EPIDEMIOLOGIC STUDIES. Methods

1494 lacc Vol. 12. No.6 EPIDEMIOLOGIC STUDIES Major and Minor Electrocardiographic Abnormalities and Risk of Death From Coronary Heart Disease, Cardiovascular Diseases and All Causes in Men and Women YOULIAN LIAO, MD, * KIANG LIU, PHD, * ALAN DYER, PHD, * JAMES A. SCHOENBERGER, MD, FACC,t RICHARD B. SHEKELLE, PHD,:j: PATRICIA COLETTE, MA,* JEREMIAH STAMLER, MD, FACC* Chicago, Illinois and Houston, Texas The independent contributions of baseline major and minor electrocardiographic (ECG) abnormalities to subsequent 11.5 year risk of death from coronary heart disease, all cardiovascular diseases and all causes were explored among 9,643 white men and 7,990 white women aged 40 to 64 years without definite prior coronary heart disease in the Chicago Heart Association Detection Project in Industry. At baseline, prevalence rates of major ECG abnormalities were higher in women than in men, with age-adjusted rates of 12.9 and 9.6% (p < 0.01), respectively. Minor ECG abnormalities were more common in men than in women (7.3 versus 4.5%, p < 0.01). Both major and minor ECG abnormalities were associated with an increased risk of death from coronary heart disease, all cardiovascular diseases and all causes. The strength of these associations was greater in men than in women. When baseline age, diastolic pressure, serum cholesterol, cigarettes smoked per day, diabetes and use of antihypertensive medication were taken into account, major abnormalities continued to be significantly related to each cause of death in both genders with much larger adjusted absolute excess risk and relative risk for men than for women. In multivariate analyses, minor ECG abnormalities contributed independently to risk of death in me)1, but not clearly so in women. The results indicate the independent association between ECG abnormalities and mortality from coronary heart disease, all cardiovascular diseases and all causes, with greater relative significance in middle-aged United States men than women. (J Am Coli CardioI1988,'12:1494-S00) The electrocardiogram (ECG) at rest has long been used in clinical examination of patients, in screening asymptomatic people and in epidemiologic surveys. The prognostic significance of the ECG for incidence of coronary heart disease, From the *Northwestern University Medical School, Chicago, Illinois. the trush Presbyterian-S!. Luke's Medical Center. Chicago and the +University of Texas School of Public Health, Houston, Texas. This study was presented at the 10th World Congress of Cardiology, September 1986, Washington, D.C. This research was supported by the American Heart Association, Dallas; Illinois Regional Medical Program, Springfield, Illinois; National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland (Grant 5 RO 1 HL-21021); Burton Marstellar Limited (Flora Information Service), London, England; Ciba-Geigy Company, Summit, New Jersey; CPC International (Best Foods), Union, New Jersey and the Chicago Health Research Foundation, Chicago. Manuscript received December I, 1986; revised manuscript received June 22, 1988, accepted July 1, 1988. Address for reprints: Jeremiah Stamler, MD. Department of Community Health and Preventive Medicine, Northwestern University Medical School, 303 East Chicago Avenue, Chicago, Illinois 60611. 1988 by the American College of Cardiology coronary heart disease mortality or total mortality has been validated in many investigations (1-12). However, most studies have dealt with men only or with men and women combined in a single observation group (1-4,6,8-12). Few studies (7,8,10,11) have examined whether the relation of ECG abnormalities to coronary heart disease and mortality is independent or confounded by other risk factors. The purpose of this study was to assess whether there is an independent relation of ECG abnormalities (major and minor abnormalities, respectively) to risk of death from coronary heart disease, all cardiovascular diseases and all causes in both men and women, and whether there is a gender differential in this regard. Methods Study population. The Chicago Heart Association Detection Project in Industry is a cross-sectional and prospective investigation of 39,573 young adult and middle-aged men and 0735-1097/88/$3.50

JACC Vol. 12, No.6 LIAO ET AL. 1495 women, most of whom are white or black. The baseline survey was done from late 1967 through early 1973 in 84 cooperating companies and organizations in the Chicago area (11,13). The cohort selected for analysis in this study was made up of the 18,218 white male and female screened participants aged 40 to 64 years at baseline. This study excluded 88 persons with evidence of definite myocardial infarction on baseline ECG, 346 persons with unknown vital status and 151 persons with missing data for one or more relevant variables. The study sample therefore consisted of 17,633 persons (9,643 men and 7,990 women). All analyses were also done with the additional exclusion of 277 persons reporting a previous medical diagnosis of heart disease or treatment for heart disease at baseline. Because findings were similar with and without this exclusion, the results presented here are for the cohort of 17,633 persons including these 277. Electrocardiograms and other baseline data. Over the years of screening, procedures for recording and analyzing the ECG at rest varied as follows: 1) use of six standard limb leads (2,586 persons); 2) standard 12 lead ECG (988 persons); 3) use for primary ECG screening of the Electrocardioanalyzer model 400 (ECA), which received signals from leads I, II, avf, VI and V s, plus a standard 12 lead ECG when ECA signals were outside normal limits (2,816 persons); 4) use of the ECA to screen, and when it indicated ECG signals within normal limits, a 5 lead ECG (I, II, avf, VI and Vs), and when the ECA reading was outside normal limits, a standard 12 lead recording (7,449 persons); and 5) ECA and 5 lead recordings (3,794 persons). The ECA was found to have high sensitivity (92%) for detecting abnormal ECGs (14), and in an earlier study abnormalities as recorded with each of the foregoing five procedures related similarly to 5 year risk of mortality for middle-aged men in this cohort (11). When ECA signals were abnormal, a regular ECG was recorded and the final assessment was based on the latter tracings. All electrocardiograms were read by one cardiologist (J.A.S.), using criteria and forms of the 1960 to 1962 National Health Examination Survey, United States Public Health Service (15). Tracings with any abnormalities were later identified and reclassified according to the Minnesota Code (16). Electrocardiographic abnormalities were divided into major and minor abnormality groups on the basis of the criteria of the Pooling Project (9). Persons with both major and minor abnormalities were classified as having major abnormalities. Criteria for major abnormalities were any of the following: 1) Complete or second degree atrioventricular (A V) block (Minnesota Code 6-1, 2). 2) Complete left or right bundle branch block, or intraventricular block (Minnesota Code 7-1, 2, 4). 3) Atrial fibrillation or flutter (Minnesota Code 8-3). 4) Frequent premature beats (Minnesota Code 8-1). 5) Significant ST segment depression (Minnesota Code 4-1,2).6) T wave deep inversion (Minnesota Code 5-1, 2). Criteria for minor abnormalities were any of the following: 1) Q wave, borderline (Minnesota Code 1-3). 2) ST segment depression, borderline (Minnesota Code 4-3). 3) Moderate T wave inversion (Minnesota Code 5-3). 4) first degree A V block (Minnesota Code 6-3).5) QRS, high voltage (Minnesota Code 3-1, 2). 6) QRS, low voltage (Minnesota Code 9-1). 7) Axis deviation (Minnesota Code 2-1, 2). Other baseline data, including blood pressure and serum cholesterol, were measured in a standardized way based on a single protocol. Data on cigarette smoking, history of physician-diagnosed diabetes and current use of antihypertensive medication were obtained from a precoded selfadministered questionnaire. Long-term follow-up of vital status. The vital status of each participant was ascertained every 5 years after baseline. The mean duration of follow-up of the cohort was 11.5 years for both men and women, varying from 0.1 year for early decedents to 18 years. The cause of death was classified from death certificates, without knowledge of baseline findings; the eighth revision of the International Classification of Diseases (adapted for use in the United States) was used, with codes 410 to 414 for coronary heart disease and codes 391 to 458 for all cardiovascular diseases (17). Statistical methods. Prevalence rates of major and minor ECG abnormalities by gender were compared for each 5 year age group with use of a chi-square test. The direct method was used to compute age-adjusted prevalence rates and to adjust for the five different ECG recording procedures. Age-adjusted cumulative mortality rates for each cause of death were calculated by the life table method (18). Multivariate-adjusted 11.5 year mortality rates were computed by multiple linear regression for each gender under the assumption of a common, linear relation between mortality and risk factors for all ECG abnormality subgroups in each gender. The Cox regression model (19) was used to examine the independent contribution of ECG abnormalities to each cause of death, with control for age, diastolic pressure, serum cholesterol, cigarettes smoked per day and history of physician-diagnosed diabetes. For the 1,329 persons taking antihypertensive medication, because baseline blood pressure measurements may not represent the true blood pressure levels, adjustment also was made for baseline usage of antihypertensive medication. Cox regression models with and without control for the five different recording procedures were fit separately for each gender. Because none of the differences for the log likelihoods between the two models with and without control for the different recording procedures were statistically significant, the results presented are the overall combined data, irrespective of the five ECG recording procedures.

1498 LIAO ET AL. JACC Vol. 12, No.6 Table 3. Adjusted 11.5 Year Mortality Rates for Coronary Heart Disease, All Cardiovascular Diseases and All Causes by Electrocardiographic Abnormalities and by Gender CHD All CV DIS All Causes Rate Per 1,000: Rate Per 1,000: Rate Per 1,000: Total Adjustment Adjustment Adjustment Number n Age Multiple* n Age Multiple* n Age Multiple* Men Major 907 133 142 134 158 168 155 228 235 215 Minor 695 55 88 83 72 113 104 118 178 162 Normal 8,041 229 31 32 296 40 39 636 85 81 Absolute excess risk Major vs. Normal III 102 128 116 150 134 Minor vs. Normal 57 51 73 65 93 81 Relative riskt Major vs. normal 4.6 3.7 4.2 3.4 2.8 2.4 Minor vs. normal 2.4 2.1 2.4 2.1 1.8 1.7 Women Major 1,047 25 23 21 43 38 37 82 73 72 Minor 361 8 22 20 11 32 29 25 70 68 Normal 6,582 54 8 10 83 12 16 281 42 48 Absolute excess risk Major vs. normal 15 11 26 21 31 24 Minor vs. normal 14 10 20 13 28 20 Relative risk Major vs. normal 2.4 1.9 2.7 2.1 1.6 1.4 Minor vs. normal 2.2 1.5 2.0 1.4 1.4 1.2 * Adjusted for age, diastolic pressure, serum cholesterol, cigarettes smoked per day, diabetes and use of antihypertensive medication; tbased on Cox regression models. Abbreviations as in Tables I and 2. from coronary heart disease, all cardiovascular disease and all causes. The strength of these associations was greater in men than in women. In multivariate analyses, minor ECG abnormalities contributed independently to risk of death only in men, but not in women. Data from a few other prospective population studies (5,20,21) have shown a lesser increase in incidence or mortality rate of coronary heart disease in women with ECG abnormalities compared with men with such abnormalities. However, the numbers of those studies were too small for in-depth comparisons, including multivariate analyses. The reasons for the observed gender differential in predictive significance of ECG abnormalities of subsequent mortality are unknown. One possibility is that a higher proportion of ECG changes reflect underlying severe atherosclerosis in men than in women (21). Several population studies in the United States (5,20,22), including the present one, have found baseline prevalence of ECG abnormalities to be nearly equal in the two genders, or even more common in women than in men. These findings seem anomalous in view of the much lower incidence of major coronary heart disease events in women than in men in "western" popula- Table 4. Relative Risks With Electrocardiographic Abnormalities and Other Risk Factors: 11.5 Year Mortality by Cause and Gender (Cox Multiple Regression Analyses) Men (n = 9,643) Women (n = 7,990) CHD CVD All CHD CVD All Major (yes vs. no) 3.7* 3.4* 2.4* 1.9* 2.1* 1.4* Minor (yes vs. no) 2.1* 2.1* 1.7* 1.5 1.4 1.2 Age (per 10 yr) 2.0* 2.1* 2.2* 3.4* 3.0* 2.1* DBP (per 20 mm Hg) 1.5* 1.5* 1.4* 1.5* 2.1* 1.0 Chol (per 40 mg/dl) I.3t 1.2* 1.1* 1.3* 1.2* 1.5* Cig/day (per pack) 1.5* 1.5* 1.6* 2.8* 2.3* 1.7* Diabetes (yes vs. no) 3.0* 2.9* 2.1* 4.0* 3.9* 2.4* Hyp med (yes vs. no) 1.3 1.4* I.3t 1.9t 1.4 1.1 *p < 0.01; tp < 0.05. Abbreviations as in Tables I to 3.

1496 LIAO ET AL. lacc Vol. 12, No.6 December 1988: 1494--500 Table 1. Univariate Analyses of Baseline Characteristics by Electrocardiographic Abnormalities and by Gender in White Men and White Women Aged 40 to 64 Years at Entry Men (n = 9,643) Women (n = 7,990) Major Minor Normal Major Minor Normal No. 907 695 8,041 1.047 361 6,582 Age (yr) 53.7 ± 6.6* 51.9 ± 6.9* 49.9 ± 6.6 52.7 ± 6.3* 51.5 ± 6.4 50.8 ± 6.3 SBP (mm Hg) 152,8 ± 23,6* 147.1 ± 20.6* 140.9 ± 19.2 147.4 ± 23.0* 140.7 ± 22.9* 136.1 ± 19.6 OBP (mm Hg) 88.8 ± ]3,5* 87.0 ± 12.2* 83.5 ± 11.2 84.8 ± 12,2* 83.2 ± 12.1* 79.7 ± 11.2 Chol (mg/dl) 215.9 ± 41.9* 216.2 ± 40.6* 212.0 ± 35,6 223.8 ± 38.8* 228.4 ± 48.2* 220.0 ± 40.5 % Smoker 43.2* 41.6 38.8 33.8 35.7 34.9 Cig/day 9.9 ± 13.3 9.9 ± 13.9 9.4 ± 13.7 6.0 ± 10.2 6.8 ± 10.6 6.3 ± 10.1 % Diabetes 5.1 4.9 3.8 3.1 5.5t 2.4 % Hyp med 16.0* 8.5* 5.4 19.2* 11.9* 6.8 *toifferences between persons with electrocardiographic (ECG) abnormalities (major or minor) and persons with a normal electrocardiogram are statistically significant (*p < 0,01, tp < 0.05). Chol = serum cholesterol; Cig/day = cigarettes smoked per day; OBP = diastolic blood pressure; Hyp med = usage of antihypertensive medication; Major = major ECG abnormalities; Minor = minor ECG abnormalities; Normal = normal ECG; SBP = systolic blood pressure. Values are mean ± SO. Results Gender differential in prevalence of major and minor electrocardiographic abnormalities. Prevalence of major EeG abnormalities was greater with age in both men and women, A positive association between age and prevalence was also present in men for minor abnormalities, but not in women, With the exception of those aged 60 to 64 years, the prevalence of major abnormalities was significantly higher in women than in men for every age group, Age-adjusted prevalence rates were 12,9 and 9,6% for women and men, respectively (p < 0,01), In contrast, men had a significantly higher prevalence rate of minor EeG abnormalities at every age and a significantly higher age-adjusted prevalence rate (7,3 versus 4,5%, p < 0,01), For any EeG abnormality, women had a slightly higher prevalence rate than did men; the respective age-adjusted rates were 17.4 and 16,9%, With each of the five EeG recording procedures, prevalence rates of major EeG abnormalities were also significantly higher in women than in men although the prevalence of minor abnormalities was higher in men than in women, Baseline characteristics for men and women with and without electrocardiographic abnormalities (Table 1). For both men and women, those with major or minor EeG abnormalities compared with those free of abnormalities were on the average older, had higher values for systolic and diastolic blood pressure and serum cholesterol, and had higher prevalence rates for smoking, diabetes and use of antihypertensive medication, These differences are generally statistically significant (p < 0,05), Relation by gender of baseline major and minor electrocardiographic abnormalities to 11.5 year risk of mortality from coronary heart disease, all cardiovascular diseases and all causes (Table 2). Both men and women with either major or minor abnormalities had higher crude mortality rates for sudden and nonsudden death from coronary heart disease, death from stroke and other cardiovascular disease, and death from cancer and other noncardiovascular disease, In Table 2. Distribution in Numbers of Death and Crude Mortality Rates by Causes, Electrocardiographic Abnormalities and Gender Men (n = 9,643) Women (n = 7,990) Causes of Death Major Minor Normal Major Minor Normal No. 907 695 8,041 1.047 361 6,582 CHD Sudden* 45 (49.6) 21 (30.2) 74 (9.2) 10 (9.6) 19 (2.9) 1 (...)t Not sudden 88 (97.0) 34 (48,9) 155 (19.3) 15 (14.3) 7 (19.4) 35 (5.3) Other CV DIS Stroke 11 (12.1) 5 (7.2) 29 (3.6) 6 (5.7) 1 (..,.)t 18 (2.7) Other 14 (15.4) 12 (17.3) 38 (4,7) 12 (11.5) 2 (...)t 11 (1.7) Non CV DIS Cancer 34 (37.5) 27 (38.8) 203 (25.2) 24 (22.9) 11 (30.5) 134 (20.4) Other 36 (39,7) 19 (27.3) 137 (17.0) 15 (14.3) 3(...,)t 64 (0,7) *Sudden death = death within 3 h of onset of illness; tnumbers too small for rate calculations, CHD = coronary heart disease; CV = cardiovascular; DIS = disease; values in parentheses are mortality rates per 1,000,

JACC Vol. 12, No.6 Dec.ember 1988: 1494-500 LlAO ET AL. 1497 180 Mortality /1000 160 140 120 100 80 60 40 20 2 4 6 8 10 12 14 16 Years Major Abn Normal Figure 1. Age-adjusted cumulative mortality from coronary heart disease by major electrocardiographic abnormalities (abn) and gender. life table analyses (Fig, I and 2), as expected, men had greater age-adjusted cumulative mortality rates than did women, with either a normal, major abnormal or minor abnormal ECG. Persons with either a major or a minor abnormality had an unfavorable survival experience compared with those without an ECG abnormality. The findings were more prominent for men than for women. For women, for each year of follow-up, the cumulative age-adjusted mortality rate from coronary heart disease was on average approximately three times greater for those with a major Figure 2. Age-adjusted cumulative mortality from coronary heart disease by minor electrocardiographic abnormalities (abn) and gender. 180 160 140 120 100 80 Mortality/WOO _Men... Women Minor Abn abnormality than for those with no abnormality, and twice as great for women with a minor abnormality as for those with none, For men, the respective rates were almost 5-fold greater for those with major abnormalities and 2,5 times greater for those with minor abnormalities. The life table curves for mortality from all cardiovascular diseases and all causes show similar results (curves not displayed). In multivariate analyses (Table 3), men and women with either major or minor ECG abnormalities had greater 11.5 year mortality rates for coronary heart disease, all cardiovascular diseases and all causes than did those with a normal ECG. The absolute excess risks were much greater in men than in women. Age-adjusted as well as multivariateadjusted relative risk of each cause of death-estimated from Cox regression models-was also much greater in men than in women. The risks of death in men with major ECG abnormalities were much greater than in men with minor abnormalities. Lesser differences were found between women with major and women with minor abnormalities. Cox regression analyses were done to examine the independent contribution of ECG abnormalities to risk of three subsequent mortality end points in men and women (Table 4). As noted previously, results were similar with and without control for the five ECG methods; the findings presented here are therefore irrespective of these five methods. Major ECG abnormalities along with the established risk factors, that is, age, diastolic blood pressure, serum cholesterol, cigarette smoking and diabetes, were independently related to all three mortality end points in both genders (except diastolic blood pressure to death from all causes in women). Minor ECG abnormalities were also associated with a significant increase in risk of death from the three end points in men, but not in women. The differences in Cox coefficients between men and women with major abnormalities for the three mortality end points were statistically significant (t == 2.17 to 3.32; p < 0.05 to <0.01). For minor abnormalities, the differences were not statistically significant (t == 0.82 to 1.49; P == 0.41 to 0.14). The interactive effects of major and minor ECG abnormalities and gender on death from coronary heart disease, all cardiovascular diseases and all causes are shown in Table 5. As a result of the gender difference in the impact of major and minor ECG abnormalities on risk, the multivariateadjusted relative risks of male to female gender for all three mortality end points were much larger in those with than in those without an ECG abnormality. 60 40 20 o Normal... Minor Abn... ::::.:::::.. :.. :... ::::::... Normal 2 4 6 8 10 12 14 16 Years Discussion Gender differential in relation of electrocardiographic abnormalities to risk of death from coronary heart disease, cardiovascular diseases and all causes. This large scale prospective study showed that major ECG abnormalities were independently associated with an increased risk of death

lacc Vol. 12, No.6 LIAO ET AL. 1499 Table 5. Adjusted Relative Risks of Men to Women for Coronary Heart Disease, All Cardiovascular Diseases and All Cause Mortality by Cox Regression Relative Risk CHD CYD All Causes Age- Multiple- Age- Multiple- Age- Multiple- Adjusted Adjusted* Adjusted Adjusted* Adjusted Adjusted* Normal 3.7 3.4 3.1 2.8 2.0 1.7 Minor 3.8 4.1 3.6 3.7 2.6 2.4 Major 6.5 5.6 4.5 3.8 3.4 2.9 * Adjusted for age, diastolic pressure, serum cholesterol. cigarettes smoked per day, use of antihypertensive medication. All the above relative risks are significantly different from 0 (p < 0.01). Abbreviations as in Tables I to 3. tions, In other populations with almost no clinical coronary heart disease in middle age, sizable prevalence rates of ECG abnormalities in women have also been reported (23,24). One explanation put forward has been that more ECG abnormalities in women than in men are functional and not due to coronary heart disease (21). This observation might also explain the absence of age dependence in the prevalence of minor ECG abnormalities in women as shown in the present study, Unfortunately, there is no possibility of direct validation of the diagnostic reliability of indexes of coronary heart disease in large scale population studies, because there is no reference test to identify with certainty ECG false positive and false negative cases (25), The present data show that the prevalence of ECG abnormalities was associated with all major coronary risk factors, such as age, blood pressure, serum cholesterol, cigarette smoking and diabetes. Therefore, the finding that many men and women with ECG abnormalities were alive years later does not necessarily mean that they are false positive cases; they may be true positive cases but with a relatively benign course. It has also been reported (26,27) that some women with an abnormal ECG do have myocardial ischemia without significant coronary atherosclerosis. It may be that prognosis with such myocardial ischemia is different from that of ischemia due to coronary atherosclerosis. In addition, the mitral valve prolapse syndrome may be playing a role. This relatively benign disease can also cause ECG changes, and it is more prevalent in women than in men (28). Overall, for these or other unknown reasons, ECG abnormalities may more often reflect prognostically nonsignificant or less significant processes other than latent coronary heart disease in women compared with men. Independent predictive value of the electrocardiogram. Traditionally, the ECG has rarely been viewed as an independent prognostic variable, the usual statement being that "the prognosis of a given abnormal ECG pattern is that of the underlying heart disease." An important concern in the prediction of cardiovascular risk is whether ECG variables add independent information not given by major non-ecg variables. Now there is convincing evidence that ECG abnormalities are not only independent prognostic markers in patients with known heart disease (29), but also independent predictors of morbidity and mortality in persons with definite heart disease. Previous studies. The Framingham study (7) reported a significant association among left ventricular hypertrophy, intraventricular conduction defect and definite nonspecific T wave or ST segment abnormality and 18 year morbidity from coronary heart disease and all cardiovascular diseases, and mortality from coronary heart disease, all cardiovascular diseases and all causes, with control for age, systolic blood pressure, serum cholesterol, cigarettes smoked and glucose intolerance in men and women aged 45 to 74 years. The 5 year follow-up of the Whitehall study (8) showed a significant association between any ECG abnormality and coronary heart disease mortality independent of age, systolic blood pressure and serum cholesterol in middle-aged male civil servants. Recently, Cedres et al. (II) reported the independent contribution of ECG abnormalities, in particular major abnormalities, in middle-aged men to risk of death from coronary heart disease, all cardiovascular diseases and all causes in three Chicago epidemiologic studies. All these and the present study indicate that these ECG abnormalities generally reflect already existent subclinical disease, and in this regard they represent a type of risk factor qualitatively different from other risk factors. Thus, it is potentially meaningful and useful to detect and identify the more vulnerable segments of the population who are especially high risk candidates for preventive management. Possible limitations. The present cohort consisted of more than 17,000 persons followed up for an average of 11.5 years. However, the numbers of deaths were still small for women with minor ECG abnormalities. This limited the statistical power for comparisons between men and women with minor abnormalities. Also, the great number ofindividual ECG items in the Minnesota Code led to a denominator for most codes that was too small for mortality investigation,

1500 LIAO ET AL. JACC Vol. 12. No.6 December 1988:1494-500 hence the resort to the approach of combining ECG items into the two groups of those with major and minor abnormalities. Studies need to be done on the gender patterns of prognosis for specific ECG abnormalities with use of larger data bases, as well as further investigations of this cohort after a longer period of follow-up. We are pleased to acknowledge the cooperation of the Chicago organizations and their employees who participated in the Chicago Heart Association Detection Project in Industry. Acknowledgment is also gratefully extended to all those involved in the Project, especially the late Louis de Boer, Raymond Restivo, Kay Westfall and Sue Shekelle of the Chicago Heart Association. the project staff and the many Chicago Heart Association volunteers. References I. Mathewson FAL, Varnam GS. Abnormal electrocardiograms in apparently healthy people. 1. Long term follow-up study. Circulation 1%0;21: 196-203. 2. Ungerleider HE. The prognostic implications of the electrocardiogram. Am J Cardiol 1960;6:35-44. 3. Blackburn H, Parlin RW. Antecedents of disease: insurance mortality experience. Ann NY Acad Sci 1966;134:965-1017. 4. Keys A. Coronary heart disease in seven countries. Circulation 1970;41(suppl 1):1-1-211. 5. Bartel A, Heyden S, Tyroler HA, Tabesh E, Cassel JC, Hames CG. Electrocardiographic predictors of coronary heart disease. Arch Intern Med 1971 ;128:929-37. 6. Blackburn H. Importance of the electrocardiogram in populations outside the hospital. Can Med Assoc J 1973;108:1262-5. 7. Shurtleff D. Some characteristics related to the incidence of cardiovascular disease and death: Framingham Study, 18-year follow-up. Section 30. In: Kannel WB, Gordon T, eds. The Framingham Study-An Epidemiological Investigation of Cardiovascular disease. Washington, D.C.: United States Department of Health, Education, and Welfare; United States Public Health Service; National Institutes of Health. Department of Health, Education, and Welfare Publ (NIH) 74-599;1974:1-40. 8. Rose G. The basic principles of multifactorial prediction and control. In: Proceedings of a Symposium, from Epidemiology to Prevention, sponsored by the World Health Organization and the European Society of Cardiology. Geneva, Switzerland, 1974: 107-10. 9. The Pooling Project Research Group. Relationship of blood pressure, serum cholesterol, smoking habit, relative weight and ECG abnormalities to incidence of major coronary events: final report of the Pooling Project. J Chronic Dis 1978;31:201-306. 10. Reunanen A, pyoriilii K, Punsar S, Aromaa A. Predictive value of ECG findings with respect to coronary heart disease mortality. Adv Cardiol 1978;21:310-2. II. Cedres LB, Liu K, Stamler J, et al. Independent contribution of electrocardiographic abnormalities to risk of death from coronary heart disease, cardiovascular diseases and all causes: findings of three Chicago epidemiologic studies. Circulation 1982;65: 146-53. 12. Cullen K, Stenhouse NS, Wearne KL, Cumpston GN. Electrocardiograms and 13 year cardiovascular mortality in Busselton study. Br Heart J 1982;47:209-12. 13. Stamler J, Rhomberg P, Schoenberger JA, et al. Multivariate analysis of the relationship of seven variables to blood pressure: findings of the Chicago Heart Association Detection Project in Industry, 1%7-1972. J Chronic Dis 1975;28:527-48. 14. Berkson DM, Stamler J, Stevens E, Soyugenc R, Smoot E. The electrocardioanalyzer. Isr J Med Sci 1%9;5:687-90. 15. US Department of Health Education and Welfare. Heart disease in Adults, United States, 1960-1962. National Health Survey, National Centerfor Health Statistics. Series 11, No.6, Washington DC, 1964:28. 16. Blackburn H, Keys A, Simonson E, RautahaIju P, Punsar S. The electrocardiogram in population studies: a classification system. Circulation 1960;21:]]60-75. 17. International Classification of Diseases, 8th revision. Adapted for use in the United States (lcda), Vol. I (PHS), 1693. Washington DC: National Center for Health Statistics, 1967:49-522. 18. Hull CH, Nie NH. SPSS Update 7-9. New Procedures and Facilities for Releases 7-9. New York: McGraw-Hili, 1981:205-19. 19. Cox DR. Regression models and life tables. J R Stat Soc Series B 1972;34: 187-220. 20. Higgins ITT, Kannel WB, Dawber TR. The electrocardiogram in epidemiological studies: reproducibility, validity, and international comparison. Br J Prev Soc Med 1965;19:53-68. 21. Reunanen A, Aromaa A, Pyoriilii K, Punsar S, Maatela J, Knekt P. The Social Insurance Institution's Coronary Heart Disease Study: baseline data and 5-year mortality experience. Acta Med Scand 1983 (suppi673): 1-120. 22. Ostrander LD, Brandt RL, Kjelsberg MO, Epstein FH. Electrocardiographic findings among the adult population of a total natural community, Tecumseh, Michigan. Circulation 1%5;31:888-98. 23. Kennelly BM, Truswell AS, Schrire V. A clinical and electrocardiographic study of Kung bushmen. S Afr Med J 1972;46:1093-7. 24. Sinnett PF, Whyte HM. Epidemiological studies in a total highland population, Tukisenta, New Guinea: cardiovascular disease and relevant clinical, electrocardiographic, radiological and biochemical findings. J Chronic Dis 1973;26:265-90. 25. Rose G. Predicting coronary heart disease from minor symptoms and electrocardiographic findings. Br J Prev Soc Med 1971 ;25:94-6. 26. Eliot RS, Bratt G. The paradox of myocardial ischemia and necrosis in young women with normal coronary arteriograms: relation to abnormal hemoglobin-oxygen dissociation. Am J Cardiol 1969;23:633-8. 27. Kemp HG, Elliott WC, Gorlin R. The anginal syndrome with normal coronary arteriography. Trans Assoc Am Physicians 1967;80:59-70. 28. Hancock EW, Cohn K. The syndrome associated with midsystolic click and late systolic murmur. Am J Cardiol 1%6;41:183-96. 29. The Coronary Drug Project Research Group. The prognostic importance of the electrocardiogram after myocardial infarction. Ann Intern Med 1972;77:677-'139.