- Email: [email protected]
Resting electrocardiographic abnormalities as predictors of coronary events and total mortality among elderly men
Resting Electrocardiographic Abnormalities as Predictors of Coronary Events and Total Mortality among Elderly Men Markku Tervahauta, MD, Juha Pekkanen, MD, Sven Punsar, MD, Aulikki Nissinen, MD, Kuopio, f’infand
To examine the prognostic significance of electrocardiographic (ECG) abnormalities among the elderly. MATERIALS AND METHODS: The Finnish cohorts of the Seven Countries Study involved 697 men aged 65 to 84 years at baseline in 1984. A 5 year follow-up was made from 1984 to 1989. Fatal myocardial infarction, nonfatal myocardial infarction, and all-cause mortality were outcome measures. RESULTS: Seventy-four fatal myocardial infarctions (Ml), 101 fatal or nonfatal MIS, and 207 deaths occurred. When electrocardiographic changes were analyzed one by one, men with Q waves (n = 98), high-amplitude R waves (n = 112), depressed ST-interval (n = 122) or T-wave changes (n = 263) had significantly (P < 0.05) higher risk of coronary events and all-cause mortality than men without these changes. Additionally, men with atrial fibrillation (n = 49) had significantly higher risk of death. Highest risk was observed among men with Q waves together with ST- or T-wave changes. Men with both ST depression and T flattening/inversions without Q waves had also increased risk, whereas this was not true for men with Q waves without concomitant ST- or T-wave changes. CONCLUSION: Electrocardiographic abnormalities suggestive of coronary heart disease are associated with a high risk for coronary events and total mortality among elderly men. Among the elderly, a reliable history of coronary heart disease may not be easily achievable, thus the ECG could potentially be used as an indicator of symptomless or atypical heart disease. Am J Med. 1996;100:641-645. PURPOSE:
From the Department of Community Health and General Practice, University of Kuopio, and Department of Environmental Epidemiology, National Public Health Instrtute. KUODIO. Finland. The present study was supported’by funding from the National Institute on Aging. USA (grant number EDC-1 1 R01 AG08762-OlAi). Finnish Academyof>cience~ the Finnish Heart Foundation, and the Finnish Medical Foundation. Rearrests for reorints should be addressed to Markku Tervahauta. MD. Department of Community Health and General Practice, University of Kuooio. PO Box 1627. FIN-70211 Kuooio. Finland. Manuscript submitted September 2i, 1995, and accepted in revised form February 15, 1996.
01996 by Excerpta All rights reserved.
Medica,
Inc.
S
ymptoms of heart diseases may more often be missing or atypical among the elderly than in younger age groups. Thus ECG changes could be valuable indicators of heart diseases in clinical practice especially among the aged. Pathological processes associated with cardiovascular diseases are probably responsible for most of the changes in electrocardiograms (ECG) l-6 among the elderly, although also aging alone appears to induce alterations in the cardiac structure and function.“,7 In resting 12-lead ECG, both among the middle aged’-” and the elderly, 12-14findings suggestive of coronary heart disease (CHD) and/or cardiovascular diseases strongly predict coronary events and total mortality. We studied how ECG changes predict CHD events and all-cause mortality among 697 Finnish men aged 65 to 84 years drawn from a population with a high morbidity and mortality from CHD.
MATERIALS AND METHODS The Seven Countries Study aimed to study the incidence of coronary heart disease and associated risk factors in several countries including Finland. The Finnish part of the study included two cohorts selected from geographically defined rural areas: Ilomantsi in eastern Finland (East), a region having a very high CHD mortality, and Poytya and Mellila in south-western Finland (West) having a lower CHD mortality. The original cohorts invited to participate in the study in 1959 consisted of all men aged 40 to 59 years living in these areas, 823 men in the East and 888 in the West.i5 The present analysis includes men who participated in the 25-year follow-up study in 1984 (n = 716, 93% of the 766 men alive of the original cohorts). For 697 men (91% of those alive) a standard 12-lead resting ECG was available. The field study included questionnaires concerning angina pectoris, previous myocardial infarction, and smoking habits, and clinical examination, which was performed by a physician according to the original study protocol.16 The ECGs were encoded independently according to the Minnesota coding rulesL7 by one of the authors (SP) and another coder. In cases of disagreement, coding was based on consensus of the two coders. Death certificates and hospital records ‘were collected for all men who died between the 25- (1984) 0002-9343/96/$15.00 PII SOOO2-9343(96)00042-3
641
ECG AND MORTALITY IN THE ELDERLY/lERVAHAUTA
ET AL
and 30-year (1989) follow-up examinations. We reviewed also the hospital records for all men who at the 30-year follow-up examination reported possible MI during the last 5 years. Additionally, for the study population, all hospital discharge diagnoses with ICD-8 codes 410-414 were identified from the National Hospital Discharge Register, and hospital records were collected and reviewed. One of the authors (MT) classified all possible myocardial infarctions during the study period, following as closely as possible the MONICA criteria.” As ECGs during the acute disease episode were not examined for the present study, ECG information was classified into five categories according to the notes by the clinician in the hospital records.lg To study the number of silent myocardial infarctions we identified men with major Q waves in ECG in the 30-year but not in the 25year follow-up examination and with no reported history of myocardial infarction in interim. Altogether 12 men had a new Q wave in their ECG suggestive of MI that had occurred between the 25 and 30-year follow-up examinations. Eight of these men had no history of MI and had thus experienced a silent MI. In the present analysis, fatal events coded as definite or possible acute myocardial infarction were considered as fatal MIS, and nonfatal cases coded as definite myocardial infarction were considered as nonfatal MIS. Three end points were defined for the analysis: fatal MI, any MI (fatal or nonfatal), and any death. We first studied the incidence of all three end points within each Minnesota code item without considering possible coexistence of other items. We then divided men into 7 hierarchical groups as follows: (1) Men with none of the major abnormalities, as listed below, according to the Minnesota coding system; (2) men with Q waves (Minnesota code l:l-3) combined with either depressed ST segment (horizontal or downward sloping at least 0.5 mm [Minnesota code 4:1-31) or T inversion (Minnesota code 5:1, 2); (3) men with Q waves, but no depressed ST segment or T inversion; (4) men with depressed ST segment (Minnesota code 4: l-3), but no codable Q waves; (5) men with T abnormality (Minnesota code 5: l3), but no codable Q waves or depressed ST segment; (6) men with high amplitude left R waves (Minnesota code 3:1,3), but no depressed ST segment, T abnormality, or codable Q wave; (7) men with arrhythmias (Minnesota code 8; all items). There were no significant interactions between ECG changes and analyses of reported history of angina pectoris (assessed with a questionnaire closely similar to that published by the World Health Organization l7 or previous MI; thus, all analyses were 642
June 1996
The American
Journal
of Medicinea
Volume
100
performed combining men with and without a prehistory of CHD. Also differences between the two study areas were statistically nonsignificant (P > 0.1). Statistical analyses were performed using SAS software.2o Differences in risk between categories of each Minnesota code item were studied using ageand area-adjusted chi square tests (Cochran-MantelHaentzel statistics) .” Logistic regression models were used to study the association of grouped ECG findings with the three end points. The latter analyses were repeated adjusting for major CHD risk factors such as serum total cholesterol, smoking, and blood pressure.
vious
RESULTS During the g-year follow-up period, 101 (15%) of the 697 men included in the present study had any MI, 74 deaths (11%) were due to fatal MI, and 207 deaths (30%) were due to all causes. ‘Fable I presents the prevalences of ECG abnormalities in 1984 and their predictive value in 1984 to 1989. Fifty-three men (8%) had in 1984 a Q wave in the ECG, suggesting previous MI (Minnesota codes l:l, 2); 26 (49%) of these men had Q-wave changes only and no other abnormalities suggestive of CHD (major ST-interval and T-wave changes, or atrial fibrillation) . Eighty-one men ( 12%) had an ischiemic ST interval with a J depression of at least 0.5 :mm. Mostly these abnormalities were concomitant with either Qwave or T-wave changes (the latter association, however, being partially technical due to the Minnesota coding rules). Two hundred and eighty three men (38%) had codable and 121 (17%) major (Minnesota code 5:1, 2) T-wave abnormalities. One hundred and twenty seven (48%) of the 263 men with codable T-wave changes had these without accompanying major CHD suggestive Q wave (Minnesota code 1: 1,2) or ST patterns (Minnesota code 4: l-3). Q waves predicted considerably increased risk of CHD end points and also of all-cause mortality. These associations were nonsignificant (P > 0.1) when adjusted for the history of CHD. QRS-axis deviation was not associated with increased risk of either CHD or all-cause mortality. High-amplitude R waves were associated with significantly increased risk for all three end points. This result remained significant even when adjusted for history of CHD in the case of fatal MI and any MI. Also major ST depressions and T inversions were related to high risk of all three end points, independently from previous history of CHD. For total mortality atria1 fibrillation was associated with high risk, which, however, became statistically nonsignificant when a’cljusting for history of CHD. Atrioventricular and ventricular con-
TABLE
I
Prevalence
and Predictive
Value
of Changes
in Resting
Electrocardiogram
among
697
Men
Aged
65 to 84 years Total
Electrocardiogram Abnormality
Minnesota Code
Q-wave items
1
QRS deviation
2
R-wave items
3
ST changes
4
n
0 1 2 3 Ptdf 3)’ 0 1 3 5 P tdf 3) 0 1 2 3 P Idf 3) 0 1 2 3 4
No codable changes Large Q and QS Intermediate Q and QS Small Q and QS
599 21 32 45
No codable changes Left t-300 to -90”) Right (+120” to -150”) Indeterminate
603 83 7 4
No codable changes High left High right Smaller left
585 60 3 49
No codable changes Horizontal or downward sloping ST segment, STJ depression Z=1 .O mm Horizontal or downward sloping ST segment, STJ depression ~0.5 mm but cl.0 mm Horizontal or downward sloping ST segment to.5 mm, STJ depression ~0.5 mm ST segment upward sloping, STJ depression 2 1 .O mm
575 33
No codable changes T-wave negative; >5 mm -1 to -5 mm Oto-lmm Positive T, T/R ratio
434 10 111 106 36
No codable changes
613
Fatal
MI
Any MI
n
(%)
n
(%)
-~ Mortality n (%I
53 9 4 19 9 28 8 18
78 13 5 24 10 31 8 18 0.015 89 15 10 12 1 14 1 25 n.s. 73 12 15 25 2 67 11 22 0.002 58 10 18 55
48
14
29
16
33
28
58
38
6
16
8
21
17
45
3
0
0
1
33
1
33
T-wave inversion/flattening
5
Atrioventricular conduction defects
6
P (df 4) 0 1 2 3 4 P tdf 4) 0
Complete (39 atrioventricular block PQ interval 0.22 s or more PR (PQ) interval co.12 s
7
1 3 5 Pfdf3) 0 1 2
Complete left bundle branch block Complete right bundle branch block Incomplete left bundle branch block lntraventricular block; QRS 0.12 s or more R-Rprime Incomplete left bundle branch block
13 33
0 7
0 21
0 7
0 21
5 17
38 52
9 8
0 0
0 0
2 0
22 0
3 2
33 25
19 5
2 1
11 20
2 1
11 20
No codable changes Any combination of codable arrhythmias Frequent premature (10% or more) atrial, nodal or ventricular beats Atrial fibrillation Sinus tachycardia t>lOO/min) Sinus bradycardia (c50/min) Other arrhythmias
571 2
Ventricular conduction defects
3 4 5 6 PM61 Arrhythmias
8 0 1 3 7 8 9 P (df 6)
No codable changes
4 76 4 610
167 28 1.0 48 1.5 47 1.5 33 0.027 175 29 29 35 2 29 1 25 n.s. 165 28 213 38 3 100 16 36 0.034 136 24 25 76
1 8 1
25 11 25
64n”lO
co.001 101 23 7 70 59 53 29 27 11 31 co.001 179 29
1 13 1
2 23 3 n.s. 174
25 17 25
89n.5’15
61ns 11 1 50
83=15 1 50
50 30 75 29
5 26 1 20 n.s. 160 28 2 100
40
4
10
6
15
12
49 11 22 2
5 1 2 0 ns.
10 9 9 0
6 2 3 0
12 18 14 0
24
ns.
30
49 45 : 14 1 50 0.003
* P values from age and area adjusted chi-square test. MI = myocardial infarction; n.s. = not significant (P > 0.01).
June 1996
The American
Journal
of Medicine@
Volume
100
643
ECG AND MORTALITY IN THE ELDERLY/lERVAHAUTA
ET AL
No major abnormalities
E Q-waves
and ST/T
patterns 0
Q-waves
ST-T
Fatal MI
@>Any MI 0 El Total mortality
only
depression only
T inversion/flattening
only
High left R
Arrhythmias
0.01
0.1
1
10
100
Odds ratio (95% Cl) Figure. Risk for fatal myocardial infarction (MI), any MI, and total mortality among 65- to 84-year old men by hierarchical classification of electrocardiographic (ECG) abnormalities. Risk presented in odds ratios based on estimates from age- and area-adjusted logistic regression models. Hierarchical groups are as follows: Men with none of the major abnormalities, as listed below, according to the Minnesota coding system; men with Q waves (Minnesota code l:l-3) combined with either depressed ST segment (horizontal or downward slospingat least 0.5 mm [Minnesota code 4:1-31) or T inversion (Minnesota code 5:1, 2); men with Q waves, but no depressed ST segment or T inversion; men with depressed ST segment (Minnesota code 4:1-31, but no codable Q waves; men with T abnormality (Minnesota code 5:1-31, but no codable Q waves or depressed ST segment; men with high-amplitude left R waves (Minnesota code 3:1, 31, but no depressed ST-segment or T-wave abnormality or codable Q waves; and men with arrhythmias (Minnesota code 8; all items).
than among younger age groups. Also, among the aged, heart disease may be symptomless due to decreased physical activity. Thus, especially among the aged, ECG findings could potentially be used in clinical practice as indicators of symptomless heart disease. Among the elderly, major ECG findings associated with CHD/CVD have significantly predicted coronary events and total mortality.12-14~21 In contrast, minor ECG changes do not appear to be associated with increased risk of mortality.‘2~‘3 Atrial fibrillation, DISCUSSION first-degree atrioventricular block, 13V21 left bundle Among the elderly, prognostic significance of spe- branch block, 21 and left ventricular hypertrophy12 cific ECG abnormalities is mostly due to the under- have been found to be significant predictors of morlying heart disease.” However, among the elderly, a tality. Results of the present study are mostly in line reliable history of previous CHD is harder to obtain with previous studies. Especially when coincident,
duction defects and other arrhytmias did not show significant associations with the end points. For the ECG groupings, constructed for the present analysis, results were in accord with those reported above (Table II, Figure). Especially among men with ST-T changes, with or without concomitant Q waves, there was an increased risk compared with men with no codable abnormality. Estimates for relative risk did not change significantly when adjusted for CHD risk factors (data not shown).
644
June 1996
The American
Journal
of Medicinea
Volume
100
ECG AND MORTALITY IN THE ELDERLY/TERVAHAlJTA TABLE II Hierarchical Classification of Electrocardiogram (ECG) Abnormalities and Numbers of Studied End Points among Men Aged 65 to 84 Years Category of Men with ECG Fatal MI Any Ml Total Mortality Abnormalities* n n (%) n (%) n (%) 1 325 22 (7) 34 (10) 73 (22) 2 36 13 (36) 15 (42) 19 (53) 3 8 (13) 21 (341 4 E 2: Ki 28 (30) 55 (601 5 85 15 (18) 3 (41 9 (11) 6 39 2 (5) 4 (10) 6 (15) 7 58 3 (5) 3 (5) 18(31) Total 697 74 101 207 ’ 1. Men with none of the major abnormalities, as listed below, according to the Minnesota coding system. 2. Men with Q-waves (Minnesota code l:l-3) combined with either depressed ST-segment fhorizontal or downward sloping at least 0.5 mm (Minnesota code 4:1-311 or T-inversron (Minnesota code 5:1, 2). 3. Men with Q-waves, but no depressed ST-segment or T-inversion. 4. Men with depressed ST-segment (Minneosta code 4:1-31, but no codable Q-waves. 5. Men wrth T-abnormality (Minnesota code 5:1-3), but no codable Qwaves or depressed ST-segment. 6. Men with high amplitude left R-waves (Minnesota code 3:1, 31, but no depressed ST-segment, T-abnormality or codable Q-waves. 7. Men with arrhythmias (Minnesota code 8; all items). Ml = myocardial infarction.
ST-segment and T-wave abnormalities and Q waves suggestive of CHD, as well as high-amplitude R waves, were associated with significantly increased risk of both coronary events and mortality. Adjusting for history of previous CHD reduced the significance of Q waves, probably because of considerable colinearity, but not that of ST-segment and T-wave changes nor high R waves. Combination of ST-segment and T-wave abnormalities appeared to be highly predictive independently from Q waves. This seems to indicate, also among the elderly, the more acute nature of ST-T changes whereas Q waves may be evidence of previous damage in the heart. The study population was relatively small in number. Consequently, we were not able to study possible increases in risk associated with infrequent ECG changes. Adjusting the present analyses for classical coronary risk factors such as cholesterol, smoking, and blood pressure did not change the observed associations. These findings were in line with results from previous studies.g We conclude that ECG abnormalities suggestive of CHD are associated with high risk for coronary events and total mortality among elderly men. Among the elderly, a reliable history of previous CHD may not be obtained and heart disease symptoms may be atypical or missing due to decreased physical activity. Therefore, especially among the
ET AL
aged, ECG findings could be valuable in clinical practice as indicators of symptomless heart disease and in determining which patients should be referred for further noninvasive testing or cardiac catheterization. Additionally, ECG findings may be helpful when identifying patients at high risk for perioperative or postoperative cardiac problems involving noncardisc surgery.
REFERENCES 1. Gorman P, Calatayud J, Abraham S, Caceres CA. Effects of age and heart disease on the QRS axis during the seventh through the tenth decades. Am J Med. 1964;67:39-43. 2. Blackburn H, Vasquez C, Keys A. The aging electrocardiogram, a common agrng process or latent coronary artery disease? Am J Cardiol. 1967;20:618627. 3. Simonson E. The effect of age on the electrocardiogram. Am J Cardiol. 1972;29:64-73. 4. Mihalick M, Fisch C. Electrocardiographrc changes in the aged. Am Heart J. 1974;87:117-128. 5. Fisch C. Electrocardiogram in the aged: an independent marker of heart disease. Am J Med. 1981;70:4-6. 6. Bachman S, Sparrow D, Smith K. Effect of aging on the electrocardiogram. Am J Cardiol. 1981;48:513-515. 7. Fleg JL. Alternations in cardiovascular structure and function with advancing age. Am J Cardiol. 1986;57:33C-44C. 8. Rose G, Baxter PJ, Reid DD, McCartney P. Prevalence and prognosis of electrocardiographic findrngs in middleaged men. Br Heart J. 1978;40:636643. 9. Cedres BL, Liu K, Stamler J, et al. Independent contribution of electrocardiographic abnormalities to risk of death from coronary heart disease, cardio vascular diseases and all causes. Circulation. 1982;65:146-153. 10. Blackburn H, Taylor HL, Keys A. Coronary heart disease in seven countries XVI. The electrocardiogram in prediction of five year coronary heart disease incidence among men aged forty through fifty-nine. Circulation. 1970;4142fsuppl 1):1154-1161. 11. Cullen KJ, Stenhouse NS, Wearne KL, Cumpston GN. Electrocardiograms and thirteen year cardiovascular mortality in Busselton study Br Heart J. 1982;47:209-212. 12. Caird F, Campbell A, Jackson T. Significance of abnormalities of electrocardiogram in old people. Br Heart J. 1974;36:1012-1018. 13. Rajala S, Haavisto M, Kaltiala K, Mattila K. ECG findrngs and survival in very old people. Eur Heart J. 1985;6:247-252. 14. Dekker JM Electrocardiographic predictors of future coronary heart disease: a possible role of autonomic control. Wageningen, The Netherlands: Wageningen Agricultural University; 1994. 15. Karvonen M, Blomqvist G, Kallio V, et al. Men in rural East and West Finland. Acta Med &and. 1967;46OfsupplJ:169-190. 16. Keys A. Coronary heart disease in seven countries. Circulation. 1970;41(suppl lkl-211. 17. Rose GA, Blackburn H. Cardiovascular Survey Methods. Geneva: World Health Organrzation (World Health Organization Monograph Series No. 56); 1968. 18. FINMONICA. Monitoring of trends and determinants of cardiovascular disease in Finland (part of a joint study). The MONICA project. Helsinki: National Public Health Institute, 1986. 19. Tervahauta M, Pekkanen J, Enlund H, Nissinen A. Change in blood pressure and 5year risk of coronary heart disease among elderly men. The Finnish cohorts of the Seven Countries Study. J Hypertens. 1994;12:1183-1189. 20. SAS Institute. SAS User’s Guide: Statistics. Cary, North Carolina: SAS Institute; 1990. 21. Casrglia E, Spolaore P, Ginocchio G, et al. Mortality in relation to Minnesota code items in elderly subjects. Sex-related differences in a cardiovascular study In the elderly. Jpn Heart J. 1993;34:567-577.
June 1996
The
American Journal
of Medicine@
Volume
100
645
To examine the prognostic significance of electrocardiographic (ECG) abnormalities among the elderly. MATERIALS AND METHODS: The Finnish cohorts of the Seven Countries Study involved 697 men aged 65 to 84 years at baseline in 1984. A 5 year follow-up was made from 1984 to 1989. Fatal myocardial infarction, nonfatal myocardial infarction, and all-cause mortality were outcome measures. RESULTS: Seventy-four fatal myocardial infarctions (Ml), 101 fatal or nonfatal MIS, and 207 deaths occurred. When electrocardiographic changes were analyzed one by one, men with Q waves (n = 98), high-amplitude R waves (n = 112), depressed ST-interval (n = 122) or T-wave changes (n = 263) had significantly (P < 0.05) higher risk of coronary events and all-cause mortality than men without these changes. Additionally, men with atrial fibrillation (n = 49) had significantly higher risk of death. Highest risk was observed among men with Q waves together with ST- or T-wave changes. Men with both ST depression and T flattening/inversions without Q waves had also increased risk, whereas this was not true for men with Q waves without concomitant ST- or T-wave changes. CONCLUSION: Electrocardiographic abnormalities suggestive of coronary heart disease are associated with a high risk for coronary events and total mortality among elderly men. Among the elderly, a reliable history of coronary heart disease may not be easily achievable, thus the ECG could potentially be used as an indicator of symptomless or atypical heart disease. Am J Med. 1996;100:641-645. PURPOSE:
From the Department of Community Health and General Practice, University of Kuopio, and Department of Environmental Epidemiology, National Public Health Instrtute. KUODIO. Finland. The present study was supported’by funding from the National Institute on Aging. USA (grant number EDC-1 1 R01 AG08762-OlAi). Finnish Academyof>cience~ the Finnish Heart Foundation, and the Finnish Medical Foundation. Rearrests for reorints should be addressed to Markku Tervahauta. MD. Department of Community Health and General Practice, University of Kuooio. PO Box 1627. FIN-70211 Kuooio. Finland. Manuscript submitted September 2i, 1995, and accepted in revised form February 15, 1996.
01996 by Excerpta All rights reserved.
Medica,
Inc.
S
ymptoms of heart diseases may more often be missing or atypical among the elderly than in younger age groups. Thus ECG changes could be valuable indicators of heart diseases in clinical practice especially among the aged. Pathological processes associated with cardiovascular diseases are probably responsible for most of the changes in electrocardiograms (ECG) l-6 among the elderly, although also aging alone appears to induce alterations in the cardiac structure and function.“,7 In resting 12-lead ECG, both among the middle aged’-” and the elderly, 12-14findings suggestive of coronary heart disease (CHD) and/or cardiovascular diseases strongly predict coronary events and total mortality. We studied how ECG changes predict CHD events and all-cause mortality among 697 Finnish men aged 65 to 84 years drawn from a population with a high morbidity and mortality from CHD.
MATERIALS AND METHODS The Seven Countries Study aimed to study the incidence of coronary heart disease and associated risk factors in several countries including Finland. The Finnish part of the study included two cohorts selected from geographically defined rural areas: Ilomantsi in eastern Finland (East), a region having a very high CHD mortality, and Poytya and Mellila in south-western Finland (West) having a lower CHD mortality. The original cohorts invited to participate in the study in 1959 consisted of all men aged 40 to 59 years living in these areas, 823 men in the East and 888 in the West.i5 The present analysis includes men who participated in the 25-year follow-up study in 1984 (n = 716, 93% of the 766 men alive of the original cohorts). For 697 men (91% of those alive) a standard 12-lead resting ECG was available. The field study included questionnaires concerning angina pectoris, previous myocardial infarction, and smoking habits, and clinical examination, which was performed by a physician according to the original study protocol.16 The ECGs were encoded independently according to the Minnesota coding rulesL7 by one of the authors (SP) and another coder. In cases of disagreement, coding was based on consensus of the two coders. Death certificates and hospital records ‘were collected for all men who died between the 25- (1984) 0002-9343/96/$15.00 PII SOOO2-9343(96)00042-3
641
ECG AND MORTALITY IN THE ELDERLY/lERVAHAUTA
ET AL
and 30-year (1989) follow-up examinations. We reviewed also the hospital records for all men who at the 30-year follow-up examination reported possible MI during the last 5 years. Additionally, for the study population, all hospital discharge diagnoses with ICD-8 codes 410-414 were identified from the National Hospital Discharge Register, and hospital records were collected and reviewed. One of the authors (MT) classified all possible myocardial infarctions during the study period, following as closely as possible the MONICA criteria.” As ECGs during the acute disease episode were not examined for the present study, ECG information was classified into five categories according to the notes by the clinician in the hospital records.lg To study the number of silent myocardial infarctions we identified men with major Q waves in ECG in the 30-year but not in the 25year follow-up examination and with no reported history of myocardial infarction in interim. Altogether 12 men had a new Q wave in their ECG suggestive of MI that had occurred between the 25 and 30-year follow-up examinations. Eight of these men had no history of MI and had thus experienced a silent MI. In the present analysis, fatal events coded as definite or possible acute myocardial infarction were considered as fatal MIS, and nonfatal cases coded as definite myocardial infarction were considered as nonfatal MIS. Three end points were defined for the analysis: fatal MI, any MI (fatal or nonfatal), and any death. We first studied the incidence of all three end points within each Minnesota code item without considering possible coexistence of other items. We then divided men into 7 hierarchical groups as follows: (1) Men with none of the major abnormalities, as listed below, according to the Minnesota coding system; (2) men with Q waves (Minnesota code l:l-3) combined with either depressed ST segment (horizontal or downward sloping at least 0.5 mm [Minnesota code 4:1-31) or T inversion (Minnesota code 5:1, 2); (3) men with Q waves, but no depressed ST segment or T inversion; (4) men with depressed ST segment (Minnesota code 4: l-3), but no codable Q waves; (5) men with T abnormality (Minnesota code 5: l3), but no codable Q waves or depressed ST segment; (6) men with high amplitude left R waves (Minnesota code 3:1,3), but no depressed ST segment, T abnormality, or codable Q wave; (7) men with arrhythmias (Minnesota code 8; all items). There were no significant interactions between ECG changes and analyses of reported history of angina pectoris (assessed with a questionnaire closely similar to that published by the World Health Organization l7 or previous MI; thus, all analyses were 642
June 1996
The American
Journal
of Medicinea
Volume
100
performed combining men with and without a prehistory of CHD. Also differences between the two study areas were statistically nonsignificant (P > 0.1). Statistical analyses were performed using SAS software.2o Differences in risk between categories of each Minnesota code item were studied using ageand area-adjusted chi square tests (Cochran-MantelHaentzel statistics) .” Logistic regression models were used to study the association of grouped ECG findings with the three end points. The latter analyses were repeated adjusting for major CHD risk factors such as serum total cholesterol, smoking, and blood pressure.
vious
RESULTS During the g-year follow-up period, 101 (15%) of the 697 men included in the present study had any MI, 74 deaths (11%) were due to fatal MI, and 207 deaths (30%) were due to all causes. ‘Fable I presents the prevalences of ECG abnormalities in 1984 and their predictive value in 1984 to 1989. Fifty-three men (8%) had in 1984 a Q wave in the ECG, suggesting previous MI (Minnesota codes l:l, 2); 26 (49%) of these men had Q-wave changes only and no other abnormalities suggestive of CHD (major ST-interval and T-wave changes, or atrial fibrillation) . Eighty-one men ( 12%) had an ischiemic ST interval with a J depression of at least 0.5 :mm. Mostly these abnormalities were concomitant with either Qwave or T-wave changes (the latter association, however, being partially technical due to the Minnesota coding rules). Two hundred and eighty three men (38%) had codable and 121 (17%) major (Minnesota code 5:1, 2) T-wave abnormalities. One hundred and twenty seven (48%) of the 263 men with codable T-wave changes had these without accompanying major CHD suggestive Q wave (Minnesota code 1: 1,2) or ST patterns (Minnesota code 4: l-3). Q waves predicted considerably increased risk of CHD end points and also of all-cause mortality. These associations were nonsignificant (P > 0.1) when adjusted for the history of CHD. QRS-axis deviation was not associated with increased risk of either CHD or all-cause mortality. High-amplitude R waves were associated with significantly increased risk for all three end points. This result remained significant even when adjusted for history of CHD in the case of fatal MI and any MI. Also major ST depressions and T inversions were related to high risk of all three end points, independently from previous history of CHD. For total mortality atria1 fibrillation was associated with high risk, which, however, became statistically nonsignificant when a’cljusting for history of CHD. Atrioventricular and ventricular con-
TABLE
I
Prevalence
and Predictive
Value
of Changes
in Resting
Electrocardiogram
among
697
Men
Aged
65 to 84 years Total
Electrocardiogram Abnormality
Minnesota Code
Q-wave items
1
QRS deviation
2
R-wave items
3
ST changes
4
n
0 1 2 3 Ptdf 3)’ 0 1 3 5 P tdf 3) 0 1 2 3 P Idf 3) 0 1 2 3 4
No codable changes Large Q and QS Intermediate Q and QS Small Q and QS
599 21 32 45
No codable changes Left t-300 to -90”) Right (+120” to -150”) Indeterminate
603 83 7 4
No codable changes High left High right Smaller left
585 60 3 49
No codable changes Horizontal or downward sloping ST segment, STJ depression Z=1 .O mm Horizontal or downward sloping ST segment, STJ depression ~0.5 mm but cl.0 mm Horizontal or downward sloping ST segment to.5 mm, STJ depression ~0.5 mm ST segment upward sloping, STJ depression 2 1 .O mm
575 33
No codable changes T-wave negative; >5 mm -1 to -5 mm Oto-lmm Positive T, T/R ratio
434 10 111 106 36
No codable changes
613
Fatal
MI
Any MI
n
(%)
n
(%)
-~ Mortality n (%I
53 9 4 19 9 28 8 18
78 13 5 24 10 31 8 18 0.015 89 15 10 12 1 14 1 25 n.s. 73 12 15 25 2 67 11 22 0.002 58 10 18 55
48
14
29
16
33
28
58
38
6
16
8
21
17
45
3
0
0
1
33
1
33
T-wave inversion/flattening
5
Atrioventricular conduction defects
6
P (df 4) 0 1 2 3 4 P tdf 4) 0
Complete (39 atrioventricular block PQ interval 0.22 s or more PR (PQ) interval co.12 s
7
1 3 5 Pfdf3) 0 1 2
Complete left bundle branch block Complete right bundle branch block Incomplete left bundle branch block lntraventricular block; QRS 0.12 s or more R-Rprime Incomplete left bundle branch block
13 33
0 7
0 21
0 7
0 21
5 17
38 52
9 8
0 0
0 0
2 0
22 0
3 2
33 25
19 5
2 1
11 20
2 1
11 20
No codable changes Any combination of codable arrhythmias Frequent premature (10% or more) atrial, nodal or ventricular beats Atrial fibrillation Sinus tachycardia t>lOO/min) Sinus bradycardia (c50/min) Other arrhythmias
571 2
Ventricular conduction defects
3 4 5 6 PM61 Arrhythmias
8 0 1 3 7 8 9 P (df 6)
No codable changes
4 76 4 610
167 28 1.0 48 1.5 47 1.5 33 0.027 175 29 29 35 2 29 1 25 n.s. 165 28 213 38 3 100 16 36 0.034 136 24 25 76
1 8 1
25 11 25
64n”lO
co.001 101 23 7 70 59 53 29 27 11 31 co.001 179 29
1 13 1
2 23 3 n.s. 174
25 17 25
89n.5’15
61ns 11 1 50
83=15 1 50
50 30 75 29
5 26 1 20 n.s. 160 28 2 100
40
4
10
6
15
12
49 11 22 2
5 1 2 0 ns.
10 9 9 0
6 2 3 0
12 18 14 0
24
ns.
30
49 45 : 14 1 50 0.003
* P values from age and area adjusted chi-square test. MI = myocardial infarction; n.s. = not significant (P > 0.01).
June 1996
The American
Journal
of Medicine@
Volume
100
643
ECG AND MORTALITY IN THE ELDERLY/lERVAHAUTA
ET AL
No major abnormalities
E Q-waves
and ST/T
patterns 0
Q-waves
ST-T
Fatal MI
@>Any MI 0 El Total mortality
only
depression only
T inversion/flattening
only
High left R
Arrhythmias
0.01
0.1
1
10
100
Odds ratio (95% Cl) Figure. Risk for fatal myocardial infarction (MI), any MI, and total mortality among 65- to 84-year old men by hierarchical classification of electrocardiographic (ECG) abnormalities. Risk presented in odds ratios based on estimates from age- and area-adjusted logistic regression models. Hierarchical groups are as follows: Men with none of the major abnormalities, as listed below, according to the Minnesota coding system; men with Q waves (Minnesota code l:l-3) combined with either depressed ST segment (horizontal or downward slospingat least 0.5 mm [Minnesota code 4:1-31) or T inversion (Minnesota code 5:1, 2); men with Q waves, but no depressed ST segment or T inversion; men with depressed ST segment (Minnesota code 4:1-31, but no codable Q waves; men with T abnormality (Minnesota code 5:1-31, but no codable Q waves or depressed ST segment; men with high-amplitude left R waves (Minnesota code 3:1, 31, but no depressed ST-segment or T-wave abnormality or codable Q waves; and men with arrhythmias (Minnesota code 8; all items).
than among younger age groups. Also, among the aged, heart disease may be symptomless due to decreased physical activity. Thus, especially among the aged, ECG findings could potentially be used in clinical practice as indicators of symptomless heart disease. Among the elderly, major ECG findings associated with CHD/CVD have significantly predicted coronary events and total mortality.12-14~21 In contrast, minor ECG changes do not appear to be associated with increased risk of mortality.‘2~‘3 Atrial fibrillation, DISCUSSION first-degree atrioventricular block, 13V21 left bundle Among the elderly, prognostic significance of spe- branch block, 21 and left ventricular hypertrophy12 cific ECG abnormalities is mostly due to the under- have been found to be significant predictors of morlying heart disease.” However, among the elderly, a tality. Results of the present study are mostly in line reliable history of previous CHD is harder to obtain with previous studies. Especially when coincident,
duction defects and other arrhytmias did not show significant associations with the end points. For the ECG groupings, constructed for the present analysis, results were in accord with those reported above (Table II, Figure). Especially among men with ST-T changes, with or without concomitant Q waves, there was an increased risk compared with men with no codable abnormality. Estimates for relative risk did not change significantly when adjusted for CHD risk factors (data not shown).
644
June 1996
The American
Journal
of Medicinea
Volume
100
ECG AND MORTALITY IN THE ELDERLY/TERVAHAlJTA TABLE II Hierarchical Classification of Electrocardiogram (ECG) Abnormalities and Numbers of Studied End Points among Men Aged 65 to 84 Years Category of Men with ECG Fatal MI Any Ml Total Mortality Abnormalities* n n (%) n (%) n (%) 1 325 22 (7) 34 (10) 73 (22) 2 36 13 (36) 15 (42) 19 (53) 3 8 (13) 21 (341 4 E 2: Ki 28 (30) 55 (601 5 85 15 (18) 3 (41 9 (11) 6 39 2 (5) 4 (10) 6 (15) 7 58 3 (5) 3 (5) 18(31) Total 697 74 101 207 ’ 1. Men with none of the major abnormalities, as listed below, according to the Minnesota coding system. 2. Men with Q-waves (Minnesota code l:l-3) combined with either depressed ST-segment fhorizontal or downward sloping at least 0.5 mm (Minnesota code 4:1-311 or T-inversron (Minnesota code 5:1, 2). 3. Men with Q-waves, but no depressed ST-segment or T-inversion. 4. Men with depressed ST-segment (Minneosta code 4:1-31, but no codable Q-waves. 5. Men wrth T-abnormality (Minnesota code 5:1-3), but no codable Qwaves or depressed ST-segment. 6. Men with high amplitude left R-waves (Minnesota code 3:1, 31, but no depressed ST-segment, T-abnormality or codable Q-waves. 7. Men with arrhythmias (Minnesota code 8; all items). Ml = myocardial infarction.
ST-segment and T-wave abnormalities and Q waves suggestive of CHD, as well as high-amplitude R waves, were associated with significantly increased risk of both coronary events and mortality. Adjusting for history of previous CHD reduced the significance of Q waves, probably because of considerable colinearity, but not that of ST-segment and T-wave changes nor high R waves. Combination of ST-segment and T-wave abnormalities appeared to be highly predictive independently from Q waves. This seems to indicate, also among the elderly, the more acute nature of ST-T changes whereas Q waves may be evidence of previous damage in the heart. The study population was relatively small in number. Consequently, we were not able to study possible increases in risk associated with infrequent ECG changes. Adjusting the present analyses for classical coronary risk factors such as cholesterol, smoking, and blood pressure did not change the observed associations. These findings were in line with results from previous studies.g We conclude that ECG abnormalities suggestive of CHD are associated with high risk for coronary events and total mortality among elderly men. Among the elderly, a reliable history of previous CHD may not be obtained and heart disease symptoms may be atypical or missing due to decreased physical activity. Therefore, especially among the
ET AL
aged, ECG findings could be valuable in clinical practice as indicators of symptomless heart disease and in determining which patients should be referred for further noninvasive testing or cardiac catheterization. Additionally, ECG findings may be helpful when identifying patients at high risk for perioperative or postoperative cardiac problems involving noncardisc surgery.
REFERENCES 1. Gorman P, Calatayud J, Abraham S, Caceres CA. Effects of age and heart disease on the QRS axis during the seventh through the tenth decades. Am J Med. 1964;67:39-43. 2. Blackburn H, Vasquez C, Keys A. The aging electrocardiogram, a common agrng process or latent coronary artery disease? Am J Cardiol. 1967;20:618627. 3. Simonson E. The effect of age on the electrocardiogram. Am J Cardiol. 1972;29:64-73. 4. Mihalick M, Fisch C. Electrocardiographrc changes in the aged. Am Heart J. 1974;87:117-128. 5. Fisch C. Electrocardiogram in the aged: an independent marker of heart disease. Am J Med. 1981;70:4-6. 6. Bachman S, Sparrow D, Smith K. Effect of aging on the electrocardiogram. Am J Cardiol. 1981;48:513-515. 7. Fleg JL. Alternations in cardiovascular structure and function with advancing age. Am J Cardiol. 1986;57:33C-44C. 8. Rose G, Baxter PJ, Reid DD, McCartney P. Prevalence and prognosis of electrocardiographic findrngs in middleaged men. Br Heart J. 1978;40:636643. 9. Cedres BL, Liu K, Stamler J, et al. Independent contribution of electrocardiographic abnormalities to risk of death from coronary heart disease, cardio vascular diseases and all causes. Circulation. 1982;65:146-153. 10. Blackburn H, Taylor HL, Keys A. Coronary heart disease in seven countries XVI. The electrocardiogram in prediction of five year coronary heart disease incidence among men aged forty through fifty-nine. Circulation. 1970;4142fsuppl 1):1154-1161. 11. Cullen KJ, Stenhouse NS, Wearne KL, Cumpston GN. Electrocardiograms and thirteen year cardiovascular mortality in Busselton study Br Heart J. 1982;47:209-212. 12. Caird F, Campbell A, Jackson T. Significance of abnormalities of electrocardiogram in old people. Br Heart J. 1974;36:1012-1018. 13. Rajala S, Haavisto M, Kaltiala K, Mattila K. ECG findrngs and survival in very old people. Eur Heart J. 1985;6:247-252. 14. Dekker JM Electrocardiographic predictors of future coronary heart disease: a possible role of autonomic control. Wageningen, The Netherlands: Wageningen Agricultural University; 1994. 15. Karvonen M, Blomqvist G, Kallio V, et al. Men in rural East and West Finland. Acta Med &and. 1967;46OfsupplJ:169-190. 16. Keys A. Coronary heart disease in seven countries. Circulation. 1970;41(suppl lkl-211. 17. Rose GA, Blackburn H. Cardiovascular Survey Methods. Geneva: World Health Organrzation (World Health Organization Monograph Series No. 56); 1968. 18. FINMONICA. Monitoring of trends and determinants of cardiovascular disease in Finland (part of a joint study). The MONICA project. Helsinki: National Public Health Institute, 1986. 19. Tervahauta M, Pekkanen J, Enlund H, Nissinen A. Change in blood pressure and 5year risk of coronary heart disease among elderly men. The Finnish cohorts of the Seven Countries Study. J Hypertens. 1994;12:1183-1189. 20. SAS Institute. SAS User’s Guide: Statistics. Cary, North Carolina: SAS Institute; 1990. 21. Casrglia E, Spolaore P, Ginocchio G, et al. Mortality in relation to Minnesota code items in elderly subjects. Sex-related differences in a cardiovascular study In the elderly. Jpn Heart J. 1993;34:567-577.
June 1996
The
American Journal
of Medicine@
Volume
100
645