Aug 24, 1993 - among the white and black men of the Charleston Heart Study. Methods and Results. The 1960 baseline tracings of men ages 35 to 74 in the ...
2685
Electrocardiographic Abnormalities and 30-Year Mortality Among White and Black Men of the Charleston Heart Study Susan E. Sutherland, PhD; Peter C. Gazes, MD; Julian E. Keil, DrPH; Gregory E. Gilbert, MSPH; Rebecca G. Knapp, PhD Background. The long-term predictive significance of a single ECG tracing for mortality was explored the white and black men of the Charleston Heart Study. Methods and Results. The 1960 baseline tracings of men ages 35 to 74 in the Charleston Heart Study cohort were coded according to the Minnesota classification. Tracings were categorized as being normal or having minor or major abnormalities. The 30-year vital status was ascertained for the cohort, and the association between ECG findings and coronary and all-cause mortality was evaluated. The proportion of black men with major abnormalities at the 1960 baseline examination was almost twice that ofwhite men. Rates of all-cause mortality increased with severity of abnormalities for white and black men. The absolute excess risk for black men with major abnormalities was 23.3 per 1000 person-years and 12.8 for white men. The excess risk for coronary mortality was 73 for white men and 6.5 for black men. Conclusions. Many of the findings in this study confirm earlier associations derived from studies ofwhite populations and extend the observations to black men. However, the magnitude of the relative risk for mortality was different for white and black men. After controlling for traditional coronary disease risk factors and minor abnormalities, white men with major abnormalities were 2.72 (95% confidence interval, 1.47, 5.04) times more likely to die of coronary disease compared with black men, who were 1.95 (95% confidence interval, 0.93, 4.11) times more likely to die of coronary disease. (Circulation. among
1993;88:2685-2692.) KEY WORDs * mortality
*
race
*
electrocardiography
Although the use of resting ECGs has become common in epidemiological studies, there continues to be a paucity of information concerning the prognostic significance of ECG findings in various racial groups. While it has been documented that there are racial differences in the prevalence of some findings, such as increased R-wave amplitudes,1 left ventricular hypertrophy,2-5 and T-wave abnormalities,6 there are few studies that have examined the predictive value of these and other findings such as left-axis deviation and ST-T wave changes in subsequent mortality. A 20-year follow-up of the Evans County, Georgia Heart Study cohort found that even though the prevalence of major ECG abnormalities was greater in black men than in white men, the effect of ECG abnormalities on mortality was similar in blacks and whites.7 Rautaharju et a14 estimated left ventricular mass from ECGs among the NHANES I sample and Received March 18, 1993; revision accepted August 24, 1993. From The Charleston Heart Study (S.E.S., J.E.K., G.E.G., R.G.K.), Department of Biometry and Epidemiology, and the Department of Medicine (P.C.G.), Division of Cardiology, The Medical University of South Carolina, Charleston, SC. Correspondence to Dr Susan E. Sutherland, The Charleston Heart Study, Department of Biometry and Epidemiology, Room 908 Harborview Office Towers, Medical University of South Carolina, 171 Ashley Ave, Charleston, SC 29425.
found that left ventricular hypertrophy was a significant risk factor for cardiovascular mortality among white and black men. The completion of 30 years of follow-up in the Charleston Heart Study (CHS) provides an additional opportunity to examine the long-term effects of ECG findings in a biracial cohort. The primary goal of this study is to determine if there are racial differences in the prevalence of ECG findings and to determine if the predictiveness of these findings is different in blacks and whites. Methods The CHS, which began in 1960, was based on a random sampling of Charleston County residents who were at least 35 years of age. The sampling plan and methodology, which have been previously described, yielded 2181 respondents, representing an 84% response rate.8 The 1960 baseline examination included 653 white men, 741 white women, 333 black men, and 454 black women. Eight-lead ECGs were secured on all subjects using six limb leads, V2, and V4. In addition, medical histories, sociodemographic information, smoking histories, measurements of blood pressure, and serum cholesterol assays were obtained at the baseline examination. Body mass index (BMI) was calculated as weight divided by height squared (kg/M2) from the
2686
Circulation Vol 88, No 6 December 1993 TABLE 1. Vital Status of Black and White Men With ECG Tracings in the Charieston Heart Study Cohort (1960 to 1990)
Died Total, n* White men 642 Black men 328 High socioeconomic status black men 102 CHD indicates coronary heart disease. *Number with Minnesota-coded ECG tracings.
baseline examination and was used as a measure of obesity. Because the original sample of black participants was from a predominantly low range of socioeconomic levels, a cohort of 102 high socioeconomic status (SES) black men was recruited in 1963 to allow testing of social hypotheses concerning coronary disease. This group, assembled by peer nomination and primarily composed of professionals, proprietors, and school principals, represented a 93% response of all identified high SES black men in Charleston. Data collected on this group during their baseline examination included physical assessments, medical histories, and ECGs, which were obtained in a similar manner to the 1960 examination.
The vital status of the original CHS cohort was updated through December 1990, and underlying cause of death from the International Classification of Diseases (ICD) was recorded for each decedent. As shown in Table 1, the vital status was known for 98.5% of the male participants. The mortality end points in this study included deaths from all causes and deaths from coronary heart disease (CHD). Deaths were coded by the nosologist according to the eighth revision of the ICD for deaths occurring in 1960 through 1978 and according to the ninth revision for deaths in 1979 through 1990. ICD codes 410-414 from the ninth revision and codes 410-413 from the eighth revision were used to define CHD mortality. The ECGs recorded at the baseline examination (either in 1960 or 1963) were subsequently coded in 1987 according to the Minnesota code for 98.5% of male participants. ECGs were not coded for female participants. All tracings were classified as normal or having a major or a minor abnormality according to the criteria defined by the Pooling Project Research Group (Table 2).9 Persons with both major and minor abnormalities were classified as having a major abnormality. The
CHD 138 55 8
Other 256 166 31
Alive 235 104 63
Lost 13 3 0
specific components of the major and minor abnormalities are presented in Table 3 along with the number of occurrences of each type. The major abnormalities were represented by either ST depression (4-1 and 4-2), T-wave inversions (5-1 and 5-2), or premature beats (8-1). The greatest proportion of minor abnormalities was represented by T-wave inversions (5-3), low QRS amplitude (9-1), high amplitude R waves (3-1), and left-axis deviation (2-1). Original tracings were read by a single cardiologist (P.C.G.) before Minnesota coding and without regard to age or race. Inclusion of four specific clinical interpretations -left-axis deviation, early repolarization, left ventricular hypertrophy, and nonspecific T-wave/ST segment changes-added another dimension to the analyses by highlighting findings that might be obscured by only including major and minor classifications as defined by the Pooling Project.9 In addition, this allowed a comparison of ECG interpretations by a clinician with Minnesota coding. The clinical interpretations considered in these analyses demonstrated a high concordance with corresponding categories of the Minnesota codes. A clinical diagnosis of left-axis deviation, corresponding with Minnesota code 2-1, yielded 95% concordance. Because there were no participants with known current injury, ST elevation was considered as early repolarization that agreed with Minnesota code 9-2 in 78% of the participants. Minnesota codes 5-1 through 5-3 and 4-1 through 4-3 were compared with the cardiologist's classification of nonspecific T-wave and ST segment changes and resulted in 86% concordance. Although the criteria were slightly different for comparison of left ventricular hypertrophy, the concordance was high. The criteria used by the clinician was the point system described by Romhilt and Estes with ST-T changes.10 Even though this does not conform to just R amplitude as noted with Minnesota code 3-1, there was agreement between the clinician's assessment of left ventricular
TABLE 2. Percentage of Black and White Men With ECG Abnormalities White Men
Black Men
High Socioeconomic Status Black Men
(n=584)
(n=307)
(n=102)
467 (80.0%) 77 (13.2%) 40 (6.9%)
216 (70.4%) 49 (16.0%) 42 (13.7%)
78 (76.5%) 19 (18.6%) 5 (4.9%)
Comparison* x2= 10.4, P
