(Hypertension. 1996;28:8-15.)
© 1996 American Heart Association, Inc.
Articles |
Utility of New Electrocardiographic Models for Left Ventricular Mass in Older Adults
the Department of Public Health Sciences, Bowman Gray School of Medicine, Winston-Salem, NC (P.M.R., C.D.F.); Division of Epidemiology and Clinical Applications, National Heart, Lung, and Blood Institute, Bethesda, Md (T.A.M.); Departments of Medicine and Epidemiology, University of Washington, Seattle (D.S.); Division of Cardiology, Department of Internal Medicine, St Louis (Mo) University School of Medicine (S.H.Z.); Department of Medicine, University of California-Irvine (J.M.G.); Department of Biostatistics, University of Washington, Seattle (R.K.); Department of Internal Medicine, University of California School of Medicine at Davis (N.O.B.); and Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh (Pa) (A.N.).
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Several multivariate statistical models have recently been introduced for estimation of left ventricular mass from standard 12-lead electrocardiographic measurements. The validity of these algorithms has not been adequately evaluated. The objective of this investigation was to compare the associations between echocardiographic and electrocardiographic left ventricular mass values with clinical and subclinical indexes of coronary heart disease. The evaluation was performed with participants of the Cardiovascular Health Study, a population-based sample of 5201 men and women aged 65 years and older. Echocardiographic M-mode measurements of left ventricular mass were performed from videotape recordings with the use of a strictly standardized protocol. Electrocardiographic algorithms of the Novacode program and new algorithms derived from the Cardiovascular Health Study population were used for left ventricular mass prediction. Echocardiographic and electrocardiographic determinations of left ventricular mass were technically successful in 3410 (65.6%) and 5013 (96.4%) participants, respectively. The Novacode model overestimated echocardiographic left ventricular mass. Compared with the Novacode model, the new Cardiovascular Health Study electrocardiographic model, which includes adjustment for body weight, eliminated left ventricular mass prediction bias and improved the correlation between echocardiographic and electrocardiographic left ventricular mass from .33 to .54 in women and from .46 to .51 in men. Echocardiographic and electrocardiographic models both demonstrated similar and about equally strong associations with overt and subclinical disease and with risk factors for left ventricular hypertrophy. These observations demonstrate the potential utility of electrocardiographic models for left ventricular mass estimation.
Key Words: electrocardiography • echocardiography • hypertrophy • risk factors • aging • obesity
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Advances in echocardiographic methodology1 2 3 4 5 have enhanced the reliability of detection of LVH to a level higher than has been possible to achieve with conventional ECG criteria, known to have a low sensitivity for LVH.6 7 LVH detected by echocardiographic methodology has been shown to be an independent predictor of cardiovascular disease mortality and morbidity.8 9 10 Similarly, ECG manifestations of LVH are known to be associated with an increased risk of cardiovascular disease mortality, particularly when high-amplitude R waves are combined with repolarization abnormalities.11 12 13 14 15 16 17
Despite the advantages of echocardiograms, cost and operational considerations tend to limit their utility in large-scale population studies and clinical trials. There are substantial technical problems in securing echocardiographic data of sufficient quality for LVM determination, particularly in elderly subjects.18 Furthermore, there is considerable uncertainty in the use of echocardiography for LVH prevalence estimation because of different standards used for the adjustment of LVM to body size. For instance, LVH prevalence estimates in a study population from a hypertension clinic varied more than twofold with various echocardiographic thresholds and LVM indexing methods recommended in the literature.19
There have been serious efforts recently to improve ECG criteria for LVH.20 21 22 23 These efforts have also produced multivariate statistical models for the estimation of LVM on a continuous scale.22 24 25 26 However, these models have not been adequately validated in large independent population samples, particularly in elderly populations. The CHS offers a good opportunity for comparative assessment of the potential utility of ECG and echocardiographic determination of LVM in a large community-based sample of elderly adults. The primary objective of the present investigation was to evaluate correlations between ECG and echocardiographic estimates of LVM and various overt and subclinical indexes of cardiovascular disease. A parallel objective was to evaluate whether the ECG model used for LVM estimation in the CHS is sufficiently accurate to qualify as a substitute in those subgroups in which echocardiographic LVM determination was unsuccessful in the CHS and in other current and future studies in which echocardiographic data are not available.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Study Population
The present investigation used data from the baseline examination of the CHS, a National Heart, Lung, and Blood Institute-sponsored, population-based, multicenter cohort study of risk factors for coronary heart disease and stroke in men and women 65 years of age and older. The participants, representing a random sample of older adults (selected from four communities: Forsyth County, NC; Sacramento County, Calif; Washington County, Md; and Pittsburgh County, Pa) on the Health Care Financing Administration Medicare eligibility list, were enrolled between June 1989 and May 1990. The description of the CHS design has been published.27 The study went through the regular ethics review process at each participating institution, and informed consent was obtained from each participant according to the institutional guidelines. Potential participants were excluded if they were wheelchair bound, institutionalized, or receiving treatment for cancer. The cohort was 94.7% white, 4.7% black, and 0.6% other ethnic groups. Among 5201 participants enrolled, 2246 were men, with an average age of 73.3±5.8 years, and 2955 were women, with an average age of 72.4±5.4 years.
Eligible subjects giving informed consent answered standard questionnaires on personal habits and medical history (including hospitalizations, diagnoses, and cardiac procedures). Blood pressure was measured in the right arm of seated subjects with a random-zero sphygmomanometer after a 5-minute rest. The average of two measurements was used for analysis. Supine blood pressures in both arms and both ankles were measured in duplicate with a standard mercury sphygmomanometer and an 8-MHz Doppler probe. A low ratio of these measures (ankle-arm systolic pressure ratio <0.9) was used as a measure of arterial occlusive disease in the lower extremities.
Anthropometric measurements included weight and standing height. Venipuncture was performed early in the clinic visit after subjects had fasted 12 hours. Fasting serum glucose level was measured at a central laboratory. All participants except diabetics treated with insulin or oral hypoglycemic agents drank a 75-g oral glucose load, and repeat venipuncture was performed 2 hours later for measurement of postchallenge serum glucose and insulin levels.28
Carotid stenosis was defined by duplex ultrasonography and classified into one of two categories for this analysis: 0% to 49% and greater than 50%. Near and far wall maximal intimal-medial thicknesses of the carotid arteries were measured and averaged as an indicator of atherosclerosis; separate measurements were made for common and internal carotid arteries. CHS ultrasound methods and initial quality-control results have been published.29
ECG Methodology
A 12-lead resting ECG was obtained from all participants. ECG technicians were trained to make a special effort to reduce chest electrode placement errors, thereby reducing interindividual variability and improving the consistency of serial ECG recordings. Careful attention was paid to proper identification of the fourth and fifth intercostal spaces for correct level of the chest electrodes and the left midaxillary line for the V6 electrode location. In addition, a special electrode locator was used for positioning of the V4 electrode at a 45° angle between the midsternal and left midaxillary lines at the fifth intercostal space.30 Electrodes V3 and V5 were then located in a straight line halfway between electrodes V2 and V4, and V4 and V6, respectively.
The ECGs were recorded with MAC PC-DT ECG acquisition units (Marquette Electronics, Inc). A 10-second segment of simultaneous ECG leads was sampled at a rate of 250 samples per second per lead. The ECGs stored in the MAC PC units were transmitted daily to the Electrocardiographic Reading Center (EPICORE Center, Division of Cardiology, University of Alberta, Canada) for analysis and classification with the Novacode ECG measurement and classification program.31 32 Participants with electronic pacemakers (n=49) and with partially incomplete ECG records (n=39) (rejected leads because of excessive noise or artifacts) were excluded. A complete set of ECG data for LVM determination was available from 5013 participants (96.4%), and 3236 of them also had adequate-quality echocardiograms for LVM determination. The prevalence of major ECG abnormalities in the study population has been reported elsewhere.33 The Novacode program has algorithms for ECG classification according to the Minnesota Code,34 classification of LVH according to a variety of ECG criteria, and statistical multivariate models for estimation of echocardiographic LVM.24 25 The algorithms for estimation of LVM were as follows:
White and black men: LVM=-58.51+0.060 QS (III)+0.021 R (V5)-0.033 QS (V1)-0.296 Tp (aVR)+0.316 Tn (V6)+1.821 QRS.
White women: LVM=134.77+0.023 R (V5)-0.155 QS (I)+0.070 QS (V5)+0.112 Tp (V1)-0.123 Tp (V6)+0.032 R (aVL).
Black women: LVM=-90.71+0.050 R (I)-0.051 R (V1)-0.098 QS (V6)+0.522 Tn (I)+1.848 QRS+0.023 [R(V6)+QS(V2)].
The Novacode program algorithms for LVM were derived in the late 1980s, when echocardiographic LVM data became available from clinical trials, as the standard suitable for ECG models designed for estimation of LVM on a continuous scale. These models used standard linear regression methods for feature selection, with model R2 as a statistical measure of the goodness of fit.
Early test runs in the CHS population revealed that the Novacode LVM prediction model overestimated the echocardiographic LVM in both men and women and that LVM prediction accuracy was influenced by the presence of old MI and ventricular conduction defects. Therefore, it was decided to investigate whether improved LVM prediction models can be derived with the relatively large echocardiographic and ECG data files of the CHS study population. Statistical methodology used is described below ("Statistical Methods") and was the same as that used for the development of the earlier Novacode LVM algorithms. However, a simpler, reduced subset of ECG variables was used for feature selection. The variables chosen were those that have shown potential, as single variables or as combinations, in earlier studies: RaVL, SV3, RV5, SV1, TV5, TV6, JV5, and QRS duration, where R is the R wave amplitude; S, the absolute value of the S wave amplitude; T, the signed value of the T wave; and J, the absolute value of the J-point depression.
The following subgroups of CHS participants were used for this model development, selected according to the Minnesota Code (MC) criteria: (1) Normal ventricular conduction and no ECG evidence of an old MI (no MC 1.1 or 1.2; or 1.3 with 5.1, 5.2, or 5.3; and no MC 7.1, 7.2, or 7.4; n=2793); (2) anterior or lateral MI (MC 1.1 or 1.2; or 1.3 with 5.1, 5.2, or 5.3 in anterior or lateral lead group; n=85); (3) inferior MI (MC 1.1 or 1.2; or 1.3 with 5.1, 5.2, or 5.3 in inferior lead group; n=81); (4) left bundle branch block (MC 7.1; n=48); (5) right bundle branch block (MC 7.2; n=142); and (6) indeterminate type ventricular conduction delay (MC 7.4; n=82).
Echocardiographic Methodology
M-mode, two-dimensional, and color Doppler echocardiograms were performed in CHS participants during the baseline examination on super-VHS tape with a cardiac ultrasound machine (SSH-160A, Toshiba America) according to a previously published protocol.35 All studies were sent to a reading center at the University of California-Irvine, where the images were digitized and measurements made with customized computer algorithms. M-mode measurements of the left ventricle were made with the use of standards of the American Society of Echocardiography,36 and M-mode LVM was calculated with a formula previously reported by Devereux et al37 : LVM (g)=0.83[(VSTd+LVIDd+PWTd)3-(LVIDd)3]+0.60, where VSTd is ventricular septal thickness at end diastole, LVIDd is left ventricular internal dimension at end diastole, and PWTd is posterior wall thickness at end diastole.
Other Definitions
Prevalent MI, angina, and congestive heart failure were defined as positive answers to the question, "Has a doctor ever told you that you had (the particular disease)" confirmed by review of hospital or physicians' records. Subjects with major Q/QS waves on resting ECG were also considered to have prevalent MI, regardless of reported history. Coronary heart disease was defined as reported and confirmed MI or the presence of major Q/QS waves or reported and confirmed angina.
Diabetes was defined as self-report of physician-diagnosed diabetes, current use of insulin or oral hypoglycemic agents, fasting glucose level greater than 140 mg/dL, or 2-hour postload glucose level greater than 200 mg/dL. Number of medications included all current prescriptions for medications other than oral estrogen or progesterone. Hypertension was defined as self-reported physician diagnosis of hypertension plus use of antihypertensive medications, or seated blood pressure greater than 140/90 mm Hg. Hypertension was subclassified as borderline for subjects not on medications whose blood pressure was 140 to 159/90 to 95 mm Hg and as definite for those on medications or with blood pressure greater than 160/95 mm Hg.
Obesity was defined for the present investigation by the conventional limits for overweight according to body mass index: 25 kg/m2 in women and 27 in men.
Statistical Methods
Linear regression models were used for the development of improved ECG models for LVM estimation. The selection of optimal combinations of ECG features for each subset was performed by first ranking the R2 values for LVM regression on single variables, then on the combination of two, etc, with increasing dimensionality. One half (every second subject) in the group with normal conduction was assigned to the subgroup used for model development, and the other half to the subgroup used for testing of the stability and consistency of variable selection and overall LVM prediction accuracy. The other subgroup was considered too small to be divided into development and test groups. In the group with normal conduction, separate models were developed for men and women; for other models, men and women were combined, and sex was entered as a candidate with other covariates for variable selection. Lateral MI was included within anterior MI because of the small number of subjects in that subgroup (n=3).
Pearson correlation coefficients were determined for correlations between ECG and echocardiographic LVM in subgroups stratified by age, smoking status, and hypertensive status. Finally, multiple logistic regression analyses were performed with prevalent disease categories as the dependent variables for assessment of the strength and independence of associations noted with measures of LVM. All analyses were performed with the Statistical Analysis System (SAS) software.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Main characteristics of the study group are listed in Table 1
. The mean age was 72 years in women and 73 years in men. A substantial fraction of the study group was overweight (55% of the women and 37% of men), and mean body mass index was nearly equal in men and women (26.1 kg/m2 in women and 26.2 in men).
|
Success Rates of LVM Determination
The percentage of CHS participants in whom echocardiographic LVM could be determined was 65.6%. Thus, one third of CHS participants did not have adequate echocardiographic measurements for LVM analysis. This contrasts with 188 participants (3.6%) who did not have an ECG estimate of LVM available. This group included 49 participants with an artificial pacemaker, 2 with no ECG available, and 137 with partially missing ECG data (some leads rejected because of poor quality or a partially incomplete set of measurements necessary for various LVM models).
Feature Selection for LVM Models
Eight ECG variables, listed in "Statistical Methods," were considered in selecting features for the ECG LVM models. Two of these variables, RV5 and SV1, are components of the traditional Sokolow-Lyon criterion for LVH used, for instance, in the Minnesota Code.34 RaVL and SV3, in turn, are components of the Cornell voltage criteria for LVH.21 22 23 Various test runs indicated that LVM was approximately a function of the square root of body weight in both normal-weight and overweight men and women (Figure). Therefore, it was decided to include the square root of body weight among the covariates used for feature selection for the LVM models. Comparison of the relative contribution of these variables to LVM prediction (Table 2) indicated that body weight dominated in feature selection for LVM models, explaining the largest fraction of LVM variance. RaVL, SV3, and sex entered next into the model for normal conduction and no MI. The model R2 for this normal group was .32.
|
|
QRS duration did not enter into the LVM model in the normal group or in the group with indeterminate type bundle branch block, but it was a relatively prominent feature in some other models, particularly in those for lateral or anterior MI and left bundle branch block. JV5 amplitude entered among the best three features into the LVM models for anterior/lateral and inferior MI and indeterminate type ventricular conduction delay.
Final feature selection was done by considering the performance of the features in Table 2
, with some minor modifications for the choice (Table 3). JV5 measurement requires careful identification of the end point of QRS and tends to be unstable. Therefore, it was replaced by more stable TV5 or TV6 amplitudes, with little change in the model R2. It was retained, however, in the LVM model for the group with indeterminate type ventricular conduction delay. Test runs were also performed for examination of interaction terms for QRS duration with the other ECG variables, such as RaVL and SV3, as used in the Cornell product criterion for LVH.38 QRS interaction term with TV6 amplitude improved the LVM model performance in left bundle branch block, and it was incorporated into this model. Linear combinations of variables were given preference because their performance in general was as good as that for their sums or products. However, the coefficients for RaVL and SV3 were fairly similar, and they were entered as a linear sum as a single variable, as commonly used in the Cornell voltage criteria for LVH.21
|
Correlations Between Echocardiographic and ECG LVM
There was a substantial bias toward overestimation of LVM by the Novacode model, by about 47 g in men and 32 g in women. As expected, this bias was practically eliminated in the CHS ECG model developed in this same study group. The correlation between echocardiographic and ECG estimates of LVM in combined first and second half normal subgroups was .57 for the CHS model and .49 for the Novacode model (Table 4). The improvement in correlation was pronounced in normal women (.32 for the Novacode model and .48 for the CHS model) but not in normal men. The correlation between echocardiographic LVM and ECG LVM in the combined group with old MIs by ECG was .64 for the CHS model and .42 for the Novacode model and in the group with all ventricular conduction defects combined, .62 for the CHS model and .45 for the Novacode model.
|
Sex Differences and Age Trends in LVM
The mean value of the echocardiographic LVM was 135 g in women and 176 g in men, 41 g higher. The sex differences were 38 g for the CHS LVM and 56 g for the Novacode LVM; the differences were significant for all three LVM models (P<.001). Although not shown, echocardiographic LVM increased relatively little with age (P=.016), by about 8 g in women and 5 g in men from the youngest to the oldest age group. This increase was more pronounced (P<.001) in the Novacode ECG LVM, about 14 g in women and 24 g in men. There was no significant age trend in ECG LVM by the CHS model.
Association Between LVM, Body Size, and Obesity
An analysis of the functional relationship between LVM and body size using an allometric formula of the type LVM=
WßH
(where W is body weight and H is standing height) showed that body size explained approximately 9% of the total LVM variance in men and 17% in women (R2 values). It appeared that body weight was the dominant factor in these LVM prediction models and that body weight alone yielded R2 values for men and for women as high as did the optimal weight and height combination of the power function. Standing height alone explained at most 1% to 2% of LVM variation, and because of this low correlation, the use of regression methods to derive formulas for indexing of LVM to height may create artifacts in the apparent association between obesity and LVM. This functional relationship in the present study group is similar to that derived in a previous report for a healthy subgroup of CHS participants free of hypertension and clinical disease, including normal ejection fraction and absence of wall motion abnormalities, with exponent ß=0.51 for both men and women.39 These results suggest that LVM in older adults is approximately proportional to the square root of body weight.
LVM predicted from the equation LVM=Wß (Figure) indicated that these functional relationships were approximately similar in overweight and normal-weight subgroups of men and women. The closely parallel curves particularly in women in the Figure reflect the fact that the exponents for body weight differed relatively little, and this equation for LVM prediction could be reduced with little loss of overall accuracy into a uniform expression, LVM=W0.59, where the coefficient equals 10.8 for women and 12.6 for men. The R2 values for this LVM model were .17 for women and .09 for men, indicating that weight alone explained 17% of the total LVM variance in women and 9% in men.
LVM, Clinical Disease, and Risk Factors
In view of the considerable bias toward overestimation of LVM by the Novacode LVM model and the low correlation levels with this model, the remaining analyses of this investigation were limited to comparative evaluation between echocardiographic LVM and ECG LVM by the CHS model. The mean values of echocardiographic and ECG LVM showed similar trends toward increased LVM in prior MI, coronary heart disease, congestive heart failure, and hypertension, although the differences were smaller for ECG than for echocardiographic LVM (Table 5). The presence of diabetes was not associated with increased LVM by either estimate. Obesity was associated with an equal increase of LVM by both methods.
|
Multivariate analyses revealed that echocardiographic and ECG estimates of LVH had significant and similar independent associations with coronary heart disease, hypertension, congestive heart failure, and diabetes (Table 6). Separate multivariate logistic regression models derived for normal-weight and overweight subgroups demonstrated that these associations were approximately equally strong in normal-weight and obese subjects. In the combined model for normal-weight and overweight groups, overweight was not significantly associated with LVM estimates by either method.
|
150 and Those with LVM <150 for Presence vs Absence of Prevalent Disease, Hypertension, and Obesity From Multivariate Logistic Regression Models With Adjustment for Age, Sex, and Smoking |
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Utility of ECG Estimation of LVM
Our experience confirmed the technical difficulties involved in obtaining good-quality echocardiograms in elderly men and women. This problem and economic considerations limit the utility of the echocardiogram for evaluation of LVH status or for monitoring of LVH progression in hypertension or possible regression with hypertension control in the elderly. In earlier reports, adequate-quality echocardiograms were obtained in the hands of experts in a hospital setting in about 80% of the patients.40 41 In our study, the success rate was 66% in this cohort of older men and women. This performance level is comparable to the success rate in a similarly aged cohort reported for the Framingham Study.10 Technically, ECG data for use by LVM estimation models are easily available in nearly all hospital patients and participants in clinical trials and other epidemiological studies, as was the case in approximately 96% of the participants of the present study.
The overall correlations between the echocardiographic and ECG estimates of LVM were .62 for the CHS model of ECG LVM and .52 for the Novacode model. Considerably higher levels of correlation have been reported in hospital-based test populations that were used for the development of the Novacode LVM algorithms (r=.82 for men and .63 for women).25 This is probably due to the fact that hospital-based population samples tend to cover a substantially wider overall range of LVM because of the inclusion of valvular defects and often more severe levels of hypertension. The Novacode ECG model used for LVM estimation in the present study overestimated echocardiographic LVM in the CHS population. The CHS model of ECG LVM eliminated the LVM overestimation bias. Although these new ECG models improved correlation with echocardiographic LVM, these correlations still remained relatively low. A number of clinical and subclinical conditions, including obesity and emphysematous conditions expected in older individuals, may attenuate ECG voltages and limit the level of correlation that can be achieved. These conditions were not excluded from the present study group so that generalizability of the results could be retained.
Compared with the ECG features in the Novacode LVM algorithms, the new CHS models are substantially simpler in terms of their dimensionality. These new models contain just one or two ECG variables, whereas the Novacode models use six ECG variables each. This degree of simplification should facilitate the acceptance of these models over the old Novacode models.
Associations Between LVM and Prevalent Disease
Echocardiographic and ECG estimates of LVM had similar and approximately equally strong associations with clinical and subclinical conditions and LVM risk factors. This suggests that they both may be useful in the identification of subgroups at increased risk of cardiovascular disease. A separate study is warranted for determination of the long-term risk of cardiovascular disease associated with ECG LVM compared with that associated with echocardiographic LVM in order to fully determine to what extent these ECG models can provide a practical substitute for echocardiography in applications in which echocardiography is difficult to obtain.
Correlates of LVM
Several physiological, anthropometric, and pathophysiological factors are known to have a modifying effect on LVM and LVH status, including a modifying effect of the severity level and time course of hypertension and hypertension control efforts and medication status.7 42 43 44 45 46 Heart size increase with age has been documented in both normotensive and hypertensive subgroups of general North American populations.43 This increase in radiographic heart size with age was thought to reflect in part cardiac dilation and in part LVH. However, newer echocardiographic data suggest that the major change with aging is increased left ventricular wall thickness and not dilation.45 46 47 In the present study population involving men and women 65 years old and older, echocardiographic LVM changed little with age. This trend may be more pronounced in study populations involving a wider age range than was the case in the present study.
LVM, Obesity, and Body Size
The functional relationship between LVM and body weight was similar in overweight and normal-weight subgroups of men and women. Although the choice of the commonly used limits of defining overweight based on body mass index is to some extent arbitrary, the similarity of the functional relationship in normal-weight and overweight subgroups lends support for the generality of the results. This relationship could be reduced with little loss of overall accuracy into a uniform expression, LVM=W0.59, where the coefficient was 10.8 for women and 12.6 for men. The significance of this rather uniform expression for the association of LVM and body weight in all these subgroups can perhaps be best appreciated by taking the derivative of this equation with respect to body weight:
 |
 |
 |
, this relationship reveals that for each increment in body weight (
W), LVM=·0.59·W/W0.41. Thus, for the average weights from Table 1 for men and women, for each increment of 10 kg in body weight, LVM increases by approximately 11 g in women and 12 g in men. Of course, these are observations from cross-sectional data, and they do not necessarily imply that in a given individual, an increase or decrease in body weight due to a change in overweight level is associated with an equally pronounced change in LVM or progression or regression of LVH. This observation also illuminates one of the difficulties that can be expected in monitoring possible regression in echocardiographic LVM associated with weight reduction. An 11- to 12-g difference in LVM (approximately 1 SD in total population distribution) is associated with a substantial (10-kg) body weight difference, and an individual LVM change of this order of magnitude is barely within the resolution of the present-day echocardiographic methodology. In the CHS, the mean interreader variability (percent measurement differences) for echocardiographic LVM was 17% and the median variability was 14%.39 The implications of this degree of uncertainty can be appreciated by considering the observed distributions of echocardiographic LVM indexed to body weight in men and women and by altering one set of measurements randomly (one half increased and one half decreased) by 14%. The calculated R2 values for these two sets were .78 in women and .82 in men. This means that approximately 20% of the total observed echocardiographic LVM variance is due to measurement variation alone. This level of uncertainty is one of the factors limiting the level of correlation that can be achieved between echocardiographic and ECG LVM, and it is an even more severe problem in the assessment of the classification accuracy of ECG criteria for LVH because of the large degree of instability about a single chosen decision boundary in the traditionally dichotomous echocardiographic LVH classification.
Study Limitations
Our study population consisted of predominantly white (approximately 5% nonwhite) men and women 65 years old and older. Our results and the conclusions may not apply to younger age groups. Also, the issue of racial differences in the prevalence of LVH and the question of the validity of current ECG criteria for LVH among black men and women has been raised recently.47 48 Profound ethnic differences between white and black men and women in ECG amplitudes, the frontal plane QRS axis, and the evolution of ECG patterns with age have recently been demonstrated.49 These racial differences warrant serious consideration in future studies, and unquestionably, improved race-specific ECG models for LVM prediction need to be developed.
Sample size limitations also prevented adequate testing of the new LVM models derived in the present study in subgroups with bundle branch blocks and old MI, and independent testing will be necessary for full evaluation of their stability and accuracy in these categories.
|
Selected Abbreviations and Acronyms |
|---|
|
|
Acknowledgments |
|---|
This work was supported by contracts NO1-HC-85079, NO1-HC-85080, NO1-HC-85081, NO1-HC-85082, NO1-HC-85083, NO1-HC-85084, NO1-HC-85085, and NO1-HC-85086 from the National Heart, Lung, and Blood Institute. CHS Participating Institutions and Principal Staff: Forsyth County, NC, Bowman Gray School of Medicine of Wake Forest University: Gregory L. Burke, Tina Boyles, Alan Elster, Walter H. Ettinger, Curt D. Furberg, Edward Haponik, Gerardo Heiss, Dalane Kitzman, H. Sidney Klopfenstein, Margie Lamb, David S. Lefkowitz, Mary F. Lyles, Maurice B. Mittelmark, Cathy Nunn, Ward Riley, Grethe S. Tell, James F. Toole, and Beverly Tucker; Forsyth County, NC, Bowman Gray School of Medicine, Epidemiological Cardiology Research Center (EPICARE) (ECG Reading Center): Teresa Alexander, Beverly Benton, Stacey Gustafson, Margaret Mills, Lawrence Park, Farida Rautaharju, Pentti Rautaharju, Martha Smith, and Daniel Tesfaye; Sacramento County, Calif, University of California-Davis: Charles Bernick, William Boomer, Nemat Borhani, Andrew Duxbury, Mary Haan, Calvin Hirsch, Paul Kellerman, Lawrence Laslett, Marshall Lee, Virginia Poirier, John Robbins, and Marc Schenker; Washington County, Md, The Johns Hopkins University: M. Jan Busby-Whitehead, Joyce Chabot, George W. Comstock, Linda P. Fried, Joel G. Hill, Steven J. Kittner, Shiriki Kumanyika, David Levine, Joao A. Lima, Neil R. Powe, Thomas R. Price, Moyses Szklo, Melvyn Tockman, and Jeff Williamson; MRI Reading Center, Washington County, Md (Johns Hopkins): R. Nick Bryan, Carolyn C. Meltzer, Douglas Fellows, Melanie Hawkins, Patrice Holtz, Michael Kraut, Grace Lee, Larry Schertz, Earl P. Steinberg, Scott Wells, Linda Wilkins, and Nancy C. Yue; Allegheny County, Pa, University of Pittsburgh: Diane G. Ives, Charles A. Jengreis, Laurie Knepper, Lewis H. Kuller, Elaine Meilahn, Peg Meyer, Robert Moyer, Anne Newman, Richard Schulz, Vivienne E. Smith, and Sidney K. Wolfson; Echocardiography Reading Center (baseline), University of California-Irvine: Hoda Anton-Culver, Julius M. Gardin, Margaret Knoll, Tom Kurosaki, and Nathan Wong; Echocardiography Reading Center (follow-up), Georgetown Medical Center: John Gottdiener, Eva Hausner, Stephen Kraus, Judy Gay, Sue Livengood, Mary Ann Yohe, and Retha Webb; Ultrasound Reading Center, Geisinger Medical Center: Daniel O'Leary, Joseph Polak, and Laurie Funk; Central Blood Analysis Laboratory, University of Vermont: Edwin Bovill, Elaine Cornell, Mary Cushman, and Russell P. Tracy; Respiratory Sciences, University of Arizona (Tucson): Paul Enright; Coordinating Center, University of Washington, Seattle: Alice Arnold, Annette L. Fitzpatrick, Bonnie K. Lind, Richard A. Kronmal, Bruce M. Psaty, David S. Siscovick, Lynn Shemanski, Lloyd Fisher, Will Longstreth, Patricia W. Wahl, David Yanez, Paula Diehr, Maryann McBurnie, and Dwayne Reed; National Heart, Lung, and Blood Institute Project Office: Diane E. Bild, Teri A. Manolio, Peter J. Savage, and Patricia Smith.
|
Footnotes |
|---|
Reprint requests to Pentti M. Rautaharju, MD, PhD, 2000 W First St, Suite 505, Winston-Salem, NC 27104. E-mail pentti@phs.bgsm.wfu.edu.
A complete list of the participating institutions and principal staff appears at the end of this article.
Received December 1, 1995;
first decision January 30, 1996; first decision March 11, 1996;
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1. Reichek N, Devereux RB. Left ventricular hypertrophy: relationship of anatomic, echocardiographic and electrocardiographic findings. Circulation. 1981;63:1391-1398.
2.
Reichek N, Helak J, Plappert T, Sutton MS, Weber KT. Anatomic validation of left ventricular mass estimates from clinical two-dimensional echocardiography: initial results. Circulation. 1983;67:348-352.
3. Woythaler JN, Singer SL, Kwan LOL, Metzer RS, Reubner B, Bommer W, De Maria N. Accuracy of echocardiography versus electrocardiography in detecting left ventricular hypertrophy: comparison with postmortem mass measurements. J Am Coll Cardiol. 1983;2:305-311.[Abstract]
4. Devereux RB, Alonso DR, Lutas EM, Gottlieb GJ, Campo E, Sachs I, Reichek N. Echocardiographic assessment of left ventricular hypertrophy: comparison to necropsy findings. Am J Cardiol. 1986;57:450-458.[Medline] [Order article via Infotrieve]
5. Levy D, Savage DD, Garrison RJ, Anderson KM, Kannel WB, Castelli WP. Echocardiographic criteria for left ventricular hypertrophy: the Framingham Heart Study. Am J Cardiol. 1987;59:956-960.[Medline] [Order article via Infotrieve]
6. Crow RS, Prineas RJ, Rautaharju P, Hannan P, Liebson PR. Relation between electrocardiography and echocardiography for left ventricular mass in mild systemic hypertension. (Results from Treatment of Mild Hypertension Study.) Am J Cardiol. 1995;75:1233-1238.[Medline] [Order article via Infotrieve]
7.
Levy D, Labib SB, Anderson KM, Christiansen JC, Kannel WB, Castelli WP. Determinants of sensitivity and specificity of electrocardiographic criteria for left ventricular hypertrophy. Circulation. 1990;81:815-820.
8. Casale PN, Devereux RB, Milner M, Zullo G, Harshfield GA, Pickering TG, Saragh JH. Value of echocardiographic left ventricular mass in predicting cardiovascular morbid events in hypertensive men. Ann Intern Med. 1986;105:173-178.
9. Levy D, Garrison RJ, Savage D, Kannel WB, Castelli WP. Left ventricular mass and incidence of coronary heart disease in an elderly cohort: the Framingham Heart Study. Ann Intern Med. 1989;110:101-107.
10. Garrison RJ, Savage DD, Kannel WB, Castelli WP. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med. 1990;322:1561-1566.[Abstract]
11. Kannel WB, Gordon T, Offut D. Left ventricular hypertrophy by electro-cardiogram: prevalence, incidence, and mortality in the Framingham Study. Ann Intern Med. 1969;71:89-105.
12. Kannel WB, Gordon T, Castelli WP, Margolis JR. Electrocardiographic left ventricular hypertrophy and the risk of coronary heart disease: the Framingham Study. Ann Intern Med. 1970;72:813-822.
13. Kannel WB. Prevalence and natural history of electrocardiographic left ventricular hypertrophy. Am J Med. 1983;75(suppl 3A):4-11.
14. Kannel WB, Abbott RD. A prognostic comparison of asymptomatic left ventricular hypertrophy and unrecognized myocardial infarction: the Framingham Study. Am Heart J. 1986;111:391-397.[Medline] [Order article via Infotrieve]
15. Kannel WB, Dannenberg AL, Levy D. Population implications of electrocardiographic left ventricular hypertrophy. Am J Cardiol. 1987;60:851-931.
16. Kreger BE, Cupples LA, Kannel WB. The electrocardiogram in prediction of sudden death: Framingham Study experience. Am Heart J. 1987;113:377-382.[Medline] [Order article via Infotrieve]
17. Savage DD. Overall risk of left ventricular hypertrophy secondary to systemic hypertension. Am J Cardiol. 1987;60:8I-12I.[Medline] [Order article via Infotrieve]
18. Savage DD, Garrison RJ, Kannel WB, Anderson SJ, Feinlieb M, Castelli WP. Considerations in the use of echocardiography in epidemiology: The Framingham Study. Hypertension. 1987;4:104-114.
19. Abergel E, Tase M, Bohlender J, Menard J, Chatellier G. Which definition for echocardiographic left ventricular hypertrophy? Am J Cardiol. 1995;75:498-502.[Medline] [Order article via Infotrieve]
20. Kansal S, Roitman DI, Sheffield LT. A quantitative relationship of electrocardiographic criteria of left ventricular hypertrophy with echocardiographic left ventricular mass: a multivariate approach. Clin Cardiol. 1983;6:456-463.[Medline] [Order article via Infotrieve]
21. Casale PN, Devereux RB, Kligfield P, Eisenberg RR, Miller DH, Chauchary BS, Phillips MC. Electrocardiographic detection of left ventricular hypertrophy: development and prospective validation of improved criteria. J Am Coll Cardiol. 1985;6:572-580.[Abstract]
22.
Casale PN, Devereux RB, Alonso DR, Camp E, Kligfield P. Improved sex-specific criteria of left ventricular hypertrophy for clinical and computer interpretation of electrocardiograms: validation with autopsy findings. Circulation. 1987;75:565-572.
23. Devereux RB, Casale PN, Eisenberg RR, Miller DH, Kligfield P. Electrocardiographic detection of left ventricular hypertrophy using echocardiographic determinants of left ventricular mass as the reference standard. J Am Coll Cardiol. 1984;3:82-87.[Abstract]
24. Rautaharju PM, LaCroix AZ, Savage DD, Haynes S, Madans JH, Wolf HK, Hadden W, Keller J, Cornoni-Huntl J. Electrocardiographic estimate of left ventricular mass versus radiographic cardiac size and the risk of cardiovascular disease mortality in the epidemiologic follow-up study of the First National Health and Nutrition Examination Survey. Am J Cardiol. 1988;62:59-66.[Medline] [Order article via Infotrieve]
25. Wolf HK, Burggraf GW, Cuddy E, Rautaharju PM, Smith ER, Warren JW. Prediction of left ventricular mass from the electrocardiogram. J Electrocardiol. 1991;24:121-127.[Medline] [Order article via Infotrieve]
26. Kornreich F, Montague TJ, van Herpen G, Rautaharju PM, Smets P, Dramaix M. Improved prediction of left ventricular mass by regression analysis of body surface potential maps. Am J Cardiol. 1990;66:485-492.[Medline] [Order article via Infotrieve]
27. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A, O'Leary DH, Rautaharju PM, Tracy RP, Weiler PG. The Cardiovascular Health Study: design and rationale. Ann Epidemiol. 1991;1:263-276.[Medline] [Order article via Infotrieve]
28. National Diabetes Data Group. Classification and diagnosis of diabetes and other categories of glucose intolerance. Diabetes. 1979;28:1039-1057.[Medline] [Order article via Infotrieve]
29.
O'Leary DH, Polak JF, Wolfson SK Jr, Bond MG, Bommer W, Sheth S, Psaty BM, Sharrett AR, Manolio TA. The use of sonography to evaluate carotid atherosclerosis in the elderly: The Cardiovascular Health Study. Stroke. 1991;22:1155-1163.
30. Rautaharju PM, Wolf HK, Eifler WJ, Blackburn H. A simple procedure for positioning precordial ECG and VCG electrodes using an electrode locator. J Electrocardiol. 1976;9:35-40.[Medline] [Order article via Infotrieve]
31. Rautaharju PM, MacInnis PJ, Warren JW, Wolf HK, Rykers PM, Calhoun HP. Methodology of ECG interpretation in the Dalhousie Program: NOVACODE ECG classification procedures for clinical trials and population health surveys. Methods Inf Med. 1990;29:362-374.[Medline] [Order article via Infotrieve]
32. Rautaharju PM, Calhoun HP, Chaitman BR. Novacode serial ECG classification system for clinical trials and epidemiological studies. J Electrocardiol. 1992;24:163-172.
33. Furberg CD, Manolio TA, Psaty BM, Bild DE, Borhani NO, Newman A, Tabatznik B, Rautaharju PM. Major ECG abnormalities in persons aged 65 years and older (The Cardiovascular Health Study). Am J Cardiol. 1992;69:1329-1335.[Medline] [Order article via Infotrieve]
34.
Blackburn H, Keys A, Simonson E, Rautaharju PM, Punsar S. The electrocardiogram in population studies: a classification system. Circulation. 1960;21:1160-1175.
35. Gardin JM, Wong ND, Bommer W, Klopfenstein HS, Smith VE, Tabatznik B, Siskovick D, Lobodzinski S, Anton-Culver H, Manolio TA. Echocardiographic design of a multi-center investigation of free living elderly subjects: The Cardiovascular Health Study. J Am Soc Echocardiogr. 1992;1:63-72.
36.
Sahn DJ, DeMaria A, Kisslo J, Weyman A. The committee on M-mode standardization of the American Society of Echocardiography: results of a survey of echocardiographic methods. Circulation. 1978;58:1072-1083.
37. Devereux RB, Alonso DR, Lutas EM, Gottlieb GJ, Campo E, Sachs I, Reichek N. Echocardiographic assessment of left ventricular hypertrophy: comparison to necropsy findings. Am J Cardiol. 1986;57:450-458.
38. Malloy JM, Okin PM, Devereux RB, Kligfield O. Electrocardiographic detection of left ventricular hypertrophy by the simple QRS voltage-duration product. J Am Coll Cardiol. 1992;5:1180-1186.
39.
Gardin JM, Siscovick D, Anton-Sulver H, Lynch JC, Smith VE, Klopfenstein S, Bommer WJ, Fried L, O'Leary D, Manolio TA. Sex, age, and disease affect echocardiographic left ventricular mass and systolic function in the free-living elderly: The Cardiovascular Health Study. Circulation. 1995;91:1739-1748.
40. Kafka H, Burggraf GW, Milliken JA. Electrocardiographic diagnosis of left ventricular hypertrophy in the presence of left bundle branch block. Am J Cardiol. 1985;55:103-106.[Medline] [Order article via Infotrieve]
41. Wong M, Shah PM, Taylor RD. Reproducibility of left ventricular dimensions with M-mode echocardiography: effects of heart size, body position and transducer angulation. Am J Cardiol. 1981;47:1068-1074.[Medline] [Order article via Infotrieve]
42. Levy D, Anderson RM, Savage DD, Kannel WB, Christiansen JC, Castelli WP. Echocardiographically detected left ventricular hypertrophy: prevalence and risk factors: The Framingham Heart Study. Ann Intern Med. 1988;108:7-13.
43. Rautaharju PM, LaCroix AZ, Savage DD, Cox CS, Madans JH, Warren JW, Wolf HK, Hadden W. Heart size estimates indexed optimally to body and chest size, I: population standards; the effect of age and hypertensive status. Am J Noninvasive Cardiol. 1990;4:104-114.
44. MacMahon S, Collins G, Rautaharju PM, Cutler J, Neaton J, Prineas R, Crow R, Stamler J. Electrocardiographic left ventricular hypertrophy and the effects of antihypertensive drug therapy in hypertensive participants in the Multiple Risk Factor Intervention Trial. Am J Cardiol. 1989;63:202-210.[Medline] [Order article via Infotrieve]
45. Gardin JM, Henry WL, Savage DD, Ware JH, Burn C, Borer JS. Echocardiographic measurements in normal subjects: evaluation of an adult population without clinically apparent heart disease. J Clin Ultrasound. 1979;7:439-447.[Medline] [Order article via Infotrieve]
46.
Henry WL, Gardin JM, Ware JH. Echocardiographic measurements in normal subjects from infancy to old age. Circulation. 1979;62:1054-1061.
47.
Lee DK, Marantz PR, Devereux RB, Kligfield P, Alderman MH. Left ventricular hypertrophy in black and white hypertensives: standard electrocardiographic criteria overestimate racial differences in prevalence. JAMA. 1992;267:3294-3299.
48. Xie X, Liu K, Stamler J, Stamler R. Ethnic differences in electrocardiographic left ventricular hypertrophy in young and middle-aged employed American men. Am J Cardiol. 1994;73:564-567.[Medline] [Order article via Infotrieve]
49. Rautaharju PM, Zhou SH, Calhoun HP. Ethnic differences in ECG amplitudes in North American white, black and Hispanic men and women. J Electrocardiol. 1995;27:20-31.
This article has been cited by other articles:
 |
|
 |
|
  E. W. Hancock, B. J. Deal, D. M. Mirvis, P. Okin, P. Kligfield, and L. S. Gettes AHA/ACCF/HRS Recommendations for the Standardization and Interpretation of the Electrocardiogram: Part V: Electrocardiogram Changes Associated With Cardiac Chamber Hypertrophy A Scientific Statement From the American Heart Association Electrocardiography and Arrhythmias Committee, Council on Clinical Cardiology; the American College of Cardiology Foundation; and the Heart Rhythm Society Endorsed by the International Society for Computerized Electrocardiology J. Am. Coll. Cardiol., March 17, 2009; 53(11): 992 - 1002. [Full Text] [PDF]
|
|
|
|
 |
|
 E. W. Hancock, B. J. Deal, D. M. Mirvis, P. Okin, P. Kligfield, and L. S. Gettes AHA/ACCF/HRS Recommendations for the Standardization and Interpretation of the Electrocardiogram: Part V: Electrocardiogram Changes Associated With Cardiac Chamber Hypertrophy: A Scientific Statement From the American Heart Association Electrocardiography and Arrhythmias Committee, Council on Clinical Cardiology; the American College of Cardiology Foundation; and the Heart Rhythm Society: Endorsed by the International Society for Computerized Electrocardiology Circulation, March 17, 2009; 119(10): e251 - e261. [Full Text] [PDF]
|
|
|
|
 |
|
 M. Inzitari, B. L. Naydeck, and A. B. Newman Coronary Artery Calcium and Physical Function in Older Adults: The Cardiovascular Health Study J. Gerontol. A Biol. Sci. Med. Sci., October 1, 2008; 63(10): 1112 - 1118. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Mozaffarian, R. J. Prineas, P. K. Stein, and D. S. Siscovick Dietary Fish and n-3 Fatty Acid Intake and Cardiac Electrocardiographic Parameters in Humans J. Am. Coll. Cardiol., August 1, 2006; 48(3): 478 - 484. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Perego, P. Pizzocri, D. Corradi, F. Maisano, M. Paganelli, P. Fiorina, M. Barbieri, A. Morabito, G. Paolisso, F. Folli, et al. Circulating Leptin Correlates with Left Ventricular Mass in Morbid (Grade III) Obesity before and after Weight Loss Induced by Bariatric Surgery: A Potential Role for Leptin in Mediating Human Left Ventricular Hypertrophy J. Clin. Endocrinol. Metab., July 1, 2005; 90(7): 4087 - 4093. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. H. Drazner, J. E. Rame, E. K. Marino, J. S. Gottdiener, D. W. Kitzman, J. M. Gardin, T. A. Manolio, D. L. Dries, and D. S. Siscovick Increased left ventricular mass is a risk factor for the development of a depressed left ventricular ejection fraction within five years: The Cardiovascular Health Study J. Am. Coll. Cardiol., June 16, 2004; 43(12): 2207 - 2215. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. S. Gottdiener, R. L. McClelland, R. Marshall, L. Shemanski, C. D. Furberg, D. W. Kitzman, M. Cushman, J. Polak, J. M. Gardin, B. J. Gersh, et al. Outcome of Congestive Heart Failure in Elderly Persons: Influence of Left Ventricular Systolic Function: The Cardiovascular Health Study Ann Intern Med, October 15, 2002; 137(8): 631 - 639. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. S. Gottdiener, A. M. Arnold, G. P. Aurigemma, J. F. Polak, R. P. Tracy, D. W. Kitzman, J. M. Gardin, J. E. Rutledge, and R. C. Boineau Predictors of congestive heart failure in the elderly: the cardiovascular health study J. Am. Coll. Cardiol., May 1, 2000; 35(6): 1628 - 1637. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. M. Psaty, C. D. Furberg, L. H. Kuller, D. E. Bild, P. M. Rautaharju, J. F. Polak, E. Bovill, and J. S. Gottdiener Traditional Risk Factors and Subclinical Disease Measures as Predictors of First Myocardial Infarction in Older Adults: The Cardiovascular Health Study Arch Intern Med, June 28, 1999; 159(12): 1339 - 1347. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Tochikubo, E. Miyajima, T. Shigemasa, and M. Ishii Relation Between Body Fat–Corrected ECG Voltage and Ambulatory Blood Pressure in Patients With Essential Hypertension Hypertension, May 1, 1999; 33(5): 1159 - 1163. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Kuller, L. Fisher, R. McClelland, L. Fried, M. Cushman, S. Jackson, and T. Manolio Differences in Prevalence of and Risk Factors for Subclinical Vascular Disease Among Black and White Participants in the Cardiovascular Health Study Arterioscler Thromb Vasc Biol, February 1, 1998; 18(2): 283 - 293. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||