Major and Minor ECG Abnormalities in Asymptomatic Women and Risk of Cardiovascular Events and Mortality

Context Data are sparse regarding the prevalence, incidence, and independent prognostic value of minor and/or major electrocardiographic (ECG) abnormalities in asymptomatic postmenopausal women. There is no information on the effect, if any, of hormonal treatment on the prognostic value of the ECG.

Objective To examine association of minor and major baseline and incident ECG abnormalities with long-term cardiovascular morbidity and mortality.

Design, Setting, and Participants Post-hoc analysis of the estrogen plus progestin component of the Women's Health Initiative study, a randomized controlled primary prevention trial of 14 749 postmenopausal asymptomatic women with intact uterus who received 1 daily tablet containing 0.625 mg of oral conjugated equine estrogen and 2.5 mg of medroxyprogesterone acetate or a matching placebo. Participants were enrolled from 1993 to 1998, and the estrogen plus progestin trial was stopped on July 7, 2002.

Main Outcome Measures The Novacode criteria were used to define minor, major, and incident ECG abnormalities. Cardiovascular end points included incident coronary heart disease (CHD) and cardiovascular disease (CVD) events.

Results Among women with absent (n = 9744), minor (n = 4095), and major (n = 910) ECG abnormalities, there were 118, 91, and 37 incident CHD events, respectively. The incident annual CHD event rates per 10 000 women with absent, minor, or major ECG abnormalities were 21 (95% confidence interval [CI], 18-26), 40 (95% CI, 32-49), and 75 (95% CI, 54-104), respectively. After 3 years of follow-up, 5% of women who had normal ECG at baseline developed new ECG abnormalities with an annual CHD event rate of 85 (95% CI, 44-164) per 10 000 women. The adjusted hazard ratios for CHD events were 1.55 (95% CI, 1.14-2.11) for minor baseline, 3.01 (95% CI, 2.03-4.46) for major baseline, and 2.60 (95% CI, 1.08-6.27) for incident ECG abnormalities. There were no significant interactions between hormone treatment assignment and ECG abnormalities for risk prediction of cardiovascular end points. For prediction of CHD events, the addition of ECG findings to the Framingham risk score increased from 0.69 to 0.74 the area under the receiver operating characteristic curve. Similar findings were found for incident CVD events.

Conclusions Among asymptomatic postmenopausal women, clinically relevant baseline and incident ECG abnormalities are independently associated with increased risk of cardiovascular events and mortality, and the information is incremental to the established method of risk stratification.

Trial Registration clinicaltrials.gov Identifier: NCT00000611

Resting 12-lead electrocardiogram (ECG) abnormalities are independently associated with incident coronary heart disease (CHD) and cardiovascular disease (CVD) events.^1-14 Many of the prior studies included only men or compared men and women, but the women were not selected for age or the presence or absence of underlying heart disease.^1-13 Data are sparse regarding the prognostic significance of baseline ECG abnormalities in postmenopausal women without clinically manifest heart disease.¹⁴ Furthermore, there is no information on the prognostic significance of incident ECG abnormalities.

The Women's Health Initiative (WHI) clinical trials of hormone therapy examined whether in healthy postmenopausal women a combination of estrogen and progestin would reduce CHD and CVD events.¹⁵ The study showed that there was a significant increase in CHD rates among women taking hormone therapy compared with the placebo group.¹⁶ A subsequently published article on the risk of CHD found that clinical characteristics or biomarkers did not significantly modify the treatment-related risk of CHD end points.¹⁷

In our study, we examined the association of baseline and incident ECG findings with CHD and CVD outcomes in the placebo and hormonal treatment groups of the WHI estrogen plus progestin trial.^16,17 We hypothesized that baseline minor and major ECG abnormalities and development of new ECG abnormalities during follow-up would be associated with increased incident CHD and CVD outcomes independent of traditional risk factors. We also sought to examine whether hormone use would affect the association between the ECG findings and CVD outcomes.

Detailed eligibility criteria and recruitment methods, randomization, follow-up, data and safety monitoring, and quality assurance have been published previously.^15-17 Briefly, most participants were recruited by population-based direct mailing campaigns to women aged 50 to 79 years at initial screening. Postmenopausal women with an intact uterus were eligible for the combined estrogen and progestin trial. Participants were enrolled from 1993 to 1998, and the estrogen plus progestin trial was stopped on July 7, 2002. The protocol and consent form were approved by the institutional review boards of all participating institutions, and written informed consent was obtained from all participants.

Major exclusion criteria were related to competing risks, safety issues, and adherence.^15-17 For these analyses, we excluded women with history of prior myocardial infarction (MI), angina, congestive heart failure, coronary artery bypass graft surgery, percutaneous transluminal coronary angioplasty or stenting, permanent pacemakers, stroke or transient ischemic attacks, deep venous thrombosis, and pulmonary embolism. The sample analyzed included 14 749 women with intact uterus who received 1 daily tablet containing 0.625 mg of oral conjugated equine estrogen and 2.5 mg of medroxyprogesterone acetate (Prempro, Wyeth Ayerst Pharmaceutical, Philadelphia, Pa) or a matching placebo.

Follow-up contact for clinical events occurred every 6 months, in addition to an annual in-clinic visit. At each semi-annual contact, a standardized interview with a self-administered questionnaire was used to collect information on symptoms and safety concerns. Standard 12-lead ECGs were taken at baseline and at follow-up years 3 and 6. The data and safety monitoring board stopped the trial early at an average follow-up of 5.2 years, because health risks exceeded health benefits in the active treatment group compared with placebo.¹⁶ Race/ethnicity was determined by self-report questionnaire.¹⁵

Coronary heart disease was defined as acute MI necessitating overnight hospitalization, silent MI identified on serial ECGs, or death due to CHD. Detailed definitions for the diagnosis of acute MI and death due to CHD were previously published and are based on standardized criteria.^15-17 Cardiovascular disease end points included CHD (CHD death and nonfatal MI), coronary artery bypass graft surgery/percutaneous transluminal coronary angioplasty, and stroke (fatal and nonfatal).^16,17

Standard 12-lead ECGs were recorded at baseline and the annual visit at year 3 in the resting supine position using strictly standardized procedures in all clinical centers.¹⁴ All ECGs received at the central ECG laboratory (EPICARE Center, University of Alberta, Alberta, Edmonton, and later in Wake Forest University, Winston-Salem, NC) were inspected visually to detect technical errors, missing leads, and inadequate quality, and such records (n = 255) were rejected from ECG data files. ECG data were stored electronically and transmitted daily to the Electrocardiographic Reading Center for analysis by using the Novacode criteria measurement and classification system.¹⁸

ECG abnormalities were divided into minor and major abnormalities on the basis of the Novacode system, which is a modification of the criteria used in the Pooling Project.⁹ The Pooling Project was a US national cooperative study of 5 longitudinal investigations on the incidence of heart disease that was the first to categorize individual ECG findings into minor and major groupings, and used the Minnesota code for coding individual ECG abnormalities. We used a hierarchical categorization. Women with only minor ECG abnormalities were classified as having minor abnormalities; women with both minor and major abnormalities were classified as having major abnormalities. Women without minor or major ECG abnormalities were classified as having marginal/absent abnormalities and their ECG was considered normal.

Criteria for minor prevalent ECG abnormalities were any of the following: (1) first- and second-degree atrioventricular block (Novacode 2.1 and 2.2.1); (2) borderline prolonged ventricular excitation (Novacode 3.4.1 and 3.4.2); (3) prolonged ventricular repolarization (Novacode 4.1.1 and 4.1.2); (4) isolated minor Q and ST-T abnormalities (Novacode 5.7 and 5.8); (5) left ventricular hypertrophy without ST-T abnormalities (Novacode 6.1.0); (6) left atrial enlargement (Novacode 7.1); (7) frequent atrial or ventricular premature beats (Minnesota code 8.1); and (8) fascicular blocks (Novacode 10.1 and 10.2).

Criteria for major prevalent ECG abnormalities were any of the following: (1) atrial fibrillation or atrial flutter (Novacode 1.5); (2) high-degree atrioventricular dissociation (Novacode 2.3.1 and 2.3.2); (3) left bundle-branch block (Novacode 3.1.0 and 3.1.1); (4) right bundle-branch block (Novacode 3.2.0); (5) indeterminate conduction delay (Novacode 3.3.0 and 3.3.1); (6) Q-wave MI (Novacode 5.1, 5.2, 5.3, and 5.4); (7) isolated ischemic abnormalities (Novacode 5.5 and 5.6); (8) left ventricular hypertrophy with ST-T abnormalities (Novacode 6.1.1); and (9) other Novacode 1.4, 1.7, 1.8, 1.9, and 2.4, which refer to miscellaneous arrhythmias (eg, supraventricular tachycardia, ventricular preexcitation, ventricular tachycardia) with less than 5 participants being included in the analysis and not listed individually.

Criteria for incident ECG abnormalities were any of the following: (1) new atrial fibrillation or flutter (Novacode I 1.5); (2) new prolonged ventricular excitation (Novacode I 3.1, I 3.2, I 3.3, and I 3.4); (3) new prolonged ventricular repolarization (Novacode I 4.1.); (4) new left ventricular hypertrophy (Novacode I 6.1.1 and I 6.1.2); (5) new Q-wave MI (Novacode I 5.1, I 5.2, I 5.3, and I 5.4); and (6) new ischemic ST-T evolution (Novacode I 5.5, I 5.6.1, I 5.6.2, and I 5.7).

Associations between continuous baseline variables and ECG abnormality status were assessed by using 2-sample t tests and presented with means (SDs). Associations with categorical baseline variables were assessed with χ² tests and presented with sample sizes and percentages.

To evaluate outcome occurrence, results are presented with the number of events, percentage of total participants with the event, the rate per 10 000 women per year, and a 95% confidence interval (CI) for the rate. Kaplan-Meier survival curves were estimated and compared between ECG groups by using the log-rank test. A crude relative risk was estimated from the event rates. Adjusted hazard ratios (HRs) for the association between outcome occurrences and ECG abnormalities were modeled with Cox proportional hazards regression models, and stratified by randomization status in the Dietary Modification trial of the WHI.¹⁵ The proportional hazards assumption was checked by modeling each outcome (CHD and CVD) by a time by baseline ECG status interaction term in a proportional hazards model along with the baseline ECG main effects term. This was not significant for CHD (interaction P=.29) and borderline for CVD (interaction P=.01), indicating that the proportional hazards assumption was satisfied. Cox proportional hazards regression models were adjusted for estrogen plus progestin treatment assignment, age, ethnicity, history of treated diabetes mellitus, hypertension, current smoking, self-reported pills for cholesterol and statin use, and body mass index (calculated as weight in kilograms divided by height in meters squared). The resulting HRs and 95% CIs are presented.

In addition to the main analyses, subgroups were examined to investigate an interaction effect between baseline ECG status and the majority of the above risk factors, with an additional factor of parental history of premature MI. To evaluate this interaction, the P value of an interaction term from a Cox proportional hazards regression model with main effects of the subgroup of interest and baseline ECG status as well as their interaction were calculated. All Cox proportional hazards regression analyses were complete case and excluded a small percentage of participants (<10%) with missing data.

Additional models were constructed to look at the effect on outcome association by adding ECG data to existing CHD risk criteria; in this case, the Framingham risk score for 10-year CHD risk. To evaluate this, the likelihood ratio test was used for both the incident CHD and CVD outcomes. First, a Cox proportional hazards regression model was fit with the outcome of interest as a function of the Framingham CHD risk score.¹⁹ A second model was then fit with the Framingham score as well as the baseline ECG variable (normal, minor and major abnormalities). The difference between the –2 log likelihoods of the 2 models was then tested against a χ² statistic with 2 df, with a significance level of P<.05 being used to determine significant improvement from the Framingham-only model to the Framingham plus ECG model.

We also ran logistic models to compare the differences between the Framingham-only and Framingham plus ECG models. C statistics and receiver operating characteristic curves were completed for each of these models. Because the Framingham score calculation already involves covariates that were previously used as adjustments, the only adjustment used in these comparison analyses was randomization status in the WHI Dietary Modification trial. Calculation of the Framingham risk score also requires measured cholesterol information. Because of this, both the survival and logistic analyses involving Framingham risk score were limited to participants in the WHI blood subsample (n = 1264). All analyses were performed by using SAS version 9.1 (SAS Institute Inc, Cary, NC). All tests were 2-sided and P<.05 was considered statistically significant.

We identified 14 749 participants of the WHI estrogen plus progestin trial who at baseline had no history of prior MI, angina, coronary interventions, coronary artery bypass graft surgery, stroke, or transient ischemic attacks. A total of 7593 women were randomized to estrogen plus progestin group and 7156 were assigned to placebo. Baseline characteristics of the study group are shown in Table 1. At the baseline examination, the ECG was normal in 9744 women (66.0%), 4095 women (27.8%) had minor abnormalities, and 910 women (6.2%) had major abnormalities. Women in the group with ECG abnormalities were older, had a higher body mass index, and were more likely to have a history of treated diabetes, hypercholesterolemia, and hypertension.

The mean follow-up was 5.6 years, with a maximum of 8.6 years. A total of 246 women had CHD events and 595 had CVD events. Table 2 shows the number of women and the adjusted HRs for each component of the CHD and CVD outcomes, and total mortality. The strongest association of both minor and major ECG abnormalities was with CHD death.

The crude rates and the multivariable-adjusted HRs associated with the presence of minor ECG abnormalities were significantly higher compared with those women with absent ECG abnormalities (Table 3). The Kaplan-Meier estimates, shown in Figure 1, show a significant difference in CHD and CVD outcomes (log-rank P<.001) between absent and minor ECG abnormality groups. There was an excess of 36 (95% CI, 23-50) CVD events per 10 000 women per year compared with those women with absent ECG abnormalities.

When we examined the placebo and estrogen plus progestin groups separately (Table 4), the adjusted HRs of incident CHD and CVD were consistently higher among the hormone-treated group compared with the placebo group, although the HRs in the placebo group did not reach significance due to the small number of outcomes.

As shown in Table 3, the crude rates of incident and the multivariable-adjusted HRs associated with the presence of major ECG abnormalities were significantly higher compared with those women with absent ECG abnormalities. The Kaplan-Meier estimates, shown in Figure 1, show a significant difference in CHD and CVD outcomes (log-rank P<.001) between the absent and major ECG abnormality groups. There was an excess of 113 (95% CI, 79-150) CVD events per 10 000 women per year compared with those women with absent ECG abnormalities.

When we examined separately the placebo and estrogen plus progestin groups for the adjusted HRs of incident CHD and CVD events, we did not find a consistent difference between the hormone-treated and placebo groups (Table 4).

A total of 12 647 (86%) of 14 749 women came for their third-year visit (follow-up ECG). There were 7717 women with normal baseline ECG who had no CHD, congestive heart failure, angina, or CVD events before the 3-year follow-up and whose ECG remained normal (referent group). Incident ECG abnormalities developed in 405 women (5%). The annual event rate per 10 000 women for incident CHD was 85 (95% CI, 44-164). The multivariable-adjusted HRs associated with the presence of incident ECG abnormalities were 2.60 (95% CI, 1.08-6.27) for CHD and 2.86 (95% CI, 1.69-4.83) for CVD events.

To determine whether the presence of ECG abnormalities (minor and major) were more predictive of cardiovascular outcome in selected subgroups of women, we examined several demographic and clinical characteristics of the estrogen plus progestin trial participants (Figure 2). Despite differences noted in HRs associated with minor ECG abnormalities, there were no significant interactions between ECG abnormalities and hormone treatment in the risk for CVD events (Table 4 and Figure 2). Likewise, we observed no other significant interactions between ECG abnormalities across different strata of women. The interaction between ECG abnormalities and ethnicity was of borderline significance (Figure 2).

When added to the Cox proportional hazards regression models, which looked at CHD and CVD as a function of Framingham CHD risk score, baseline ECG abnormality status significantly improved each model and was evaluated with the likelihood ratio test (P = .004 for CHD and P = .02 for CVD). This improvement was also observed by looking at the receiver operating characteristic curves and C statistics computed from logistic regression (Figure 3). The area under the receiver operating characteristic curve increased from 0.69 (95% CI, 0.61-0.86) to 0.74 (95% CI, 0.66-0.90) for CHD and 0.68 (95% CI, 0.62-0.77) to 0.70 (95% CI, 0.65-0.79) for CVD, with the addition of baseline ECG abnormalities to a model with the Framingham risk score.

In a large cohort of postmenopausal, asymptomatic women who were without a history of prior CVD and participating in the estrogen plus progestin group of the WHI trial, we found that minor and major baseline ECG abnormalities were associated with significantly increased risks for CHD and CVD events, independent of established risk factors and hormone treatment. Several previous studies that evaluated nonspecific repolarization abnormalities of the ECG in women found significant associations between CHD outcomes and ECG abnormalities.^2,3,7,12,13 The 2 prior studies that examined the prognostic significance of minor and major ECG abnormalities in women showed similar significant association between CHD outcomes and ECG abnormalities independent of clinical variables.^8,11

A recent publication on the ECG abnormalities of women participating in the WHI study showed that repolarization abnormalities, such as wide QRS/T angle, QT prolongation, high QRS nondipolar voltage, and reduced heart rate variability, predict CHD events and mortality.¹⁴ However, these ECG abnormalities are not easily characterized or diagnosed by clinicians without computer assistance. We used the ECG classification of minor and major abnormalities because they are simple, easily defined and interpreted by clinicians, and widely applicable in the clinical setting. Our study is in agreement with the previous study findings.^{2,3,7,8,11-13} In addition, we provide new information regarding incident ECG abnormalities and the associated increased risk of coronary morbidity and mortality. We also found that increasing severity of the ECG findings, when defined as the presence of minor and major abnormalities, correlate with increasing risk for CVD events and mortality. The Kaplan-Meier estimates of cumulative hazard for CVD events showed a significant difference, with an early onset of the separation of the curves, for both ECG abnormalities (Figure 2).

We present new information showing that in a subgroup of women with complete data on the Framingham risk score, the ECG provides incremental information for risk stratification of CHD events beyond the standard risk factors (Figure 3). The risks for CVD events associated with ECG abnormalities were similar across multiple subgroups, although we did observe evidence for an interaction between ECG abnormalities and ethnicity that was of borderline statistical significance, suggesting that white women with an abnormal baseline ECG have a higher risk for CVD risk than nonwhite women did.

There is clinical and experimental evidence that there are sex-related differences in cardiac repolarization and possibly arrhythmias.²⁰ In our study, the use of estrogen and progestin did not affect the overall CVD risk assessment by the ECG significantly.

Limitations to our study should be considered. The effect of ethnicity on our findings could not be adequately evaluated because 12 429 study participants (84%) were white. Confirmation of our findings in other studies of women with larger minority participation would be useful. Second, not all patients with baseline ECG presented for a repeat ECG at year 3 and many of those who did only had a short period of follow-up because the trial was terminated prematurely. Third, there was an underascertainment of increased low-density lipoprotein and other biochemical markers that were not included in the analysis or were not measured but may be associated with ECG abnormalities. Finally, data for the Framingham risk score determination was available only in a limited number of participants and our study findings need to be confirmed in a larger population with longer follow-up.

Given the low cost, wide availability, and ease of interpretation, the ECG may be a useful tool for assisting in the prediction of future cardiovascular events in asymptomatic postmenopausal women. The presence of ECG abnormalities should prompt physicians to consider further risk stratification, more intensive therapeutic interventions, or both on modifiable risk factors for primary prevention of cardiovascular events.

Corresponding Author: Pablo Denes, MD, Departments of Medicine and Preventive Medicine, Northwestern University, Feinberg School of Medicine, 251 E Huron, Feinberg 8-340, Chicago, IL 60611 (p-denes@northwestern.edu).

Author Contributions: Mr Larson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Denes, Lloyd-Jones, Prineas, Greenland.

Acquisition of data: Denes, Larson, Prineas, Greenland.

Analysis and interpretation of data: Denes, Larson, Lloyd-Jones, Greenland.

Drafting of the manuscript: Denes, Larson, Lloyd-Jones.

Critical revision of the manuscript for important intellectual content: Denes, Larson, Lloyd-Jones, Prineas, Greenland.

Statistical analysis: Denes, Larson, Lloyd-Jones.

Obtained funding: Greenland.

Administrative, technical, or material support: Prineas, Greenland.

Study supervision: Lloyd-Jones, Greenland.

A Short List of Women's Health Initiative Investigators:

Program Office: National Heart, Lung, and Blood Institute, Bethesda, Md: Barbara Alving, Jacques Rossouw, Shari Ludlam, Linda Pottern, Joan McGowan, Leslie Ford, and Nancy Geller.

Clinical Coordinating Center: Fred Hutchinson Cancer Research Center, Seattle, Wash: Ross Prentice, Garnet Anderson, Andrea LaCroix, Charles L. Kooperberg, Ruth E. Patterson, and Anne McTiernan; Wake Forest University School of Medicine, Winston-Salem, NC: Sally Shumaker; Medical Research Labs, Highland Heights, Ky: Evan Stein; and University of California, San Francisco: Steven Cummings.

Clinical Centers: Albert Einstein College of Medicine, Bronx, NY: Sylvia Wassertheil-Smoller; Baylor College of Medicine, Houston, Tex: Jennifer Hays; Brigham and Women's Hospital, Harvard Medical School, Boston, Mass: JoAnn Manson; Brown University, Providence, RI: Annlouise R. Assaf; Emory University, Atlanta, Ga: Lawrence Phillips; Fred Hutchinson Cancer Research Center, Seattle, Wash: Shirley Beresford; George Washington University Medical Center, Washington, DC: Judith Hsia; Harbor-University of California Los Angeles Research and Education Institute, Torrance, Calif: Rowan Chlebowski; Kaiser Permanente Center for Health Research, Portland, Ore: Evelyn Whitlock; Kaiser Permanente Division of Research, Oakland, Calif: Bette Caan; Medical College of Wisconsin, Milwaukee: Jane Morley Kotchen; MedStar Research Institute/Howard University, Washington, DC: Barbara V. Howard; Northwestern University, Evanston, Ill: Linda Van Horn; Rush Medical Center, Chicago, Ill: Henry Black; Stanford Prevention Research Center, Stanford, Calif: Marcia L. Stefanick; State University of New York at Stony Brook, Stony Brook: Dorothy Lane; The Ohio State University, Columbus: Rebecca Jackson; University of Alabama at Birmingham, Birmingham: Cora E. Lewis; University of Arizona, Tucson/Phoenix: Tamsen Bassford; University at Buffalo, Buffalo, NY: Jean Wactawski-Wende; University of California at Davis, Sacramento: John Robbins; University of California at Irvine, Irvine: F. Allan Hubbell; University of California at Los Angeles, Los Angeles: Howard Judd; University of California at San Diego, LaJolla/Chula Vista: Robert D. Langer; University of Cincinnati, Cincinnati, Ohio: Margery Gass; University of Florida, Gainesville/Jacksonville: Marian Limacher; University of Hawaii, Honolulu: David Curb; University of Iowa, Iowa City/Davenport: Robert Wallace; University of Massachusetts/Fallon Clinic, Worcester: Judith Ockene; University of Medicine and Dentistry of New Jersey, Newark: Norman Lasser; University of Miami, Miami, Fla: Mary Jo O’Sullivan; University of Minnesota, Minneapolis: Karen Margolis; University of Nevada, Reno: Robert Brunner; University of North Carolina, Chapel Hill: Gerardo Heiss; University of Pittsburgh, Pittsburgh, Pa: Lewis Kuller; University of Tennessee, Memphis: Karen C. Johnson; University of Texas Health Science Center, San Antonio: Robert Brzyski; University of Wisconsin, Madison: Gloria E. Sarto; Wake Forest University School of Medicine, Winston-Salem, NC: Denise Bonds; Wayne State University School of Medicine/Hutzel Hospital, Detroit, Mich: Susan Hendrix.

Funding/Support: The Women's Health Initiative program was funded by the National Heart, Lung, and Blood Institute of the National Institutes of Health, US Department of Health and Human Services.

Role of the Sponsors: The funding organization had representation on the Steering Committee, which governed the design and conduct of the study, the interpretation of the data, and the preparation and approval of the manuscript. A National Heart, Lung, and Blood Institute Program officer reviewed the manuscript prior to publication.