Sudden death risk in overt coronary heart disease: The Framingham Study

PROGRESS

IN CARDIOLOGY

Sudden death risk in overt coronary disease: The Framingham Study

heart

William B. Kannel, M.D., L. Adrienne Cupples, Ph.D., and Ralph B. D’Agostino, Ph.D. Boston, Mass.

Sudden death is a prominent and lethal feature of coronary heart disease (CHD), which claims more than 350,000 lives annually in the United States1 About half of the huge annual mortality rate attributed to CHD is in the form of sudden death. While many of the sudden deaths encountered in the general population occur unexpectedly, in the absence of clinically overt CHD, almost half of these catastrophic events occur in those who have overt CHD. It is well recognized that the sudden death risk in those who have CHD is many times that of the general population of the same age. This report examines the occurrence of sudden death in participants in the Framingham Study, who developed CHD while under observation. It is based on three decades of observation. It has the advantage of being population-based and includes all coronary events, such as unrecognized myocardial infarctions and newly diagnosed angina not under medical care. Because sudden death is such a common feature of coronary disease there is a need to obtain further insights into the way it evolves in the high-risk segment of the population with clinically overt CHD. Clues to pathogenesis may be provided, which may be of preventive importance. METHODOLOGY

The cohort under surveillance over the past 30 years is comprised of 5209 subjects, aged 30 to 62 years on entry. They are a representative sample of adults of the town, who have been followed biennially and examined for new development of clinical From the Section of Preventive Medicine and Epidemiology, Evans Department of Clinical Research, the School of Public Health, University Hospital, Boston University Medical Center, and the Department of Mathematics, Boston University College of Liberal Arm. Supported by contract Nos. NIH-NOl-HV-92922 and NIH-NOI-HV529’71. Received for publication Aug. 11,1986, accepted Sept. 16, 1966. Reprint requests: WiIlii B. Rannel, M.D., and Section of Preventive Medicine and Epidemiology, Boston University Medical Center, Doctor’s Office Building, Suite 1105, 726 Harrison Ave., Boston, MA 02118.

manifestations of CHD, including sudden death. The sampling procedure, response rates, completeness of follow-up, and diagnostic criteria have been reported in detail e1sewhere.2*3 New coronary events documented included angina, coronary insufhciency syndrome, myocardial infarction, and coronary fatalities including sudden death.2 The circumstances surrounding each death were carefully ascertained with the use of information from the medical examiner’s reports, hospital records, death certificates, and interviews with decedent’s spouses and physicians. The assignment of sudden death to coronary disease is largely by inference, in as much as few other diseases can cause death in a matter of minutes. When aortic dissection, ruptured aneurysm, and pulmonary embolus can be excluded, an etiologic relationship to coronary disease is usually warranted. Death within minutes in persons not ill with any potentially lethal illness permits a reasonably certain designation of sudden coronary death. This was the definition adopted in the Framingham Study rather than the less specific 24-hour deflnition advocated by the World Health Organization.’ Subjects were placed at risk of sudden death by age and sex at the time of the first manifestation of CHD and were compared to those who, on biennial examination, remained free of the event. Comparisons were made, adjusting for age, by the direct method. The relation of risk factors to the occurrence of sudden death was examined by placing surviving subjects at risk at the time of the appearance of any manifestation of CHD and was compared with those of the same age remaining free of 0. Regression coefficients, standardized to place them on an equal footing for the different units of measurement, were estimated by an iterative maximum likelihood method.’ These regression coefllcients, calculated by this logistic regression model, measure the strength of the relationship of the specified risk factors to occurrence of sudden death. The model 799

March

800

Kannel,

Cupples,

and D’Agostino

1

American

Ail Sudden Death

Heart

1987

Journal

Sudden Unex ected Death PI) Men O--QWomen

, 45-5455-6465-7475-84

45-5455-6465-7475-84

Age (1)

Fig.

Persons

Free

Age

of CHD

1. Incidence of sudden death by age and sex: 30-year follow-up, Framingham

expresses the probability of an event by: 1/ (1 + exp - [(a + bi xi)], where xi are the risk factors of interest and bi the coefficients. OBSERVATIONS

At time of the initial examinations of the 5209 subjecta, aged 30 to 62 years, 5127 were found to be free of overt CHD. Over the 30 years of follow-up, there were 760 men and 574 women who developed clinically overt CHD. There were 1019 deaths among the men and 857 deaths among the women (Table I). Among those who died, 615 men and 379 women died of cardiovascular disease, and 350 men and 196 women died of CHD. During the 30 years, there were 160 Budden deaths among the men and 73 among the women. The incidence of sudden death in those without prior CHD doubled with each decade beyond age 45 years (Fig. 1). Women lagged behind men in incidence of sudden death by about 20 years. In contrast to those without interim overt CHD, there is no age trend in incidence of sudden death, the CHD event putting young and old men in equal jeopardy (Fig. 2). In men, but not women, the occurrence of CHD eliminates the advantage of young age. The relative risk of sudden death, compared to that in subjects without CHD, diminishes with advancing age (Fig. 2; Table II), Among persons who sustined a clinical manifestation of CHD, the risk was 3 to 12 times that of the general population of the same age. In men, the risk was an average 6.7 times that of persons without an interim CHD event. Having CHD does not eliminate the female advantage over men in susceptibility to sudden death (Table II). However, the relative risk, compared to that in those

Study.

1. Number of cardiovascular or overall deaths: 30-year folIow-up, Framingham Study

Table

Age 35-64 yr Cause of death

Men

Sudden deaths CHD deaths Cardiovascular deaths Overall deaths

90 175 240 452

Women 20 54 106 314

Age 65-94 yr Men 70 175 275 567

Women 53 142 273 543

without CHD, was similar in the two sexes. Among those with a myocardial infarction, the sudden death risk was greater than that of those who had angina pectoris. However, even angina carries almost a five-fold increased risk. Among those who developed a myocardial infarction, 28% in men and 35% in women went unrecognixed. For those who experienced silent or unrecognized myocardial infarctions sudden death risk was just as great as for those sustaining a symptomatic recognized myocardial infarction (Table III). The fraction of all deaths classified as sudden deaths was also similar. Some 40% of sudden deaths occurred in the 4 % of the general population with overt CHD (Table Iv). A larger percentage of the sudden deaths in men (45%), compared to those in women (33%), occurred with prior coronary disease. The proportion of coronary attacks presenting as sudden death increased from 13% at ages 35 to 64 years to 20% at ages 65 to 94 years (Table V). The fraction of CHD deaths that were sudden deaths varied from 28% to 51% depending on sex

Volume Number

113 3

Sudden

death

cl No

in coronary

disease

801

CAD

n Prior CAD

45-54

65-74

55-64

75-84

Age at Biennlal Examination Fig. 2. Risk of sudden death by coronary disease status: 30-year follow-up, Study. Table II. Risk of sudden death by CHD status: 30-year follow-up,

Framingham CHD

Age

grout

Men present

Absent

(~4

45-54 55-64 65-74 75-84 All ages adjusted

2.3 4.2 6.6 10.8 4.4

Risk

29.6 31.4 28.0 28.6 29.5

12.2 1.5 4.2 2.6 6.7

Table III. Risk of sudden death in recognized vs unrecognized myocardial infarction: 30-year follow-up, Framingham Study, subjects 33 to 87 years of age IO-year

Myocardial infarction

Unrecognized Recognized

20 22

Men

10 12

status

Sudden deaths Women

5 5

There are few modifiable

predisposing risk factors for sudden death once clinically manifest CHD appears. In those without interim CHD, virtuahy alI of the major risk factors were related to incidence of sudden death. After the

Women present

Absent

Risk

5.9 14.6 17.8

ratio

4.9 6.6 2.3

Table IV. Proportion of sudden deaths with and without prior overt CHD: 30-year follow-up, Framingham Study, subjects35 to 94 years of age prior

and interim CHD status (Table VI). Curiously, in those with prior CHD the fraction of CHD deaths presenting as sudden deaths appeared to be lower than in those without interim CHD. Predisposing factors.

Study

1.2 2.2 1.1

age-adjusted mortality rate/1000 at risk

All CHD deaths Men Women

26 27

ratio

men in the Framingham

Subjects

Men Women Total

No.

72 24 96

With CHD

Without prior CHD %

No.

%

45 33 41

88 49 137

55 61 59

Total No.

160 73 233

onset of overt CHD, none of the major modifiable CHD risk factors were predictive of sudden death in men (Table VII). In women, only diabetes was a significant predisposing factor, although the coefficient for cigarette smoking was sizable. In persons with already established CHD, factors which reflect ischemic myocardial damage were the chief predictors of sudden death. ECG abnormalities indicating myocardial infarction, left ventricular

802

Kannel,

Cupples,

Table V. Percentage Study

and D’Agostino

of coronary

attacks

American

presenting

as sudden death by age and sex: 30-year follow-up,

35-64yr

Coronary attacks* No. of sudden deaths % of sudden deaths *CHD

events

other

than

angina:

65-94yr Men

Women

353 51 14

142 18 13

186 37 20

166 31

539 88

308

19

16

16

at risk-persons

Age-adjusted

free

status

Absent Present Differences

not statistically

deaths by Framing-

Women

49

Table VIII. ECG contributors to sudden death in subjects with overt CHD: 30-year follow-up, Framingham Study Standardized logistic regression coefficients

percentage

Men

Women

51.2 41.8

41.7 28.2

Men

Women

ECG-MI ECG-LVH

0.391* 0.336*

0.355t 0.153

IV block NSA-ECG

0.180$

0.294t

0.265t

0.261

abnormality

significant.

Table VII. Risk factors for sudden death in subjects with CHD: 30-year follow-up, Framingham Study, subjects 35 to 94 years of age Standardized logistic coefficients Risk factors

Men 0.007

Age

Systolic blood pressure Cholesterol Cigarettes Glucose intolerance ECG-LVH

-0.014

-0.165 0.215

0.201t

-0.027 0.2253

0.224* hypertrophy; $.n < 0.001;

0.592* 0.104

0.084

VPC

= ventricular

0.3257 0.108

premature

Concurrent covariates: age, systolic blood pressure, cholesterol, cigarettes, glucose, atrioventricular block, heart rate, cardiac failure. Abbreviations: MI = myocardial infarction; LVH = left ventricular hypertrophy; IV = intraventricular; NSA = nonspecific abnormality. *p < 0.001; tp < 0.05; $0.05 < p < 0.10.

onset of CHD in particular (Table IX). Most sudden deaths (75% ) associated with cardiac failure had coexistent CHD. Having both cardiac failure and CHD increased risk of sudden death more than having either alone. Although the operative risk factors are different, multivariate risk profiles can be devised which predict sudden death almost as efficiently in those who have overt CHD as in those without CHD (Table X).

Women

-0.002 -0.024 0.376$

VPC Cardiac failure LVH = left ventricular tractions. *p < 0.01; tp < 0.05;

Men

of CHD.

EGG CHD

All ages

Women

population

Journal

Framingham

Men

Table VI. Proportion of CHD deaths as sudden prior coronary disease status: 30-year follow-up, ham Study

Prior

Heart

con-

sp < 0.08.

hypertrophy, conduction disturbance, or repolarization abnormality were significant sudden death predictors in one or both sexes (Table VIII). Ventricular ectopy was a risk factor for sudden death in men, associi&d with a doubled incidence of sudden death (Table VII). An irritable myocardium is often a reflection of a poor ejection fraction. The strong in%uence of myocarcbal damage was further mdicated by the marked excess incidence of sudden death in cardiac failure in general and that following

DlSCUS!3lON

Since such a high proportion of sudden deaths arise from the small segment of the population with overt CHD, there is a clear need to focus preventive measures on those who have the disease. The risk is sub&&al and justifies vigorous preventive meamires. Examin ation of secular trends in the proportion of CHD deaths which were sudden deaths in the Framingham cohort does not indicate much progress over the past three decades. The data presented indicate that the preventive measures required for those who already have overt CHD are those which minimize myocardial damage during an attack and protect the injured, irritable

Volume Number

113 3

Sudden

death in coronary

Table IX. Risk of sudden death by CHD and cardiac failure status: 30-year follow-up, Framingham to 94 years of age Biennial

age-adjusted

status

Neither CHD alone CHF alone Both % of CHF sudden deaths with CHD CHF = congestive Trends significant

No. of events

3.6 25.1 24.7 34.2

73 5

18

Study, subjects 35

Women Rate/1000

87

803

sudden deaths

Men Coronary failure

disease

No. of events

Rate/1000

47 19 3 8

18.3

1.5 4.6 6.1 13.9 72.1

heart failure. at p < 0.001.

myocardium. Prevention of coronary attacks in general, through correction or avoidance of the standard risk factors, would, however, appear to be the most effective approach to the problem. There is some indication that the high risk of sudden death in persons who have already sustained a myocardial infarction or developed angina can be minimized. Those who stop smoking have been found to have only half the coronary mortality risk of those who continue to smoke.5 Use of beta blockers appears to be useful in reducing sudden death rates following a myocardial infarction. 6-gAttempt to control complex and frequent ventricular premature beats by drugs has so far been disappointing, possibly because it does not address the more fundamental underlying problem of the damaged, ischemic, irritable myocardium. It is possible that calcium channel blockers may be useful in prevention of sudden death. Subgroups of patients who are convalescing from a myocardial infarction, who are at high risk of death within the first year, can be identified based on clinical findings, exercise ECG testing, and ambulatory monitoring. lo*l1 For these high-risk candidates, beta blockers, calcium antagonists, platelet inhibitors, antiarrythmic agents, and anticoagulants may be considered.12p l3 Attention to the standard CHD risk factors is of greater efficacy once the early high mortality period following myocardial infarction is past. Long-term preventive management designed to retard further progression of the underlying atherosclerosis would appear to be rational in improving the outlook following onset of CHD, despite the lack of controlled trial evidence of efficacy. The surest way to prevent sudden deaths would appear to be to prevent coronary attacks and damage from myocardial ischemia.

X. Predictive value of risk profiles in subjects with and without CHD

Table

Proportion of all deaths in top quintile multivariate risk (X) CHD

Men

Women

Absent Present

56 47

64 59

Risk variables: Age, systolic blood pressure, cigarettes, cholesterol, glucose, congestive heart failure (ever), ECG-myocardial infarction, ECG-left ventricular hypertrophy, intraventricular block, nonspecific ECG abnormality, Heart rate.

SUMMARY

Over 30 years of surveillance of 5127 subjects free of CHD, 760 men and 574 women developed overt CHD, and there were 160 sudden deaths in men and 73 in women. Among those who sustained clinically manifest CHD, the sudden death risk was increased 6.7 times that of those without an interim event. Although the reZotiue risk was comparable in the two sexes, CHD did not eliminate the.female advantage over men. Myocardial infarction imposed a greater sudden death risk than angina pectorjs, and silent infarctions were as dangerous as symptomatic infarctions. Onset of CHD put young and old at equal risk of sudden death. Some 40 % of sudden deaths occurred in the 4 % of the general population with overt CHD. The proportion of coronary attacks presenting as sudden death increased from 13 % at ages 35 to 64 years to 20 % at ages 65 to 94 years. The fraction of CHD deaths classified as sudden deaths was lower in those with than without interim CHD. In those with established CHD, factors reflecting ischemic myocardial

March

Kannel, Cupples, and D’Agostino

American

damage and cardiac failure were the chief predictors of sudden death. The proportion of CHD deaths presenting as sudden deaths has not declined in subjects with prior CHD over three decades, despite a national decline in the overall CHD mortality rate. REFERENCES

1.

2.

3. 4. 5.

Kannel WB, McGee DL, Schatzkin A. An epidemiological perspective of sudden death. 26-year follow-up in The Framingham study. Drugs 1984;28(suppl l):l-16. Shurtleff D. Some characteristics related to the incidence of cardiovascular disease and death: The Framingham study, lb-year follow-up. 1974; DHEW Publication No. (NIH) 74599. Dawber TR, Meadors GF, Moore FE Jr. Epidemiological approaches to heart disease: The Framingham study. Am J Public Health 1951;41:279-286. Walker S, Duncan D. Estimation of the probability of an event as a function of several independent variables. Biometrika 1967;54:167-179. Kannel WB. Update on the role of cigarette smoking in coronary heart disease. AM HEART J 1981;101:319-328.

F. Gilhnn,

M.D. Hyattsuille,

6. Wilhelmsson C, Vedin JA, Wilhelursen L, Tibblin G, Werko L. Reduction of sudden deaths after myocardial infarction by treatment with alprenolol. Lancet 1974;2:1157. 7. Furberg CD, Friedwald WT. The effects of chronic administration of beta-blockade on long-term survival following myocardial infarction. Excerpta Medica: 1978. 8. Multicenter study. Improvement in prognosis of myocardial infarction by long-term B-adrenoceptor blockade using practolol. Br Med J 1975;3:735. 9. Beta-Blocker Heart Attack Trial Research Group. A randomized trial of propronalol in patients with acute myocardial infarction. I. Mortality results. JAMA 1982;247:1707. 10. Bigger JT, Heller CA, Wenger TL, et al. Risk stratification after acute myocardial infarction. Am J Cardiol 1978;

42~202.

11. Kennedy HL. Comparison of ambulatory electrocardiography and exercise testing. Am J Cardiol 1981;47:1359. 12. Frommer PL, Furberg C. Beta-blocking drugs in the prevention of sudden cardiac death. In Morganroth J, Horowitz LN, eds.: Sudden cardiac death 1985. Grune & Stratton, Inc: New York, 1985:249-256. 13. Friedman L. Platelet active agents in the prevention of sudden death. In Morganroth J, Horowitz LN, eds.: Sudden cardiac death 1985. Grune & Stratton, Inc: New York, 1982257-265.

Md.

Death rates for ischemic heart disease have declined in the United States since the late 1960’~~ (Figs. 1 and 2). Several studies have attributed this to declining sudden coronary heart disease death and declinii incidence and case fatality of acute myocardial infarction.2 In several cities, hospitalization rate, used as an indicator of acute myocardial infarction occurrence, declined during the 1970% for at least some age-sex groups. III this report, data from the National Hospital Discharge Survey for a 14year period are presented. These data show no consistent downward trend in hospital discharge

From the Office of Analysis and Epidemiology Program, National Center for Health Statistics. Received for publication July 10, 1986; accepted Aug. 11, 1986. Reprint requests: Richard F. Gillurn, M.D., Office of Analysis and Epidemiology Program, National Center for Health Statistics, Center Bldg., Room 2-27, 3700 East-West Highway, Hyattsviile, MD 20782.

804

1967 Journal

kl

Acute my states, 19704983 Richard

Heart

rates for acute myocardial infarction, but a decline in hospital case fatality rates for the United States between 1970 and 1983. In addition, data by age, sex, race, and region are presented and methodologic problems of national and local surveillance for coronary heart disease trends are discussed. METHODOLOGY The Nattonal Hospital Discharge Survey. As previously described,” the National Hospital Discharge Survey encompasses patients discharged from noninstitutional hospitals exclusive of Federal hospitals located in the SO States and the District of Columbia. Included in the list of eligible hospitals are those licensed hospitals with at least six beds where the average length of stay for all patients is less than 30 days. Hospital discharges are selected for abstracting by a multi&aged sampling procedure, described in detail elsewhere.3+4 From a universe of over 8,000