Comparison of Mortality Risk for Electrocardiographic Abnormalities in Men and Women With and Without Coronary Heart Disease (from the Cardiovascular Health Study) Pentti M. Rautaharju, MD, PhDa,*, Sijian Ge, MScb,d, Jennifer C. Nelson, PhDc,d, Emily K. Marino Larsen, MScb,d, Bruce M. Psaty, MD, PhDd,e, Curt D. Furberg, MD, PhDa, Zhu-Ming Zhang, MDa, John Robbins, MD, MHSf, John S. Gottdiener, MDg, and Paulo H.M. Chaves, MD, PhDh Mortality risk associated with electrocardiographic (ECG) abnormalities has been commonly reported to be lower in women than in men. We compared coronary heart disease (CHD) and all-cause mortality risk for ECG variables during a mean 9.1-year follow-up in 4,912 participants in the Cardiovascular Health Study who were >65 years of age. The hypothesis was that mortality risk for ECG abnormalities is not lower in women than in men. Five ECG variables were significant mortality predictors in Cox regression models that were adjusted for demographic, clinical, and medication variables. Gender differences were significant and mortality risk was higher in women for ECG estimates of left ventricular mass for both end points and for nondipolar QRS voltage for all-cause mortality. When evaluated simultaneously in multiple ECG variable risk models in subgroups that were stratified by baseline CHD status, no gender difference was significant. In the latter models, ST depression was a strong predictor of CHD mortality in groups with and without previous CHD. Other significant ECG predictors were previous myocardial infarction in the previous CHD group and nondipolar QRS voltage in the CHD-free group. Four ECG abnormalities were significant predictors of all-cause mortality in the CHD-free group, with risk increases of 18% to 50%. The risk of all-cause mortality in the previous CHD group was significantly increased for ST depression (by 64%), the ECG estimate of left ventricular mass (by 48%), and previous myocardial infarction (by 34%). In conclusion, we found no evidence that the relative risk of mortality for ECG abnormalities is lower in women than in men. © 2006 Elsevier Inc. All rights reserved. (Am J Cardiol 2006;97:309 –315)
Population studies have in general found a lower or nonsignificant prognostic value for electrocardiographic (ECG) abnormalities in women compared with men.1–5 Many studies have not had an adequate statistical power to evaluate mortality risk for specific ECG abnormalities in women because of too few fatal events in subgroups of women when stratified by coronary heart disease (CHD) status. The Department of aPublic Health Sciences, Wake Forest University, Winston-Salem, North Carolina; the bCollaborative Health Studies Coordinating Center, the cCenter for Health Studies, Group Health Cooperative, the dDepartment of Biostatistics and the eCardiovascular Health Research Unit, Departments of Medicine, Epidemiology and Health Services, University of Washington, Seattle, Washington; the fUniversity of California Davis, Sacramento, California; and the gDivision of Cardiology, University of Maryland Hospital, and the hDepartments of Medicine and Epidemiology, Johns Hopkins University, Baltimore, Maryland. Manuscript received May 27, 2005; revised manuscript received and accepted August 22, 2005. This study was supported by Contracts N01-HC-85079 through N01HC-85086, N01-HC-35129, and N01 HC-15103 from the National Heart, Lung, and Blood Institute, Bethesda, Maryland. *Corresponding author: Tel/fax: 954-385-5622. E-mail address:
[email protected] (P.M. Rautaharju). 0002-9149/06/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2005.08.046
There is an increased awareness that, with the changing age structure of the population, adverse cardiovascular events are more common in women than previously believed. A recent study reported that, at a given level of the estimated risk of heart disease, women were more likely to be considered by physicians to be at a lower risk category than men and thus less likely to receive adequate preventive counseling or therapeutic intervention.6 It seems important to reevaluate the relative mortality risk also for ECG abnormalities in older women compared with men. The hypothesis of the present investigation was that the mortality risk for ECG abnormalities is not less significant in women than in men.
Methods The Cardiovascular Health Study (CHS) is a populationbased, longitudinal study of men and women who were ⱖ65 years of age and lived in 4 communities in the United States. www.AJConline.org
310
The American Journal of Cardiology (www.AJConline.org)
The study was designed to identify factors that are associated with onset and evolution of CHD and stroke. During the baseline year, which started in June 1989, 5,201 men and women were recruited, and an expanded group of 687 African-Americans was recruited in year 5 of the study. Eligible participants gave informed consent, and the study protocols were approved by the review board of each participating institution. Participants with major ventricular conduction defects or electronic pacemakers (n ⫽ 575), with atrial fibrillation or flutter (n ⫽ 185), without electrocardiogram or with some incomplete derived ECG data (n ⫽ 216) were excluded. After these exclusions, ECG data from various special programs that were used for ECG classification and to derive a comprehensive set of quantitative ECG data were available for the present report from 4,912 of the 5,888 participants. Of this group of 4,912 men and women, 827 (16.8%) were classified as having previous CHD and the remaining 4,085 (83.2%) were considered free of CHD at the baseline of the study. Data collection: Details of the study design have been described previously.7 CHD at baseline was defined by a history of myocardial infarction, angina pectoris that was confirmed by retrospective review of hospitalization medical records,8 or self-report of coronary angioplasty or bypass surgery. All fatal events were reviewed and adjudicated by the CHS mortality review committee. In addition to death certificates, hospital records, interviews with next of kin, and for nonhospitalized patients, primary physician questionnaires were reviewed to determine cause of death. CHD death was defined as definite fatal myocardial infarction or definite fatal CHD.9 The 11-year mortality follow-up period covered in the present investigation was completed June 30, 2001, with a mean follow-up of 9.1 years. Electrocardiographic methods: Standard 12-lead electrocardiograms were recorded in all subjects at study baseline (year 5 for the expanded minority group) according to strictly standardized procedures for ECG acquisition and quality control. Special attention was paid to locating chest electrodes in precise positions.10 All electrocardiograms were processed in a central laboratory (EPICARE Center, University of Alberta, Edmonton, Alberta, Canada, and later at Wake Forest University, Winston-Salem, North Carolina). Electrocardiograms were initially processed by the Dalhousie ECG program.11 All electrocardiograms were visually inspected for technical errors and inadequate quality, and ECG wave measurements were validated on interactive graphics terminals. ECG processing was later repeated with the 2001 version of the Marquette 12-SL program (GE Marquette, Milwaukee, Wisconsin). Electrocardiograms were classified by the Novacode program12 using variables that were derived from the median complex of the Marquette measurement matrix. Different ECG variables were initially considered as candidates for risk analysis (Table 1). The QTrr interval (milliseconds) is the rate- and gender-adjusted QT interval
that is obtained as a linear function of the RR interval,13 which eliminates the substantial residual correlation with heart rate if Bazett’s formula is used. The other ECG variables from ST- and T-wave measurements are generally familiar to electrocardiographers: Cornell voltage, QRS/T angle, ST amplitude in lead V5, and T amplitudes in leads V5 and V1. The mean rather than peak amplitudes were used in this initial exploration. These were derived from orthonormal expansion using modified Chebyshev’s coefficients14,15 to obtain statistically independent variables. Three of the variables listed in Table 1 are based on QRS complex measurements: myocardial infarction by Novacode12 ECG criteria and 2 variables related to left ventricular hypertrophy, namely Cornell voltage16 (R wave in lead aVL plus S wave in lead V3) as a dichotomized (2-level) variable and ECG estimate of left ventricular mass (ECGLVM) as a continuous variable.17 The 3 remaining variables, T-wave roundness, nondipolar QRS, and T wave, were obtained from principal component analysis. T-wave roundness is the ratio of the second to the first eigenvector. It is familiar to vector cardiographers as the roundness of the spatial T-vector loop, which is sometimes called T-wave complexity.18 Nondipolar QRS and T-wave components reflect the presence of signal power not contained in the first 3 orthogonal (XYZ) components. Statistical analysis: Correlations between ECG variables were performed to investigate the potential collinearity problem, which may make it difficult to judge why a certain ECG variable was retained and some other variables were removed in the multivariable selection process. Although the correlation levels varied to a certain degree with gender and CHD status, overall the QRS complex/T angle appeared to dominate the correlation among variables that also reflect alterations of the sequence of repolarization (T amplitudes in leads V5 and V1 and T-wave roundness). Correlation between ECG-LVM and Cornell voltage was also significant, as expected. Two fatal end points, CHD mortality and all-cause mortality, were evaluated for their association with ECG variables using Cox proportional hazards models. CHD death was defined as definite fatal myocardial infarction or definite fatal CHD, which was reviewed and adjudicated by the CHS mortality review committee. For each end point, 2 models were developed: (1) single ECG variable models, with each containing 1 ECG variable, and (2) multiple ECG variable models, including all significant ECG variables (at p ⬍0.10) that were chosen by backward selection. All models were adjusted for demographic and clinical risk factors (age, gender, race, body mass index, hypertensive status, and diabetes mellitus) and drug use (diuretics,  blockers, antiarrhythmic classes 1a to 1c and 3, and calcium channel blockers). Significance of an interaction between gender and ECG variables was evaluated for each mortality end point using a model that combined gender groups and
Coronary Artery Disease/Mortality Risk for ECG Abnormalities
311
Table 1 Key demographic, clinical, and electrocardiographic variables by gender and baseline cardiovascular health status Description (units)
Demographic and clinical variables Age (yrs) African-American Body mass index (kg/m2) Hypertensive at baseline or on antihypertensives Systolic blood pressure (mm Hg) Current estrogen user (noncream) Diabetes mellitus Angina pectoris at baseline Myocardial infarction at baseline Coronary angioplasty Coronary artery bypass surgery Current smoker Electrocardiographic variables QTrr interval (ms)* QRS/T angle (°) Mean ST amplitude in lead V5 (V) Mean T amplitude in lead V5 (V) Mean T amplitude in lead V1 (V) Cornell voltage (V)† LVM (g) T-vector loop roundness (%) Nondipolar QRS voltage (V) Nondipolar T-wave voltage (V)
All Subjects (n ⫽ 4,912)
CHD Free at Baseline (n ⫽ 4,085)
Previous CHD (n ⫽ 827)
Men (n ⫽ 1,530)
Women (n ⫽ 2,555)
Men (n ⫽ 411)
Women (n ⫽ 416)
72.6 ⫾ 5.5 14.8% 26.6 ⫾ 4.7 42.9% 136 ⫾ 21.7 11.3% 15.0% 14.2% 7.9% 1.5% 3.2% 12.6%
72.8 ⫾ 5.7 14.0% 26.4 ⫾ 3.7 38.5% 136 ⫾ 21.1 — 16.1% 0% 0% 0% 0% 12.3%
72.2 ⫾ 5.3 15.2% 26.7 ⫾ 5.2 41.7% 136 ⫾ 21.8 12.1% 12.0% 0% 0% 0% 0% 13.2%
73.6 ⫾ 5.6 8.8% 26.4 ⫾ 3.7 45.3% 133 ⫾ 20.9 — 25.2% 81.5% 54.3% 10.0% 28.2% 9.2%
73.5 ⫾ 5.8 21.2% 27.4 ⫾ 5.5 64.3% 139 ⫾ 24.3 6.7% 19.8% 87.3% 39.4% 7.8% 10.4% 13.2%
420 ⫾ 17.0 74 ⫾ 33.4 24 ⫾ 42.5 128 ⫾ 82.7 55 ⫾ 63.5 1,355 ⫾ 555 150 ⫾ 27.7 17 ⫾ 15.9 49 ⫾ 20.6 9 ⫾ 4.1
421 ⫾ 16.3 81 ⫾ 32.4 32 ⫾ 43.5 136 ⫾ 81.4 81 ⫾ 60.6 1,458 ⫾ 565 170 ⫾ 23.9 15 ⫾ 14.6 52 ⫾ 20.2 8 ⫾ 3.2
418 ⫾ 17.0 67 ⫾ 30.3 24 ⫾ 38.0 135 ⫾ 77.2 37 ⫾ 54.7 1,265 ⫾ 509 135 ⫾ 19.9 17 ⫾ 16.1 45 ⫾ 19.7 10 ⫾ 4.1
424 ⫾ 17.2 95 ⫾ 36.3 9 ⫾ 49.1 88 ⫾ 87.3 89 ⫾ 75.6 1,506 ⫾ 637 174 ⫾ 24.8 19 ⫾ 17.1 57 ⫾ 22.7 8 ⫾ 3.1
421 ⫾ 18.5 78 ⫾ 37.4 7 ⫾ 48.2 96 ⫾ 96.9 42 ⫾ 66.3 1,386 ⫾ 601 140 ⫾ 21.7 20 ⫾ 17.7 50 ⫾ 20.8 10 ⫾ 4.8
Values are means ⫾ SD or percentages. * Rate- and gender-adjusted QT interval as a linear function of the RR interval. † R wave in lead aVL plus S wave in lead V3.
tested for the significance of the interaction term in the single ECG variable models. Gender interaction was significant for ECG-LVM for both end points and for QRS nondipolar voltage for all-cause mortality. For these variables with a significant interaction with gender, relative risks are listed separately for men and women. Before final risk model derivation, we decreased the total number of ECG variables from 11 to 5. This variable decrease was done to facilitate the development of more stable and parsimonious models. Specifically, we compared relative risk levels in the initial single ECG variable models and considered correlations between each pair of ECG variables. T amplitudes in leads V5 and V1 and the T-wave roundness index were correlated with the more predictive QRS/T angle and thus dropped. Similarly, Cornell voltage that correlated with the more predictive ECG-LVM was dropped at this stage. Nondipolar T-wave voltage and QTrr interval were not significant predictors in the single ECG variable models and were dropped from further consideration. From the remaining ECG variables, previous myocardial infarction, QRS/T angle, ST depression, nondipolar QRS voltage, and ECG-LVM were retained for selection of the dominant ECG mortality predictors in the multiple ECG variable models. The upper decile (gender-specific cutpoint) in the CHD-free group yielded a 10% prevalence of each abnor-
mal category in the total group; however, for previous myocardial infarction, prevalence was 8.7% in men and 4.8% in women. Relative risks and mortality rates are presented with 95% confidence intervals. The proportional hazards assumption of the Cox model was checked graphically for each variable of interest and the demographic covariates used in the models. All statistical analyses were performed with SPSS for Windows (SPSS, Inc., Chicago, Illinois).
Results Key study group characteristics (Table 1) include the ethnic distribution of the study population. Of the 4,912 participants, 827 (16.8%) were classified as having CHD and 4,085 (83.2%) were considered free of CHD at baseline. Among participants with previous CHD, nearly 3 times more men than women had undergone coronary artery bypass surgery. A considerably higher percentage of women with previous CHD than men were hypertensive (64.5% vs 45.3%, respectively), and diabetes was 4% to 5% more prevalent in men than in women. The QRS/T angle was considerably smaller in women than in men in the previous CHD group (78° vs 95°) and the CHD-free group (67° vs
312
The American Journal of Cardiology (www.AJConline.org)
Table 2 Relative risks (95% confidence intervals) of coronary heart disease and all-cause mortality from single electrocardiographic variable models* Category/Subgroup Previous myocardial infarction No infarct (no NC 5.1–5.4) Infarct (NC 5.1–5.4) ST depression in lead V5 ⬎⫺20 V (men and women) ⱕ⫺20 V‡ (men and women) QRS/T angle ⬍126° in men, ⬍107° in women ⱖ126° in men, ⱖ107° in women QRS nondipolar voltage ⬍77 V in men, ⬍68 V in women ⱖ77 V in men, ⱖ68 V in women ECG-LVM ⬍201 g in men, ⬍161 g in women ⱖ201 g in men ⱖ161 g in women
CHD Mortality† (452 deaths)
All-Cause Mortality (1,856 deaths)
1 Men and women 1.82 (1.38–2.40)¶
1 Men and Women 1.61 (1.37–1.88)¶
1 Men and women 2.00 (1.60–2.50)¶
1 Men and Women 1.62 (1.43–1.82)¶
1 Men and women 1.60 (1.26–2.02)¶
1 Men and women 1.43 (1.26–1.63)¶
1 Men and women 1.66 (1.29–2.13)储
1 Men 1.00 (0.81–1.24) Women 1.40 (1.17–1.68)¶
1 Men 1.24 (0.81–1.82) Women 2.14 (1.45–3.17)¶
1 Men 1.32 (1.06–1.65)储 Women 1.81 (1.48–2.22)¶
* Adjusted for age, gender, CHD status, race, body mass index, hypertension status, diabetes, and drug usage (diuretics,  blockers, antiarrhythmic classes 1a to 1c and 3, and calcium channel blockers). Relative risks are listed separately for men and women for ECG variables with a significant gender interaction. † Includes CHD deaths due to reinfarction among those with previous infarction. ‡ Equal to ⫺50 V at ST segment at the J point ⫹60 ms. § p ⬍0.05; 储 p ⬍0.01; ¶ p ⬍0.001, risk for individual variables. NC ⫽ Novacode.
81°). Another notable gender difference in ECG variables was that the mean T-wave amplitude in lead V1 was ⬎50% smaller in women than in men in the previous CHD and CHD-free groups. ECG predictors of CHD mortality: Statistical tests revealed that, in the single ECG variable models in combined gender groups that also combined the CHD-free and previous CHD groups, gender differences were significant for ECG-LVM for both end points and for QRS nondipolar voltage for all-cause mortality (Table 2). Relative risks for variables with a significant interaction with gender are listed separately for men and women and those for other variables are listed together for combined gender groups. Relative risk of CHD mortality for ECG-LVM was significant in women only, with a greater than twofold increase in risk. Relative risks for the other 4 ECG variables, i.e., previous myocardial infarction, ST depression in lead V5, QRS/T angle, and nondipolar QRS voltage, were considerably increased in men and women, ranging from 60% for wide QRS/T angle to a twofold increase for ST depression. For ST depression and QRS/T angle, risk levels for CHD mortality in men and women were as strong as the risk level for previous myocardial infarction by ECG criteria. With all-cause mortality as an end point, interaction with gender was significant for 2 ECG variables, namely nondipolar QRS voltage and ECG-LVM. ECG-LVM was a significant predictor in men and women, with increased risks of 32% in men and 81% in women. Nondipolar QRS voltage was a significant predictor of all-cause mortality in women only, with a risk increase of 40%. For 3 other ECG variables, i.e., previous myocardial infarction, ST depression,
and wide QRS/T angle, all-cause mortality risk also increased significantly, by approximately 40% to 60%. Dominant ECG predictors of mortality: ECG variables that were retained as significant predictors of CHD mortality when all ECG variables were considered simultaneously in multiple ECG variable models were designated as dominant ECG predictors (Table 3). A mean value ⱕ20 V for ST depression (equivalent to ⫺50 V depression at the J point ⫹60 ms) was a dominant, strong ECG predictor of CHD mortality in men and women in the CHD-free and previous CHD groups. Previous myocardial infarction by electrocardiography was the other dominant predictor of CHD mortality (including death due to reinfarction) in the previous CHD group. No ECG variable among the 5 evaluated could be identified as a clearly dominant predictor of all-cause mortality in either baseline CHD group (Table 4). In these multiple ECG variable models, 4 ECG variables were retained in the final model of the CHD-free group, with risk increases up to 50% (for previous myocardial infarction) and decreases for the other 3 variables (ST depression, ECG-LVM, and QRS/T angle). These same ECG variables, except QRS/T angle, were retained as significant in the previous CHD group.
Discussion We found no evidence to suggest that the mortality risk for ECG variables is lower in women than in men. The effect of some ECG predictors differed by gender (ECG-LVM and
Coronary Artery Disease/Mortality Risk for ECG Abnormalities
313
Table 3 Relative risks (95% confidence intervals) for predictors of coronary heart disease mortality from multiple electrocardiographic variable models* by coronary heart disease status at baseline Category/Subgroup Previous myocardial infarction No infarct (no NC 5.1–5.4) Infarct (NC 5.1–5.4) ST depression in lead V5 ⬎⫺20 V (men and women) ⱕ⫺20 V† (men and women) QRS/T angle ⬍126° in men, ⬍107° in women ⱖ126° in men, ⱖ107° in women Nondipolar QRS voltage ⬍77 V in men, ⬍68 V in women ⱖ77 V in men, ⱖ68 V in women ECG-LVM ⬍201 g in men, ⬍161 g in women ⱖ201 g in men, ⱖ161 g in women
CHD Free at Baseline (294 deaths, 7.2%)
Previous CHD (158 deaths, 19.1%)
Removed
1 Men and women 1.62 (1.13–2.30)§
1 Men and women 1.74 (1.28–2.36)储
1 Men and women 2.35 (1.67–3.30)储
Removed
Removed
Removed
Removed
Removed
Removed‡
* All 5 ECG variables were entered simultaneously into the risk model and adjusted for age, gender, CHD status, race, body mass index, hypertension status, diabetes mellitus, and drug usage (diuretics,  blockers, antiarrhythmic classes 1a to 1c and 3, and calcium channel blockers). † Equal to ⫺50 V at ST segment at the J point ⫹60 ms. ‡ p ⬍0.05; § p ⬍0.01; 储 p ⬍0.001. Removed ⫽ removed from the final risk model as nonsignificant in backward selection procedure; other abbreviation as in Table 2. Table 4 Relative risks (95% confidence intervals) for independent electrocardiographic predictors of all-cause mortality from multiple electrocardiographic variable models* by coronary heart disease status and baseline Category/Subgroup Previous myocardial infarction No infarct (no MC 5.1–5.4) Infarct (NC 5.1–5.4) ST depression in lead V5 ⬎⫺20 V (men and women) ⱕ⫺20 V† (men and women) ECG-LVM ⬍201 g in men, ⬍161 g in women ⱖ201 g in men, ⱖ161 g in women QRS/T angle ⬍126° in men, ⬍107° in women ⱖ126° in men, ⱖ107° in women Nondipolar QRS voltage ⬍77 V in men, ⬍68 V in women ⱖ77 V in men, ⱖ68 V in women
CHD Free at Baseline (1,419 deaths, 34.7%)
Previous CHD (437 deaths, 52.8%)
1 Men and women 1.50 (1.20–1.87)§
1 Men and women 1.34 (1.07–1.69)‡
1 Men and women 1.35 (1.14–1.59)§
1 Men and women 1.64 (1.31–2.04)储
1 Men and women 1.27 (1.05–1.55)§
1 Men and women 1.48 (1.14–1.98)‡
1 Men and women 1.18 (0.99–1.40)‡
Removed
Removed
Removed
* All 5 ECG variables were entered simultaneously into the risk model and adjusted for age, gender, CHD status, race, body mass index, hypertension status, diabetes mellitus, and drug usage (diuretics,  blockers, antiarrhythmic classes 1a to 1c and 3, and calcium channel blockers). † Equal to ⫺50 V at ST segment at the J point ⫹60 ms. ‡ p ⬍0.05; § p ⬍0.01; 储 p ⬍0.001. Abbreviations as in Tables 2 and 3.
nondipolar QRS voltage), and for these variables relative risk was significant in women only or was stronger in women than in men. From repolarization variables, ST depression and QRS/T angle were significant mortality predictors in men and women, with a risk level as strong as that for previous myocardial infarction by electrocardiography. Widening of the QRS/T angle and its associated increased risk occur when repolarization becomes increasingly concordant to depolarization.19,20 T-wave roundness may also
reflect altered direction of terminal T vectors.21 QRS nondipolar voltage was found to be a surprisingly strong predictor of CHD mortality in men and women. T-wave nondipolar voltage was not a significant predictor, in contrast to a previous report.22 This investigation also sought to identify dominant ECG variables that were retained as significant predictors when evaluated for mortality risk simultaneously with other significant ECG variables and adjusted for multiple demo-
314
The American Journal of Cardiology (www.AJConline.org)
graphic and clinical variables. ST depression in lead V5 was a dominant predictor of CHD mortality in CHD free men and women and was also a strong predictor of CHD death in the previous CHD group. In addition, previous myocardial infarction by electrocardiography was a dominant predictor of CHD mortality in men and women with previous CHD. The search for dominant ECG predictors of all-cause mortality was less conclusive. Reports on ECG risk predictors that have included women have commonly included younger groups, from 30 to 60 years old, and the follow-up period has generally been shorter than the 11-year follow-up in the present investigation. The prevalence of infarction-related ECG abnormalities is generally low in younger groups, and even some studies with relatively large samples of adult women have reported too few fatal end points to perform a meaningful risk analysis.1 A recent study from the Women’s Health Initiative identified QRS/T angle, ECG myocardial infarction, and increased nondipolar QRS voltage as the strongest predictors of CHD mortality, with a greater than twofold risk increase for each.23 Low heart rate variability and ST-T abnormalities were also strong CHD mortality risk predictors. That study in a large group of postmenopausal women with a 9-year follow-up used methods that were closely similar to those in the present study. An Italian co-operative multicenter project, the Italian Risk Factors and Life Expectancy (RIFLE) Pooling Project,2 included a large group of CHD-free men (n ⫽ 12,180) and women (n ⫽ 10,373) who were 30 to 69 years of age. In this CHD-free group, the prevalence of any codable Q waves was low even among subjects who were 50 to 69 years of age, namely 1.1% in men and 1.2% in women. Possible myocardial infarction was not significantly associated with CHD or all-cause mortality in men or women. It was significantly associated with mortality from cardiovascular disease in men but not in women. Mortality risk of all 3 end points was very high for major Q waves associated with major T-wave abnormalities in men. The end points were too few in women for a meaningful risk analysis. Gender differences in the prognostic value of ST-T abnormalities have been compared in some studies. In the RIFLE study,2 ST-T abnormalities were strongly associated with all mortality end points in multiple ECG variable models in men but not in women. In a Belgian study, the relative multivariate risks for CHD deaths were 3.49 in men and 2.61 in women and those for all-cause mortality were 2.52 in men and 1.89 in women.4 The Chicago Heart Association Detection Project in Industry, which assessed 9,203 men and 7,818 women who were 40 to 64 years old, found a much higher multivariable-adjusted relative risk of CHD mortality for ST-T abnormalities in men than in women.5 From more recent reports, the Copenhagen City Heart Study on a large cohort of men and women who were 35 to 74 years of age compared predictive value for 5 groups of
mutually exclusive ECG abnormalities.24 As in the present study, no significant interaction was found between gender and ECG abnormalities in the individual categories of ECG abnormalities evaluated. The multivariable-adjusted 7-year risk of ischemic heart disease in the pooled gender group was significantly increased for isolated negative T waves by 56% and for ST depression with negative T waves by greater than twofold. The risk increases for all-cause mortality were 61% for isolated negative T waves and 68% for ST depression with negative T waves. Relative risks were particularly high if repolarization abnormalities were combined with a high-amplitude QRS complex.
Appendix Steering Committee Chairman: Curt D. Furberg, MD, PhD, Wake Forest University School of Medicine. National Heart, Lung, and Blood Institute Project Office: Jean Olson, MD, MPH.
Participating Sites and Principal Investigators: Wake Forest University School of Medicine: Gregory L. Burke, MD; Wake Forest University, ECG Reading Center: Pentti M. Rautaharju, MD, PhD; University of California, Davis: John Robbins, MD, MHS; The Johns Hopkins University: Linda P. Fried, MD, MPH; The Johns Hopkins University, MRI Reading Center: Nick Bryan, MD, PhD, and Norman J. Beauchamp, MD; University of Pittsburgh: Lewis H. Kuller, MD, DrPH; University of California, Irvine, Echocardiography Reading Center (baseline): Julius M. Gardin, MD; Georgetown Medical Center, Echocardiography Reading Center (follow-up): John S. Gottdiener, MD; New England Medical Center, Boston, Ultrasound Reading Center: Daniel H. O’Leary, MD; University of Vermont, Central Blood Analysis Laboratory: Russell P. Tracy, PhD; University of Arizona, Tucson, Pulmonary Reading Center: Paul Enright, MD; Retinal Reading Center, University of Wisconsin: Ronald Klein, MD; University of Washington, Coordinating Center: Richard A. Kronmal, PhD. 1. Reunanen A, Aromaa A, Pyõrãlã K, Punsar S, Maatela J, Knekt P. The Social Insurance Institution’s Coronary Heart Disease Study. Acta Med Scand 1983;673(suppl):1–120. 2. Menotti A, Seccaraccia F, and the RIFLE Research Group. Electrocardiographic Minnesota Code findings predicting short-term mortality in asymptomatic subjects. The Italian RIFLE Pooling Project (Risk Factors and Life Expectancy). G Ital Cardiol 1997;27:40 – 49. 3. Kannel WB, Abbott R. A prognostic comparison of asymptomatic left ventricular hypertrophy and unrecognized myocardial infarction: the Framingham Study. Am Heart J 1986;111:391–397. 4. De Bacquer D, De Backer G, Kornitzer M, Blackburn H. Prognostic value of ECG findings for total, cardiovascular disease, and coronary heart disease death in men and women. Heart 1998;80:570 –577. 5. Liao Y, Liu K, Dyer A, Schoenberger JA, Shekelle RB, Collette P, Stamler J. Sex differential in the relationship of electrocardiographic ST-T abnormalities to risk of coronary death: 11.5 year follow-up
Coronary Artery Disease/Mortality Risk for ECG Abnormalities
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
findings of the Chicago Heart Association Detection Project in Industry. Circulation 1987;75:347–352. Mosca L, Linfante AH, Benjamin EJ, Berra K, Hayes SN, Walsh BW, Fabunmi RP, Kwan J, Mills T, Simpson SL. National Study of Physician Awareness and Adherence to Cardiovascular Disease Prevention guidelines. Circulation 2005;111:499 –510. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A. The Cardiovascular Health Study. Design and rationale. Ann Epidemiol 1991; 1:263–276. Psaty BM, Kuller LH, Bild D, Burke GL, Kittner SJ, Mittelmark M, Price TR, Rautaharju PM, Robbins J. Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study. Ann Epidemiol 1995;5:270 –277. Ives DG, Fitzpatrick AL, Bild DE, Psaty BM, Kuller LH, Crowley PM, Cruise RG, Theroux S. Surveillance and ascertainment of cardiovascular events: the Cardiovascular Health Study. Ann Epidemiol 1995; 5:278 –285. Rautaharju PM, Park L, Rautaharju FS, Crow R. A standardized procedure for locating and documenting ECG chest electrode positions. Consideration of the effect of breast tissue on ECG amplitudes in women. J Electrocardiol 1998;31:17–29. 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. Rautaharju PM, Park LP, Chaitman BR, Rautaharju F, Zhang ZM. The Novacode criteria for classification of electrocardiographic abnormalities and their clinically significant progression and regression. J Electrocardiol 1998;31:157–187. Rautaharju PM, Zhang ZM. Linearly scaled, rate-invariant normal limits for QT interval. Eight decades of incorrect application of power functions. J Cardiovasc Electrophysiol 2002;13:1211–1218. Rautaharju PM, Warren J, Wolf HK. Waveform vector analysis of orthogonal electrocardiograms. Quantification and data reduction. J Electrocardiol 1973;6:103–111. Rautaharju PM, Punsar S, Blackburn H, Warren J, Menotti A. Waveform patterns in Frank-lead rest and exercise electrocardiograms of healthy elderly men. Circulation 1973;48:541–548.
315
16. Casale PN, Devereux RB, Kligfield P, Eisenberg RR, Miller DH, Chaudhary BS, Phillips MC. Electrocardiographic detection of left ventricular hypertrophy: development and prospective validation of improved criteria. J Am Coll Cardiol 1985;6:572–580. 17. Rautaharju PM, Park LP, Gottdiener JS, Siscovick D, Boineau R, Smith V, Powe NR. Race- and gender-specific ECG models for left ventricular mass in older populations. Factors influencing overestimation of left ventricular hypertrophy prevalence by ECG criteria in African-Americans. J Electrocardiol 2000;33:205–218. 18. Priori SG, Mortara DW, Napolitano C, Diehl L, Paganini V, Cantu F, Cantu G, Schwartz PJ. Evaluation of the spatial aspects of T-wave complexity in the long-QT syndrome. Circulation 1997;96: 3006 –3012. 19. Kors JA, de Bruyne MC, Hoes AW, van Herpen G, Hofman A, van Bemmel JH, Groebbee DE. T axis as an independent indicator of risk of cardiac events in elderly people. Lancet 1998;352:601– 605. 20. Rautaharju PM, Clark Nelson J, Kronmal RA, Zhang Z-M, Robbins J, Gottdiener JS, Furberg CD, Manolio T, Fried L. Usefulness of T-axis deviation as an independent risk indicator for incident cardiac events in older men and women free from coronary heart disease (the Cardiovascular Health Study). Am J Cardiol 2001;88: 118 –123. 21. Okin PM, Devereux RB, Fabsitz RR, Lee ET, Galloway JM, Howard BV. Principal component analysis of the T wave and prediction of cardiovascular mortality in American Indians: the Strong Heart Study. Circulation 2002;105:714 –719. 22. Zabel M, Malik M, Hnatkova K, Papademetriou V, Pittaras A, Fletcher RD, Franz MR. Analysis of T-wave morphology from the 12-lead electrocardiogram for prediction of long-term prognosis in male US veterans. Circulation 2002;105:1066 –1070. 23. Rautaharju, PM, Kooperberg, C, Larson JC, LaCroix, A. Electrocardiographic abnormalities predicting coronary heart disease events and mortality in postmenopausal women. The Women’s Health Initiative. Circulation 2005; in press. 24. Larsen CT, Dahlin J, Blackburn H, Scharling H, Appleyard M, Sigurd B, Schnohr P. Prevalence and prognosis of electrocardiographic left ventricular hypertrophy, ST segment depression and negative T-wave. Eur Heart J 2002;23:315–324.