present study and up to 98% of individuals in other population-based31 or clinical37 studies. This yield is appreciably higher than the ability to obtain LV linear measurements by classic 2-D guided M-mode techniques in about 80% of relatively unselected adults38 or to measure LV EF by 2-D planimetry in about two thirds of an elderly population-based sample.39 Wall motion scoring from 2-D images could be used to derive EF estimates in an even higher proportion (3400/ 3501 or 97%) of participants in the present study, confirming the ability to assess LV systolic function qualitatively by 2-D echocardiography in a high proportion of Cardiovascular Health Study participants.39 In addition to documentation of the prognostic significance of EF by 2 easily-applied echocardiographic methods, the present study provides evidence that segmental LV dysfunction assessed by 2-D wall motion scoring is a strong predictor of cardiovascular death in middle-aged to elderly adults who have not suffered myocardial infarction or had other definite evidence (eg, coronary angiography, revascularization procedures or ECG Minnesota code) of coronary heart dis-
ease. One possible explanation is that some members of this population, with very high prevalences of diabetes and hypertension, may have had previous, clinically unrecognized myocardial infarction. Data from Framingham have previously documented an adverse prognostic significance of clinically “silent” myocardial infarctions.40 Despite the clear evidence of a relation between several measures of LV systolic function and cardiovascular and all-cause mortality, several limitations of the present study merit consideration. First, echocardiograms were recorded from 1993 to 1995, before introduction of harmonic imaging and other advances in ultrasonic technology. Because of this, it is attractive to speculate that the yield of LV EF measurements and the accuracy of both linear dimension measurements and 2-D wall motion scores would be higher with contemporary technology. Second, studies were done in remote field centers rather than established echocardiographic laboratories, potentially leading to lower technical quality of recordings. However, extensive procedures were implemented to train field center sonographers in standardized methodology, to provide feedback from the reading center to sonographers, and to implement standardized, centralized reading procedures, with final measurements verified by investigators with experience with nearly 20,000 research echocardiograms. Finally, the present study was undertaken in a population of American Indians with high prevalences of diabetes and obesity that may not be representative of other populations. However, the prevalences of mildly and severely reduced LV EFs in the present study are almost identical to those in another population-based study in white and black adults.41 We thank the Indian Health Service facilities, Strong Heart Study participants, and participating tribal communities for their extraordinary cooperation and involvement, which made this study possible; Betty Jarvis, RN, Tauqeer Ali, MD and Marcia O’Leary for coordination of study centers; Tauqeer Ali, MD, Helen Beaty, Joan Carter, Michael Cyl, and Neil Sikes for expert echocardiogram performance; Elizabeth A. Wood for design and maintenance of computer data bases; and Virginia Burns for manuscript preparation.
References 1. Mock MB, Ringqvist I, Fischer LD, et al. Survival of medically treated patients in the coronary artery surgery study (CASS) registry. Circulation 1982;66:562–71. 2. Freeman AP, Walsh WF, Giles RW, et al. Early and long-term results of coronary artery bypass grafting with severely depressed left ventricular performance. Am J Cardiol 1984;54:749 –54.
American Heart Journal Volume 146, Number 3
3. Marantz PR, Tobin JN, Wassertheil-Smoller S, et al. Prognosis in ischemic heart disease: can you tell as much at the bedside as in the nuclear laboratory? Arch Intern Med 1992;152:2433–7. 4. Rashid H, Exner DV, Mirsky I, et al. Comparison of echocardiography and radionuclide angiography as predictors of mortality in patients with left ventricular dysfunction (studies of left ventricular dysfunction). Am J Cardiol 1999;84:299 –303. 5. Teichholz LE, Kreulen T, Herman MV, et al. Problems in echocardiographic volume determinations: echocardiographic-angiographic correlations in the presence or absence of asynergy. Am J Cardiol 1976;37:7–11. 6. Asanoi H, Sasayama S, Kameyama T. Ventriculoarterial coupling in normal and failing heart in humans. Circ Res 1989;65:83–93. 7. Van Reet RE, Quinones MA, Poliner LR, et al. Comparison of twodimensional echocardiography with gated radionuclide ventriculography in the evaluation of global and regional left ventricular function in acute myocardial infarction. J Am Coll Cardiol 1984;3: 243–52. 8. Rifkin RD, Koito H. Comparison with radionuclide angiography of two new geometric and four nongeometric models for echocardiographic estimation of left ventricular ejection fraction using segmental wall motion scoring. Am J Cardiol 1990;65:1485–90. 9. Koren MJ, Devereux RB, Casale PN, et al. Relation of left ventricular mass and geometry to morbidity and mortality in men and women with essential hypertension. Ann Intern Med 1991;114: 345–52. 10. Levy D, Garrison RJ, Savage DD, et al. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med 1990;322:1561– 6. 11. Liao Y, Cooper RS, McGee DL, et al. The relative effects of left ventricular hypertrophy, coronary artery disease, and ventricular dysfunction on survival among black adults. JAMA 1995;273: 1592–7. 12. Verdecchia P, Schillaci G, Borgioni C, et al. Prognostic significance of serial changes in left ventricular mass in essential hypertension. Circulation 1998;97:48 –54. 13. Lee ET, Welty TK, Fabsitz R, et al. The Strong Heart Study: a study of cardiovascular disease in American Indians: design and methods. Am J Epidemiol 1990;132:1141–55. 14. Welty TK, Lee ET, Yeh JL, et al. Cardiovascular disease risk factors among American Indians: The Strong Heart Study. Am J Epidemiol 1995;142:269 – 87. 15. WHO Expert Committee on Diabetes Mellitus: second report (Technical Report Series 646). Geneva: World Health Organization; 1980. 16. Howard BV, Lee ET, Yeh J, et al. Coronary heart disease prevalence and its risk factors among American Indians: The Strong Heart Study. Am J Epidemiol 1995;142:269 – 87. 17. Lee ET, Cowan LD, Welty TK, et al. All-cause mortality and cardiovascular disease mortality in three American Indian populations, aged 45-74 years, 1984-1988: the Strong Heart Study. Am J Epidemiol 1998;147:995–1008. 18. Howard BV, Lee ET, Cowan LD, et al. The rising tide of cardiovascular disease in American Indians: The Strong Heart Study. Circulation 1999;99:2389 –95. 19. Devereux RB, Roman MJ, de Simone G, et al. Relations of left ventricular mass to demographic and hemodynamic variables in American Indians: The Strong Heart Study. Circulation 1997;96: 1416 –23.
Devereux et al 533
20. Devereux RB, Roman MJ, Paranicas M, et al. Impact of diabetes on cardiac structure and function: The Strong Heart Study. Circulation 2000;101:2271– 6. 21. Devereux RB, Roman MJ. Evaluation of cardiac and vascular structure by echocardiography and other noninvasive techniques. In: Laragh JH, Brenner BM, editors. Hypertension: Pathophysiology, Diagnosis, Treatment. New York: Raven Press; 1995. p. 1969 – 85, 2nd Ed. 22. Sahn DJ, De Maria A, Kisslo J, et al. The Committee on M-mode Standardization of the American Society of Echocardiography: recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation 1978;58:1072– 83. 23. Schiller NB, Shah PM, Crawford M, et al. American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms: recommendations for quantitation of the left ventricle by two-dimensional echocardiography. J Am Soc Echocardiogr 1989;2:358 – 67. 24. Shiina A, Tajik AJ, Smith HC, et al. Prognostic significance of regional wall motion abnormality in patients with prior myocardial infarction: a prospective correlative study of two-dimensional echocardiography and angiography. Mayo Clin Proc 1986;61:254 – 62. 25. Devereux RB, Alonso DR, Lutas EM, et al. Echocardiographic assessment of left ventricular hypertrophy: comparison to necropsy findings. Am J Cardiol 1986;57:450 – 8. 26. Palmieri V, Dahlo ¨ f B, DeQuattro V, et al. Reliability of echocardiographic assessment of left ventricular structure and function: The PRESERVE Study. J Am Coll Cardiol 1999;34:1625–32. 27. Quinones MA, Greenberg BH, Kopelen HA, et al. 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– 44. 28. Roman MJ, Pickering TG, Schwartz JE, et al. The association of carotid atherosclerosis and left ventricular hypertrophy. J Am Coll Cardiol 1995;25:83–90. 29. Devereux RB, Roman MJ, Liu JE, et al. Congestive heart failure despite normal left ventricular systolic function in a populationbased sample: The Strong Heart Study. Am J Cardiol 2000;86: 1090 – 6. 30. Devereux RB, Roman MJ, Paranicas M, et al. A population-based assessment of left ventricular systolic dysfunction in middle-aged and older adults: The Strong Heart Study. Am Heart J 2001;147: 439 – 46. 31. Palmieri V, Bella JN, Arnett DK, et al. Effect of type 2 diabetes mellitus on left ventricular geometry and systolic function in hypertensive subjects: the Hypertension Genetic Epidemiology Network (HyperGEN) Study. Circulation 2001;103:102–7. 32. de Simone G, Devereux RB, Roman MJ, et al. Assessment of left ventricular function by the mid-wall fractional shortening-end-systolic stress relation in human hypertension. J Am Coll Cardiol 1994;23:1444 –51. 33. Gaasch WH, Zile MR, Hosino PK, et al. Stress-shortening relations and myocardial blood flow in compensated and failing canine hearts with pressure-overload hypertrophy. Circulation 1989;79: 872–3. 34. Shimizu G, Hirota Y, Kita Y, et al. Left ventricular midwall mechanics in systemic arterial hypertension. Myocardial function is depressed in pressure-overload hypertrophy. Circulation 1991;83: 1676 – 84.
534 Devereux et al
35. Devereux RB, de Simone G, Pickering TG, et al. Relation of left ventricular midwall function to cardiovascular risk factors and arterial structure and function. Hypertension 1998;31:929 –36. 36. Devereux RB, Roman MJ, Lee ET, et al. Differences in echocardiographic findings and systemic hemodynamics among non-diabetic American Indians in different regions: The Strong Heart Study. Ann Epidemiol 2000;10:324 –32. 37. Devereux RB, Roman MJ, Palmieri V, et al. Left ventricular wall stresses and stress-mass-heart rate products in hypertensive patients with electrocardiographic left ventricular hypertrophy: The LIFE Study. J Hypertens 2000;18:1129 –38. 38. Hammond IW, Devereux RB, Alderman MH, et al. The prevalence and correlates of echocardiographic left ventricular hypertrophy
American Heart Journal September 2003
among employed patients with uncomplicated hypertension. J Am Coll Cardiol 1986;7:639 –50. 39. Gardin JM, Siscovick D, Anton-Culver H, et al. Sex, age and disease affect left ventricular mass and systolic function in the freeliving elderly: The Cardiovascular Health Study. Circulation 1995; 91:1739 – 48. 40. Kannel WB, Abbott RD. A prognostic comparison of asymptomatic left ventricular hypertrophy and unrecognized myocardial infarction: The Framingham Study. Am Heart J 1986;111:391–7. 41. Devereux RB, Bella JN, Palmieri V, et al. Prevalence and correlates of left ventricular systolic dysfunction in a bi-racial sample of hypertensive adults: The HyperGEN Study. Hypertension 2001;38: 417–23.
Background
Although echocardiography is commonly used to assess left ventricular (LV) systolic function, few data are available concerning the prognostic significance of LV ejection fraction (EF) calculated from linear echocardiographic measurements or 2-dimensional (2-D) wall motion scores in population-based samples.
Methods Echocardiography was used in the second Strong Heart Study (SHS) examination to calculate LV EF in 2948 American Indians without prevalent coronary heart disease; 2923 had 2-D wall motion scores. Results
Mildly and severely reduced LV EF occurred in 10% and 2% of participants, was associated with older age, male sex, higher systolic pressure, heart rate and markers of renal disease and inflammation. During 37 ⫾ 9 months follow-up, cardiovascular death occurred in 2%, 5% and 12% of participants with normal, mildly reduced and severely reduced EF; all cause mortality rates were 6%, 10% and 32% (both P ⬍ .001). In Cox proportional hazards analyses, adjusting for covariates, cardiovascular death was higher with mildly reduced EF (risk ratio [RR] 2.9, 95% CI 1.6 –5.4, P ⫽ .0007) and especially with severely reduced EF (RR 6.9, 95% CI 3.0 –15.9, P ⬍ .0001); all-cause mortality was increased with severe LV dysfunction (RR 4.8, 95% CI 2.8 – 8.1, P ⬍ .001) and marginally with mildly reduced EF (odds ratio 1.4, 95% CI 0.95–2.15, P ⫽ .08). Segmental LV dysfunction and mildly and severely reduced EF from 2-D wall motion scores were associated with 3.3-fold (95% CI 1.1–9.4, P ⫽ .02), 3.5-fold (95% CI 2.1–5.8) and 3.8-fold (95% CI 1.9 –7.6) (all P ⬍ .001) increased rates of cardiovascular death.
Conclusions LV EF from linear echocardiographic measurements as well as segmental LV dysfunction and EF from 2-D wall motion scores strongly and independently predict cardiovascular mortality. Reduced EF by simple echocardiographic method has estimated population-attributable risks of about 35% for cardiovascular death and 12% for all-cause mortality in a population-based sample of middle-aged to elderly adults. (Am Heart J 2003;146:527–34.) It is well recognized that left ventricular (LV) ejection fraction (EF) bears a strong negative relation to rates of cardiovascular morbidity and mortality. The initial large-scale studies documenting this relationship utilized LV angiograms performed at cardiac catheterFrom aCornell Medical Center, New York, NY, bUniversity of Oklahoma Health Sciences Center, Oklahoma City, Okla, cMissouri Breaks Industries Research, Timber Lake, SD, dNational Heart Lung and Blood Institute, Bethesda, Md, eMedStar Research Institute, Washington, DC, andfMayo Clinic, Rochester, Minn. Supported in part by grants U01-HL41642, U01-HL41652 and U01-HL41654 from the National Heart, Lung and Blood Institute and M10RR0047-34 (GCRC) from the National Institutes of Health, Bethesda, Md. This manuscript presents views of the authors and not necessarily those of the Indian Health Service. Submitted October 4, 2002; accepted February 5, 2003. Reprint requests: Richard B. Devereux, MD, New York Presbyterian Hospital, 525 East 68th Street, NY, NY 10021. E-mail: [email protected] © 2003, Mosby, Inc. All rights reserved. 0002-8703/2003/$30.00 ⫹ 0 doi:10.1016/S0002-8703(03)00229-1
ization and coronary arteriography.1,2 Subsequent studies have shown that LV EF measured by nuclear angiography or 2-dimensional (2-D) echocardiography also strongly predicts cardiovascular events.3,4 However, large-scale application of these techniques in clinical practice or population-based research is limited by the need for radiation exposure and variably complex and time-consuming methods of measurement. In clinical practice, estimates of LV systolic function are commonly derived from simple echocardiographic measurements of linear LV dimensions or from 2-D regional wall motion scores, which have been validated by comparison with contrast or nuclear angiographic reference standards.5– 8 Numerous studies have shown that measurements of LV mass from linear echocardiographic LV dimensions predict cardiovascular events and death,9 –12 but no data exist on the prognostic significance of LV EF calculated from simple LV linear dimension measurements or from 2-D echocar-
American Heart Journal September 2003
528 Devereux et al
diographic wall motion scores. Accordingly, the present study was undertaken to relate baseline measurements of EF by these echocardiographic methods to event rates in the population-based cohort of American Indian participants in the Strong Heart Study (SHS).
Methods The SHS is a population-based survey of cardiovascular risk factors and prevalent and incident cardiovascular disease in American Indians. As previously described,13,14 members, aged 45 to 74 years, of 13 tribes in Arizona, Oklahoma and North and South Dakota were recruited (overall participation rate 62%) for a first examination from 1989 to 1992. The second SHS examination was conducted from 1993 to 1995 to assess change over time in body habitus, blood pressure and other baseline measures and to add echocardiography among 3630 participants (89% of survivors). Standardized measurements of seated blood pressure; aspects of body habitus including body mass index, waist/hip ratio and percent body fat by bioelectric impedance; fasting glucose, insulin, lipid and lipoprotein concentrations; and 2-hour glucose tolerance test and glycosylated hemoglobin levels were obtained. Diabetes mellitus was diagnosed by World Health Organization criteria15; individuals with normal fasting sugars include those with normal or impaired glucose tolerance. Prevalent definite coronary heart disease was identified by published criteria.16 Deaths were identified through regular review of hospital, governmental and Indian Health Service records and regular contact with participants and their informants, with vital status known for 99.8% of cohort members. Causes of death were classified as cardiovascular or due to other causes by an expert physician panel on the basis of death certificates, medical records and informant interviews, as previously described.17,18
Echocardiographic methods Imaging and Doppler echocardiograms were performed as previously described.19,20 A standardized protocol was followed under which the parasternal acoustic window was used to record ⱖ10 consecutive beats of 2-D and M-mode recordings of LV internal diameter and wall thicknesses just below the mitral leaflet tips in long and short-axis views, long-axis views of the mitral valve and color flow recordings to search for mitral and aortic regurgitation, and 2-D longaxis views of the aortic root. The apical window was used to record ⱖ10 cycles of 2 and 4-chamber images and color Doppler recordings to assess LV wall motion and identify valvular regurgitation.
Echocardiographic measurements Correct orientation of planes for imaging and Doppler recordings was verified as previously described.21 Measurements of LV internal dimension and septal and posterior wall thicknesses were made at end-diastole and end-systole by American Society of Echocardiography recommendations22,23 on up to 3 cardiac cycles. Segmental wall motion was graded on a scale from normal to dyskinetic in 5 segments at the
base of the left ventricle, 5 at mid-cavity level and 4 at the apex.24
Calculation of derived variables End-diastolic LV dimensions were used to calculate LV mass by a formula shown to predict necropsy LV weight accurately (r ⫽ 0.90, P ⬍ .001)25 and to show good reproducibility ( ⫽ 0.93, P ⬍ .001) in a separate series of 183 hypertensive patients.26 LV mass from 2-D echocardiographic linear measurements has also been shown to have strong predictive value for cardiovascular death.27 Relative wall thickness was calculated as posterior wall thickness/internal radius. Systolic fractional shortening in percent of the LV internal dimension and end-systolic wall stress were calculated by standard methods.21 End-diastolic and end-systolic LV volumes, calculated by the Teichholz method,5,6 were used to calculate LV ejection fraction (EF). Partition values were used to separate SHS participants with normal LV EF from those with mild LV dysfunction (EF ⬍54%, the lower limit of the 95% CI in 362 apparently normal members of a reference population28), and those with mild LV dysfunction from those with severe dysfunction (EF ⬍40%).29,30 To estimate EF from segmental wall motion, normal segments were assigned a value of 4.5, 3.5 to mildly hypokinetic, 2.5 to moderately hypokinetic, 1.5 to severely hypokinetic and 0 to akinetic segments.31 Values were summed, with adjustment if needed for unscorable segments, resulting in an EF of 63% if all segments were normal, 49% if all were mildly hypokinetic, 35% if all were moderately hypokinetic and 21% if all were severely hypokinetic. The lower limit of the 95% CI for EF from wall motion scores in normal-weight, normotensive, nondiabetic SHS participants was 51%; EFs of 40% to 50% were considered mildly reduced, whereas 2-D EFs ⬍40% were considered severely reduced.
Measures of myocardial performance The primary approach to assess myocardial contractile efficiency was examination of LV midwall shortening in relation to circumferential end-systolic stress measured at the level of the LV midwall.32–34 Midwall shortening from echocardiographic measurements was expressed as a percent of the value predicted from circumferential end-systolic stress using equations derived from previously-studied normal subjects32; this variable is termed stress-corrected midwall shortening.35
Statistical analyses Data are expressed as mean ⫾ SD. Differences between groups were assessed by ANOVA with the Scheffe´ test. Because of differences between SHS participants in Arizona, Oklahoma and North and South Dakota for clinical and echocardiographic variables,14,36 indicator variables comparing Arizona and Oklahoma participants to those in North and South Dakota were entered as covariates. Independence of differences in echocardiographic variables from effects of clinical covariates was assessed using general factorial ANOVA and the Sidak test with consideration of age, sex, systolic pressure, antihypertensive treatment, diabetes, lowdensity lipoprotein [LDL] cholesterol, cigarette smoking, center and (for variables not already adjusted for body size) height and body mass index. Relations of LV EF from linear
American Heart Journal Volume 146, Number 3
Devereux et al 529
Table I. Characteristics of participants by level of ejection fraction
Age (y) Sex (% female) Smoking (%) Systolic BP (mm Hg) Diastolic BP (mm Hg) Hypertension medication (%) BMI (kg/m2) Height (cm) Fasting glucose (mg/dL) 2-Hour glucose (mg/dL) Plasma creatinine (mg/dL) Albumin/creatinine (log10) LDL cholesterol (mg/dL) HDL cholesterol (mg/dL) C-Reactive protein (mg/dL) Fibrinogen (mg/dL)
Normal EF (n ⴝ 2598)
Mild dysfunction (n ⴝ 293)
Severe dysfunction (n ⴝ 57)
59 ⫾ 8 68 31 129 ⫾ 20 75 ⫾ 10 29 31.1 ⫾ 6.2 164 ⫾ 9 153 ⫾ 79 170 ⫾ 96 0.97 ⫾ 0.73 3.16 ⫾ 1.92 119 ⫾ 34 41.6 ⫾ 13.2 6.49 ⫾ 9.19 360 ⫾ 83
60 ⫾ 8 42‡ 36 134 ⫾ 23‡ 77 ⫾ 11* 36* 31.3 ⫾ 6.7 168 ⫾ 9‡ 153 ⫾ 74 170 ⫾ 91 1.31 ⫾ 1.60‡ 3.67 ⫾ 2.27‡ 116 ⫾ 34 40.9 ⫾ 13.7 8.31 ⫾ 13.76* 370 ⫾ 84
62 ⫾ 9 54* 27 142 ⫾ 26‡§ 77 ⫾ 10‡ 45* 27.9 ⫾ 4.2†㛳 165 ⫾ 9§ 150 ⫾ 70 159 ⫾ 76 2.74 ⫾ 3.10‡¶ 5.23 ⫾ 2.60‡¶ 105 ⫾ 37* 46.4 ⫾ 19.5*§ 9.03 ⫾ 11.30 391 ⫾ 83*
BMI, Body mass index; BP, blood pressure. *Statistical significance P versus group with normal EF, ⬍.05. †Statistical significance P versus group with normal EF, ⬍.01. ‡Statistical significance P versus group with normal EF, ⬍.001. §Statistical significance P versus group with mild dysfunction, ⬍.05. 㛳Statistical significance P versus group with mild dysfunction, ⬍.01. ¶Statistical significance P versus group with mild dysfunction, ⬍.001.
dimensions or wall motion scores and of segmental or global LV dysfunction to adverse outcomes were assessed by Cox proportional hazards analysis with the same covariates. Survival curves were plotted to display outcome in groups defined by LV function. Two-tailed P ⬍ .05 was considered significant.
Results Characteristics of participants Of 3501 participants who underwent echocardiography, 269 had prevalent coronary heart disease (myocardial infarction or imaging or angiographic evidence of coronary obstruction verified by medical record review by a physician panel) and were excluded; 2948 (91%) of the remainder had LV EF measurements needed for the present study. Excluded individuals were older (mean 62 vs 60 years), heavier (body mass index 32.0 ⫾ 7.7 vs 31.0 ⫾ 6.2 kg/m2) and had worse lung function (forced expiratory volume in 1 second 74% vs 76% of predicted) (all P ⬍ .05). Women comprised 1910 (65%) participants: 1045 lived in Arizona, 984 in Oklahoma and 920 in North and South Dakota; 1482 (50%) had diabetes and 866 (30%) were on antihypertensive treatment. Doppler echocardiography revealed 1⫹ mitral regurgitation in 553 (18.8%), 2⫹ regurgitation in 46 (1.6%), 3⫹ regurgitation in 8 (0.3%) and 4⫹ regurgitation in 5 (0.2%); 50 (1.7%) participants had experienced congestive heart failure.
LV systolic function EF calculated from linear echocardiographic measurements was in the normal range for our laboratory (⬎54.0%) in 2598 participants (88%), mildly reduced (40%–54%) in 293 participants (10%), and severely reduced (⬍40%) in 2%; 21 (1.2%), 7 (2.4%), and 12 (21.1%) of these participants, respectively, had experienced heart failure (P ⬍ .001). Visual scoring of wall motion from 2-D images was normal in 2704 (92%) participants and revealed segmental wall motion abnormalities in 162 (6%) and global LV dysfunction in 57 (2%). Segmental and global LV dysfunction from wall motion scores were present in 2% and 0.1% of participants with normal EF from linear measurements, 25% and 15% of those with mildly subnormal linear Efs, and 70% and 21% of those with severely reduced linear EFs. EF from wall motion scores was mildly reduced in 78 participants (2.6%) and severely reduced in 33 (1.1%). Stress-corrected LV midwall shortening, a measure of myocardial function, was depressed in 295 individuals (10%).
Characteristics of participants with different levels of EF Participants with mildly and severely reduced EFs were older, disproportionately male, and had higher systolic pressure than those with normal EF (Table I). Compared to individuals with normal EF, heart rate was higher with mildly or severely reduced EF, and
American Heart Journal September 2003
530 Devereux et al
Table II. Echocardiographic findings by level of ejection fraction
Septal thickness (cm) Internal diameter (cm) Posterior wall thickness (cm) Relative wall thickness LV mass (g) LV mass/height2.7 (g/m2.7) Cardiac index (L/min/m2) Total peripheral resistance (dyne-s-cm⫺5) Pulse pressure/stroke index (mm Hg/mL/m2)
Normal EF (n ⴝ 2598)
Mild dysfunction (n ⴝ 293)
Severe dysfunction (n ⴝ 57)
0.92 ⫾ 0.002 4.89 ⫾ 0.008 0.86 ⫾ 0.002 0.35 ⫾ 0.001 153 ⫾ 0.63 40.4 ⫾ 0.169 2.52 ⫾ 10.86 1695 ⫾ 8.45 1.50 ⫾ 0.09
0.93 ⫾ 0.007 5.32 ⫾ 0.025‡ 0.86 ⫾ 0.006 0.33 ⫾ 0.003‡ 178 ⫾ 1.95‡ 46.0 ⫾ 0.523‡ 2.39 ⫾ 33.17† 1752 ⫾ 26.06 1.59 ⫾ 0.027†
0.96 ⫾ 0.016* 6.09 ⫾ 0.062‡¶ 0.91 ⫾ 0.014†㛳 0.30 ⫾ 0.007‡§ 235 ⫾ 4.77‡¶ 62.6 ⫾ 1.28‡¶ 2.23 ⫾ 81.47† 1845 ⫾ 65.43 1.74 ⫾ 0.067†
BMI, Body mass index; BP, blood pressure. *Statistical significance P versus group with normal EF, ⬍.05. †Statistical significance P versus group with normal EF, ⬍.01. ‡Statistical significance P versus group with normal EF, ⬍.001. §Statistical significance P versus group with mild dysfunction, ⬍.05. 㛳Statistical significance P versus group with mild dysfunction, ⬍.01. ¶Statistical significance P versus group with mild dysfunction, ⬍.001.
body mass index was lower with severe LV dysfunction. There was no difference among groups in fasting or 2-hour postload glucose, but there were stepwise increases from the normal EF to the severe dysfunction group in plasma creatinine and urine albumin/creatinine ratios. LDL cholesterol was lower and high-density lipoprotein cholesterol higher in participants with severe LV dysfunction, whereas C-reactive protein and fibrinogen levels tended to be elevated with reduced EFs.
Echocardiographic findings in relation to EF LV wall thicknesses and, especially, internal diameter increased progressively from participants with normal LV EF to those with mildly and severely reduced EFs (Table II). As a result, LV relative wall thickness fell parallel to lower EF, whereas LV mass and mass/ height2.7 increased stepwise with poorer LV systolic function. Cardiac index was lower and peripheral resistance tended to be higher with severe LV dysfunction; pulse pressure/stroke index, an indirect measure of arterial stiffness, was higher in participants with worse LV systolic function.
Relation of LV EF from linear dimensions to prognosis During the mean follow-up of 37 ⫾ 9 months there were a total of 204 deaths, including 63 (32%) attributed to cardiovascular causes; a total of 206 participants suffered fatal or nonfatal cardiovascular events. In univariate analyses, there were stepwise increases from the normal EF group to that with mildly or severely reduced EF in the incidences of cardiovascular death (2% to 5% and 12%, P ⬍ .001), all-cause mortality (6% to 10% and 32%, P ⬍ .001) or cardiovascular events (6% to 13% and 33%, P ⬍ .001). Of note, non-
cardiovascular death rate was substantially more common with severely reduced EF than with either normal or mildly reduced linear EF (20% vs 5% and 4%, P ⬍ .001). The small group of participants who had experienced heart failure had, compared to those without heart failure, higher rates of cardiovascular death (16% vs 1.9%) and all-cause mortality (34% vs 6.5%, both P ⬍ .001). In Cox proportional hazard analyses that adjusted for age, sex, body mass index, height, systolic pressure, antihypertensive treatment, diabetes, LDL cholesterol and cigarette smoking, the likelihoods of cardiovascular death were higher with mildly reduced LV EF (risk ratio [RR] 2.9, 95% CI 1.6 –5.4, P ⫽ .0007) and especially with severely reduced EF (RR 6.9, 95% CI 3.0 – 15.9, P ⬍ .0001), independent of associations with older age (P ⫽ .004), lower body mass index (P ⬍ .05), antihypertensive treatment (P ⫽ .01) and diabetes (P ⫽ .008) (Figure 1). In analyses with the same covariates, all-cause mortality was markedly elevated with severely reduced EF (RR 4.8, 95% CI 2.8 – 8.1, P ⬍ .0001) and marginally so with mildly reduced EF (RR 1.4, 95% CI 0.95–2.15, P ⫽ .08), independent of associations with older age (P ⬍ .0001), male sex (P ⬍ .007), lower body mass index (P ⫽ .007), antihypertensive treatment (P ⫽ .01), diabetes (P ⫽ .004) and LDL cholesterol (P ⫽ .02) (Figure 2). Considered as a continuous variable, EF from linear LV dimensions was the strongest predictor of cardiovascular mortality (Wald statistic 17.6, P ⬍ .0001); when EF from 2-D wall motion scores was substituted for that from linear dimensions, it was also the strongest predictor of cardiovascular mortality (Wald statistic 30.2, P ⬍ .0001).
American Heart Journal Volume 146, Number 3
Figure 1
Freedom from cardiovascular death (vertical axis), adjusted for covariates described in the text, is significantly lower in SHS participants with mildly reduced EF (40%–54%) and especially with severely reduced EF (⬍40%) compared to those with normal EF from linear echocardiographic LV dimensions.
Relation of LV dysfunction from wall motion scores to prognosis The proportion of SHS participants suffering cardiovascular death was higher in groups with segmental or global LV dysfunction by 2-D scores (8.5% and 8.1%) than in that with visually normal LV systolic function (1.8%, P ⬍ .001). All-cause mortality increased stepwise from the group with visually normal LV function (6.4%) to those with segmental (15.3%) or global dysfunction (22.6%) (P ⬍ .001). In an initial survival analysis that considered segmental and global LV dysfunction and the covariates enumerated above, segmental dysfunction (RR 3.3, 95% CI 1.1–9.4, P ⫽ .02), but not global dysfunction (RR 2.9, 95% CI 0.4 –21.1, P ⫽ .33), independently predicted cardiovascular death. In a similar model, all-cause mortality was associated with both segmental LV dysfunction (RR 1.7, 95% CI 1.1– 2.7, P ⫽ .023) and global LV dysfunction by visual scoring (RR 4.1, 95% CI 2.3–7.2, P ⬍ .001), independent of effects of age, sex, body mass index, antihypertensive medication, diabetes and LDL cholesterol. A final Cox regression analysis considered mildly and severely decreased EF from 2-D wall motion scores in addition to the above covariates; the likelihood of cardiovascular death was similarly increased with mildly reduced 2-D EF (RR 3.5, 95% CI 2.1–5.8) and severely reduced 2-D EF (RR 3.8, 95% CI 1.9 –7.6) (both P ⬍ .001) (Figure 3).
Devereux et al 531
Figure 2
Freedom from all-cause death (vertical axis), adjusted for covariates described in the text, is significantly lower in SHS participants with severely reduced EF (⬍40%) compared to those with normal EF or mildly reduced EF (40%–54%) from linear echocardiographic LV dimensions.
Discussion The present study provides the first population-based evidence that measurement of EF from LV linear dimensions measured from 2-D or 2-D guided M-mode echocardiography strongly predicts the subsequent likelihood of both cardiovascular and all-cause mortality independent of associations of these adverse outcomes with older age, male sex, diabetes, and other established risk factors. In a sample of ambulatory adults, free of established coronary heart disease at baseline and sufficiently healthy to participate in an extensive evaluation, there were moderate prevalences of both mildly and severely reduced EF (10% and 2%, respectively). As previously reported,6,7 reduced LV function in SHS participants was associated with older age, male sex, higher systolic blood pressure, diabetes and renal dysfunction. In analyses adjusting for these covariates, relative risks of cardiovascular and all-cause mortality were 2.9 and 1.4-fold higher in the 10% of SHS participants with mildly reduced EF and 6.9 and 4.8-fold higher in the 2% of participants with severely reduced EF. These findings suggest that the population-attributable risk of cardiovascular death associated with reduced LV EF is approximately 35%, and the population-attributable risk of all-cause mortality associated with reduced LV EF is approximately 12%. The ability to assess LV systolic function by echocardiography in a population-based sample is facilitated by the use of simple LV linear dimensions that can be
American Heart Journal September 2003
532 Devereux et al
Figure 3
Freedom from cardiovascular death (vertical axis), adjusted for covariates described in the text, is similarly lower in SHS participants with mildly reduced EF (40%–54%) or severely reduced EF (⬍40%) compared to those with normal EF from 2-D echocardiographic wall motion scores.
obtained in a high proportion of individuals. Linear measurements of end-diastolic and end-systolic LV internal diameter were obtained from 2-D echocardiographic long-axis images— or M-mode recordings when the beam was correctly oriented and yielded better interface definition—in 91% of SHS participants in the present study and up to 98% of individuals in other population-based31 or clinical37 studies. This yield is appreciably higher than the ability to obtain LV linear measurements by classic 2-D guided M-mode techniques in about 80% of relatively unselected adults38 or to measure LV EF by 2-D planimetry in about two thirds of an elderly population-based sample.39 Wall motion scoring from 2-D images could be used to derive EF estimates in an even higher proportion (3400/ 3501 or 97%) of participants in the present study, confirming the ability to assess LV systolic function qualitatively by 2-D echocardiography in a high proportion of Cardiovascular Health Study participants.39 In addition to documentation of the prognostic significance of EF by 2 easily-applied echocardiographic methods, the present study provides evidence that segmental LV dysfunction assessed by 2-D wall motion scoring is a strong predictor of cardiovascular death in middle-aged to elderly adults who have not suffered myocardial infarction or had other definite evidence (eg, coronary angiography, revascularization procedures or ECG Minnesota code) of coronary heart dis-
ease. One possible explanation is that some members of this population, with very high prevalences of diabetes and hypertension, may have had previous, clinically unrecognized myocardial infarction. Data from Framingham have previously documented an adverse prognostic significance of clinically “silent” myocardial infarctions.40 Despite the clear evidence of a relation between several measures of LV systolic function and cardiovascular and all-cause mortality, several limitations of the present study merit consideration. First, echocardiograms were recorded from 1993 to 1995, before introduction of harmonic imaging and other advances in ultrasonic technology. Because of this, it is attractive to speculate that the yield of LV EF measurements and the accuracy of both linear dimension measurements and 2-D wall motion scores would be higher with contemporary technology. Second, studies were done in remote field centers rather than established echocardiographic laboratories, potentially leading to lower technical quality of recordings. However, extensive procedures were implemented to train field center sonographers in standardized methodology, to provide feedback from the reading center to sonographers, and to implement standardized, centralized reading procedures, with final measurements verified by investigators with experience with nearly 20,000 research echocardiograms. Finally, the present study was undertaken in a population of American Indians with high prevalences of diabetes and obesity that may not be representative of other populations. However, the prevalences of mildly and severely reduced LV EFs in the present study are almost identical to those in another population-based study in white and black adults.41 We thank the Indian Health Service facilities, Strong Heart Study participants, and participating tribal communities for their extraordinary cooperation and involvement, which made this study possible; Betty Jarvis, RN, Tauqeer Ali, MD and Marcia O’Leary for coordination of study centers; Tauqeer Ali, MD, Helen Beaty, Joan Carter, Michael Cyl, and Neil Sikes for expert echocardiogram performance; Elizabeth A. Wood for design and maintenance of computer data bases; and Virginia Burns for manuscript preparation.
References 1. Mock MB, Ringqvist I, Fischer LD, et al. Survival of medically treated patients in the coronary artery surgery study (CASS) registry. Circulation 1982;66:562–71. 2. Freeman AP, Walsh WF, Giles RW, et al. Early and long-term results of coronary artery bypass grafting with severely depressed left ventricular performance. Am J Cardiol 1984;54:749 –54.
American Heart Journal Volume 146, Number 3
3. Marantz PR, Tobin JN, Wassertheil-Smoller S, et al. Prognosis in ischemic heart disease: can you tell as much at the bedside as in the nuclear laboratory? Arch Intern Med 1992;152:2433–7. 4. Rashid H, Exner DV, Mirsky I, et al. Comparison of echocardiography and radionuclide angiography as predictors of mortality in patients with left ventricular dysfunction (studies of left ventricular dysfunction). Am J Cardiol 1999;84:299 –303. 5. Teichholz LE, Kreulen T, Herman MV, et al. Problems in echocardiographic volume determinations: echocardiographic-angiographic correlations in the presence or absence of asynergy. Am J Cardiol 1976;37:7–11. 6. Asanoi H, Sasayama S, Kameyama T. Ventriculoarterial coupling in normal and failing heart in humans. Circ Res 1989;65:83–93. 7. Van Reet RE, Quinones MA, Poliner LR, et al. Comparison of twodimensional echocardiography with gated radionuclide ventriculography in the evaluation of global and regional left ventricular function in acute myocardial infarction. J Am Coll Cardiol 1984;3: 243–52. 8. Rifkin RD, Koito H. Comparison with radionuclide angiography of two new geometric and four nongeometric models for echocardiographic estimation of left ventricular ejection fraction using segmental wall motion scoring. Am J Cardiol 1990;65:1485–90. 9. Koren MJ, Devereux RB, Casale PN, et al. Relation of left ventricular mass and geometry to morbidity and mortality in men and women with essential hypertension. Ann Intern Med 1991;114: 345–52. 10. Levy D, Garrison RJ, Savage DD, et al. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med 1990;322:1561– 6. 11. Liao Y, Cooper RS, McGee DL, et al. The relative effects of left ventricular hypertrophy, coronary artery disease, and ventricular dysfunction on survival among black adults. JAMA 1995;273: 1592–7. 12. Verdecchia P, Schillaci G, Borgioni C, et al. Prognostic significance of serial changes in left ventricular mass in essential hypertension. Circulation 1998;97:48 –54. 13. Lee ET, Welty TK, Fabsitz R, et al. The Strong Heart Study: a study of cardiovascular disease in American Indians: design and methods. Am J Epidemiol 1990;132:1141–55. 14. Welty TK, Lee ET, Yeh JL, et al. Cardiovascular disease risk factors among American Indians: The Strong Heart Study. Am J Epidemiol 1995;142:269 – 87. 15. WHO Expert Committee on Diabetes Mellitus: second report (Technical Report Series 646). Geneva: World Health Organization; 1980. 16. Howard BV, Lee ET, Yeh J, et al. Coronary heart disease prevalence and its risk factors among American Indians: The Strong Heart Study. Am J Epidemiol 1995;142:269 – 87. 17. Lee ET, Cowan LD, Welty TK, et al. All-cause mortality and cardiovascular disease mortality in three American Indian populations, aged 45-74 years, 1984-1988: the Strong Heart Study. Am J Epidemiol 1998;147:995–1008. 18. Howard BV, Lee ET, Cowan LD, et al. The rising tide of cardiovascular disease in American Indians: The Strong Heart Study. Circulation 1999;99:2389 –95. 19. Devereux RB, Roman MJ, de Simone G, et al. Relations of left ventricular mass to demographic and hemodynamic variables in American Indians: The Strong Heart Study. Circulation 1997;96: 1416 –23.
Devereux et al 533
20. Devereux RB, Roman MJ, Paranicas M, et al. Impact of diabetes on cardiac structure and function: The Strong Heart Study. Circulation 2000;101:2271– 6. 21. Devereux RB, Roman MJ. Evaluation of cardiac and vascular structure by echocardiography and other noninvasive techniques. In: Laragh JH, Brenner BM, editors. Hypertension: Pathophysiology, Diagnosis, Treatment. New York: Raven Press; 1995. p. 1969 – 85, 2nd Ed. 22. Sahn DJ, De Maria A, Kisslo J, et al. The Committee on M-mode Standardization of the American Society of Echocardiography: recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation 1978;58:1072– 83. 23. Schiller NB, Shah PM, Crawford M, et al. American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms: recommendations for quantitation of the left ventricle by two-dimensional echocardiography. J Am Soc Echocardiogr 1989;2:358 – 67. 24. Shiina A, Tajik AJ, Smith HC, et al. Prognostic significance of regional wall motion abnormality in patients with prior myocardial infarction: a prospective correlative study of two-dimensional echocardiography and angiography. Mayo Clin Proc 1986;61:254 – 62. 25. Devereux RB, Alonso DR, Lutas EM, et al. Echocardiographic assessment of left ventricular hypertrophy: comparison to necropsy findings. Am J Cardiol 1986;57:450 – 8. 26. Palmieri V, Dahlo ¨ f B, DeQuattro V, et al. Reliability of echocardiographic assessment of left ventricular structure and function: The PRESERVE Study. J Am Coll Cardiol 1999;34:1625–32. 27. Quinones MA, Greenberg BH, Kopelen HA, et al. 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– 44. 28. Roman MJ, Pickering TG, Schwartz JE, et al. The association of carotid atherosclerosis and left ventricular hypertrophy. J Am Coll Cardiol 1995;25:83–90. 29. Devereux RB, Roman MJ, Liu JE, et al. Congestive heart failure despite normal left ventricular systolic function in a populationbased sample: The Strong Heart Study. Am J Cardiol 2000;86: 1090 – 6. 30. Devereux RB, Roman MJ, Paranicas M, et al. A population-based assessment of left ventricular systolic dysfunction in middle-aged and older adults: The Strong Heart Study. Am Heart J 2001;147: 439 – 46. 31. Palmieri V, Bella JN, Arnett DK, et al. Effect of type 2 diabetes mellitus on left ventricular geometry and systolic function in hypertensive subjects: the Hypertension Genetic Epidemiology Network (HyperGEN) Study. Circulation 2001;103:102–7. 32. de Simone G, Devereux RB, Roman MJ, et al. Assessment of left ventricular function by the mid-wall fractional shortening-end-systolic stress relation in human hypertension. J Am Coll Cardiol 1994;23:1444 –51. 33. Gaasch WH, Zile MR, Hosino PK, et al. Stress-shortening relations and myocardial blood flow in compensated and failing canine hearts with pressure-overload hypertrophy. Circulation 1989;79: 872–3. 34. Shimizu G, Hirota Y, Kita Y, et al. Left ventricular midwall mechanics in systemic arterial hypertension. Myocardial function is depressed in pressure-overload hypertrophy. Circulation 1991;83: 1676 – 84.
534 Devereux et al
35. Devereux RB, de Simone G, Pickering TG, et al. Relation of left ventricular midwall function to cardiovascular risk factors and arterial structure and function. Hypertension 1998;31:929 –36. 36. Devereux RB, Roman MJ, Lee ET, et al. Differences in echocardiographic findings and systemic hemodynamics among non-diabetic American Indians in different regions: The Strong Heart Study. Ann Epidemiol 2000;10:324 –32. 37. Devereux RB, Roman MJ, Palmieri V, et al. Left ventricular wall stresses and stress-mass-heart rate products in hypertensive patients with electrocardiographic left ventricular hypertrophy: The LIFE Study. J Hypertens 2000;18:1129 –38. 38. Hammond IW, Devereux RB, Alderman MH, et al. The prevalence and correlates of echocardiographic left ventricular hypertrophy
American Heart Journal September 2003
among employed patients with uncomplicated hypertension. J Am Coll Cardiol 1986;7:639 –50. 39. Gardin JM, Siscovick D, Anton-Culver H, et al. Sex, age and disease affect left ventricular mass and systolic function in the freeliving elderly: The Cardiovascular Health Study. Circulation 1995; 91:1739 – 48. 40. Kannel WB, Abbott RD. A prognostic comparison of asymptomatic left ventricular hypertrophy and unrecognized myocardial infarction: The Framingham Study. Am Heart J 1986;111:391–7. 41. Devereux RB, Bella JN, Palmieri V, et al. Prevalence and correlates of left ventricular systolic dysfunction in a bi-racial sample of hypertensive adults: The HyperGEN Study. Hypertension 2001;38: 417–23.