Epidemiology of heart failure

William B. Kannel, MD, and Albert J. Belanger. Boston, Muss.

Analysis of 34 years of follow-up of Framingham Study data provides clinically relevant insights
into the prevalence, incidence, secular trends, prognosis, and modifiable risk factors for the
occurrence of heart failure in a general population sample. Heart failure was found to be highly
prevalent, affecting about 1% of persons in their 50s and rising progressively with age to afflict
10% of persons in their 80s. The annual incidence also increased with age, from about 0.2% in
persons 45 to 54 years, to 4.0% in men 85 to 94 years, with the incidence approximately
doubling with each decade of age. Women lagged slightly behind men in incidence at all ages.
Male predominance was because of a higher rate of coronary heart disease, which confers a
fourfold increased risk of heart failure. Heart failure, once manifest, was highly lethal, with 37%
of men and 33% of women dying within 2 years of diagnosis. The g-year mortality rate was 82%
for men and 67% for women, which corresponded to a death rate fourfold to eightfold greater
than that of the general population of the same age. Sudden death was a common mode of exitus
and accounted for 28% of the cardiovascular deaths in men and 14% in women with heart failure.
Hypertension and coronary disease were the predominant causes for heart failure and accounted
for more than 80% of all clinical events. Factors reflecting deteriorating cardiac function were
associated with a substantial increase in risk of overt heart failure. These include low vital
capacity, sinus tachycardia, and ECG evidence of left ventricular hypertrophy. Modifiable
predisposing risk factors for heart failure include hypertension, impaired glucose tolerance, an
elevated total to high-density lipoprotein cholesterol ratio, obesity, and cigarette smoking. In
subjects with coronary disease risk increases progressively from angina to recognized
myocardial infarctions to unrecognized infarctions in men. In women angina also carried half the
failure risk of a myocardial infarction, and in both sexes unrecognized infarctions were at least as
dangerous as symptomatic ones. Using simple office procedures and laboratory tests, it is
possible to identify high-risk candidates for heart failure early in its course for preventive
management before irreversible myocardial damage has occurred. (AM HEART J 1991;121:951.)

In spite of major reductions in deaths from cardio- first onset of heart failure in relation to predisposing
vascular disease in the United States during the past risk factors and noninvasive clinical indicators of
three decades, it continues to be life threatening. impaired myocardial function. Details of the meth-
Hypertension, coronary disease, and damage to the odology, sampling, follow-up criteria for heart fail-
cardiac valves continue to produce heart failure. As ure, risk factor measurements, and statistical meth-
people are saved from premature deaths from these ods used have been reported in detail elsewhere.2s 3
conditions, more heart failure will occur. Because
INCIDENCE AND PREVALENCE
heart failure is an end-stage of heart disease that en-
sues after the myocardium has exhausted all its The National Heart, Lung and Blood Institute es-
reserve capacity and compensatory mechanisms, the timates that more than 2 million Americans are
possibilities for salvage are 1imited.l A preventative afflicted with heart failure and that 400,000 new
approach that corrects modifiable predisposing fac- events occur yearly, requiring 900,000 hospitaliza-
tors and early functional disturbance holds more tions each year.4 Framingham Study estimates of the
promise. This article provides some insights into the prevalence indicate that the occurrence of heart fail-
evolution of heart failure based on 34 years of follow- ure increases progressively with age from about 1%
up of the Framingham cohort for the development of prevalence in those aged 50 to 59 years to a preva-
lence of about 10% in persons 80 to 89 years. The
prevalence approximately doubles with each decade
From the Boston University School of Medicine.
within that age range (Fig. 1). The prevalence is
Reprint requests: William B. Kannel, MD, Professor of Medicine and Pub-
lic Health, Baton University School of Medicine, RU-Framingham Study.
higher in hypertensive persons in each age group. The
5 Thurber St.. Framingham. MA 01701. incidence of new onset of heart failure also increases
4/O/26029 dramatically with advancing age. During 34 years of

951
 March 1991
952 Kannel and Belanger American Heart Journal

Age: 50-59 60-69 70-79 80-89

Person-bienniums 20520 19298 8994 2084

Person-bienniums with CHF 166 451 438 190
Percent prevalence 0.8 2.3 4.9 9.1

Percent 6
prevalence
4
I T 1
21 ‘I 1

0 I , / I,ll,l
50-59 60-69 70-79 80-89

Fig. 1. Prevalence of heart failure by age. Thirty-four-year follow-up experience from the Framingham
Study: men and women.

85 QWomen
80 /
Annual rate per 1,000
75 :
70
Age Men Women ,:i
65
60
55 55-64
45-54 24 13 / fMen
50
Rate per 45 65-74 8 5 ‘/
1,000 40
35 85-94
75-84 54
14 85
13 i / ;/
30
25 35-64 3
20 65-94 10
15
10

i :
45-54 55-64 65-74 75-84 85-94
Fig. 2. Incidence of heart failure by age and sex: 30-year follow-up from the Framingham Study.

follow-up, 289 men and 289 women developed heart
failure. As for prevalence, the incidence approxi-
mately doubled with each decade of age. The inci-
dence in men exceeded that in women at virtually all
ages. This was very likely because of a higher inci-
dence of coronary heart disease in men than in
women (Fig. 2). Overt heart failure was diagnosed in
the Framingham cohort based on a number of major
and minor clinical factors.2, 3 These resulted in crite-
ria that were rather severe and did not include per-
sons with impaired subclinical cardiac function now
detectable by noninvasive modern technology. Thus
the incidence rates are probably understated. The
rate of increment increases sharply beyond the age of
75 years.
ETIOLOGY
The preponderant causes of heart failure during
more than three decades of follow-up were long-
standing hypertension and coronary heart disease.
Some 76 % of men and 79 % of women had hyperten-
sion or received antihypertensive treatment. About
46 % of men and 27 % of women had a background of
coronary heart disease. Rheumatic heart disease was
Volume 121
Number 3, Part 1 Epidemiology of heart failure 953

Heart Rate
25 -

20 -
0 r85fmin

Age-
adjusted Men Women
biennial 15 -
rate per
1000
10

’
II-l-i
Normal Mild Definite
Hypertension
Fig. 3. Risk of heart failure by hypertensive
Normi 31
Mild Definil
Hypertension
and heart rate status: 34-year follow-up from Framingham
Study. Subjects 35 to 94 years. -

considered a cause in only 2% to 3%) and other Table I. Prevalence of causes of heart failure: 3%year fol-
causes were impugned in 11% to 17 % (Table I). low-up

INDICATORS OF IMPAIRED FUNCTION % of CHF

There are several intrinsic measurable factors, Men Women

which in the asymptomatic state reflect deteriorating
Hypertension* 76.4 79.1
myocardial function. These include an enlarged heart, Coronary diseaset 45.8 27.4
ECG abnormalities, a poor vital capacity, and a rapid Rheumatic HD 2.4 3.2
heart rate. Each was associated with an increased risk Other 1 11.2 16.8
of developing heart failure, and when these occurred CHF, Congestive heart failure; HD, heart disease.
jointly, the incidence of heart failure was quite sub- *Blood pressure > 160/95 mm Hg or on antihypertensive medication.
stantial. The risk of heart failure tended to increase tAny clinical manifestation of coronary heart disease.
tNone of the above and cardiomyopathy.
progressively with heart rate throughout the range
observed in both sexes, more prominently in men
than in women. Above a rate of 85 beats per minute,
the risk of failure in men was almost doubled com- pathogenesis of left ventricular hypertrophy in the
pared with that of persons with lower heart rates at evolution of heart failure is an important consider-
all blood pressure levels (Fig. 3). Before the occur- ation in devising means for prevention and effective
rence of symptoms, a low or falling vital capacity, treatment of the condition. ECG-LVH heralded the
which probably reflected pulmonary vascular en- occurrence of serious cardiovascular disease of all
gorgement resulting from malfunction of the left varieties, with risk ratios greater for heart failure
ventricle, was associated with a risk of heart failure than any other cardiovascular sequelae (Table III).
in both smokers and nonsmokers (Table II). ECG-LVH contributed more powerfully to the oc-
Each of the commonly encountered ECG abnor- currence of heart failure than did blocked intraven-
malities-left ventricular hypertrophy (LVH), intra- tricular conduction or nonspecific repolarization ab-
ventricular conduction disturbance, and nonspecific normalities. ECG-LVH characterized by increased
repolarization abnormality-is associated with sub- voltage unaccompanied by repolarization abnormal-
stantial increased risk of developing heart failure. ity carried a lesser risk of failure than that associated
This is true even when hypertension and other risk with such evidence of “strain.” In multivariate anal-
factors of cardiovascular disease are taken into ac- ysis ECG-LVH exhibits an independent effect in
count. Cardiac hypertrophy is a prominent feature of both young and old persons. In both mild and severe
the evolution of heart failure. Understanding of the hypertension, the most common cause of cardiac hy-
Table II. Risk of congestiveheart failure by vital capacity dial damage and anatomic hypertrophy. Whf-c-1both
and cigarette smokingstatus; 34-year follow-up of subjects, are present, the risk is substantially higher that) when
35 to 94 years either is present alone (Table V). Echocardiographlc
Age-adjusted biennial rate per 1000 evidence of LVH was more common than the ECG
for cigarette smohers variety in subjects with heart failure and occurred tn
63 % of women and 77% of men with ;I failed
Vital Men Women
capacity myocardium. In persons with heart failure, I, VH was
tertiks No Yes No Yes the most common echocardiographic abnormality
encountered. The prognostic significance of echocar-
4-381 26.5 14.7 9.3 12.0
diographic hypertrophy in comparison with the ECG
382-483 10.3 13.6 4.9 2.8
484-918 5.3 6.7 0.8 1.7
variety remains to be determined.
Although risk of heart failure associated with
ECG-LVH is independent of associated blood pres-
sure, the risk among persons with ECG-LVH is
Table Ill. Prognostic impact of ECG-LVH for cardiovascu- markedly influenced by the accompanying blood
lar outcomes: 30-year follow-up from the Framingham
pressure level. This suggests that the risk of heart
Study
failure associated with ECG-LVH may be reduced by
Risk ratios * correction of hypertension and other risk factors for
heart failure. There has been a substantial decline in
Men Women
the prevalence of ECG-LVH during the several
Cardiounscular 35-64 65-94 3.5-64 65-94 decades of the Framingham Study, which indicates
sequelae Yr yr y r yr that this is a preventable, if not a correctable, abnor-
mality. Clearly, ECG-LVH must be regarded as a
Heart failure 17.5 6.4 17.0 6.7
Brain infarction 6.0 3.2 7.0 6.5
grave prognostic sign for impending heart failure in
Peripheral arterial disease 2.7 1.9 3.5 2.7 the course of hypertension and coronary heart dis-
Coronary disease 3.0 3.0 5.8 3.7 ease.
*Definite ECG-LVH vs none; age-adjusted; all risk ratios significant at
MODIFIABLE PREDISPOSING RISK FACTORS
p < 0.001.
The predisposing correctable risk factors for heart
failure include hypertension, diabetes, cigarette
Table IV. Risk of congestiveheart failure by ECG-LVH and smoking, obesity, a poor total to high-density li-
hypertensive status; 34-year follow-up of subjects,35 to 94 poprotein (HDL) cholesterol ratio, and hematocrit
years level (Table VI). Hypertension is a major contributor
Age-adjusted biennial rate per 1000 to the incidence of heart failure in the general pop-
for hypertensive status ulation because of its high prevalence and the fact
that it increases risk by about threefold. The risk of
Men Women
heart failure increases progressively with the severity
ECG-LVH* Normal BHBP HBP Normal BHBP HBP of the hypertension. This risk is as strongly related to
systolic as diastolic pressure and is increased in per-
None 5.2 8.7 15.9 4.2 6.2 9.1 sons with isolated systolic hypertension. In fact, in
Possible 0.0 11.9 22.6 13.5 23.9 23.1
Definite 96.4 58.7 83.7 73.7 107.3 42.6
persons whose diastolic pressures had never ex-
ceeded 90 mm Hg over 34 years of follow-up in the
BHBP, Borderline high blood pressure (140-159/90-94 mm Hg); HBP, high Framingham cohort, the risk increased progressively
blood pressure (z 160/> 95 mm Hg).
*AI1 ECG-LVH trends significant at p < 0.001. with systolic blood pressure at all ages (Fig. 4). Com-
paring the impact of the various components of blood
pressure using a ratio of the fifth to the first quintile,
pertrophy, the risk of heart failure was greatly to place them on an equal footing for the different
increased when ECG-LVH appeared, particularly range of values suggests that systolic pressure is the
when associated with repolarization abnormality and best single predictor in women, whereas in men the
voltage increase (Table IV). X-ray film evidence of pulse pressure rivals systolic pressure as the best
anatomic hypertrophy carried a lesser risk than did predictor of heart failure (Table VII).
the ECG variety in both sexes at all ages. Because Diabetes predisposes patients to heart failure be-
each independently contributes to risk, it is likely cause of its association with accelerated coronary
that the ECG version reflects both ischemic myocar- atherosclerosis, hypertension, and obesity. In addi-
“Ohme 121
Number 3, Part 1 Epidemiology of heart failure 955

tion, it appears to damage the myocardium directly. Table V. Risk of heart failure by ECG and X-ray film-LVH:
Diabetes increases the risk of heart failure from two- 32-year follow-up from the Framingham Study
fold to sevenfold, with a greater impact in women Age-adjusted biennial rate per 1000
than in men (Table VIII). Epidemiologic data do not
35-64 yr 65-9-J yr
support the contention that heart failure resulting
X-ray-LVH X-ray-L VH
from diabetes is solely a consequence of associated
coronary disease. Functional, structural, and meta- ECG-LVH Absent Present Absent Present
bolic aberrations have been demonstrated in the di-
abetic heart, which suggests that some form of Men
Absent 3 16 14 56
diabetic cardiomyopathy exists. Diabetes was ob-
Present 71 128 102 135
served to predispose to heart failure whether or not Women
there is interim overt coronary heart disease and does Absent 2 17 12 21
so independent of the often associated hypertension. Present 16 83 89 85
Hematocrit level is a significant factor in women at
all ages, with both low and high values associated
with an increased risk of failure. Cigarette smoking
Table VI. Impact of modifiable risk factors on heart failure
increases moderately the risk in younger men and
incidence: 34-year follow-up from the Framingham Study
older women. Despite a strong association of the se- subjects,35 to 94 years old
rum total cholesterol with the occurrence of coronary
heart disease, cholesterol value was significantly re- Age-adjusted Qs/Ql risk ratio
lated to the incidence of heart failure only in men Risk factors Men Women
under the age of 65 years. The total to HDL choles-
terol ratio, however, is powerfully related to the rate Systolic blood pressure 2.3s 3.0s
of occurrence of heart failure in both sexes (Table Diastolic blood pressure 1.5$ 1.61
Blood glucose 2.2s 1.7s
IX). Obesity is a significant contributor to heart fail-
Serum cholesterol 1.3* o.l3*
ure, particularly in women. Cigarettes 1.0* 1.3*
Relative weight 1.3* 1.7t
PROGNOSIS
*Not significant.
If there is no correctable mechanical problem such tp < 0.05.
as a valvular deformity, heart failure, once overtly lp < 0.01.
manifested, is an extremely lethal condition. Within §p < 0.001.

2 years of its diagnosis in the Framingham cohort,

37 % of men and 38 % of women died. During 6 years
of follow-up, the mortality rate was a depressing 82 %
for men and 67% for women. This mortality rate is
four to eight times that of the general population of
the same age. Sudden death is a common feature of
cardiac mortality in failure.5 Within 6 years of onset,
25 % of men and 13 % of women in the Framingham
Study experienced sudden death. Sudden deaths oc-
curred at five times the rate for the general popula-
tion, even excluding those persons with overt coro-
nary heart disease. The presence of both coronary
disease and heart failure imposes a further augmen-
tation of sudden death risk. The high rates of sudden
death associated with heart failure implicate myo-
cardial malfunctioning per se as the cause, because
that risk persists on adjustment for hypertension as-
sociated coronary diseases and other risk factors.
In addition to the tragic mortality that heart fail-
ure imposes, considerable morbidity and recurrences
are common. Within 6 years 63 % and 70 % of women
and men, respectively, had one or more recurrences
of overt heart failure after its initial diagnosis. The
presence of heart failure increases the risk of a stroke
fourfold, rivalling only atria1 fibrillation as a risk fac-
tor for stroke. Coronary attacks occur at several times
the rate for the general population, probably because
of the presence of coronary disease underlying the
existing failure. An examination of the secular trends
in heart failure in the 195Os, 196Os, and 1970s in the
Framingham Study gives no clear indication of a
downward trend in any age group in spite of an
increasing institution of vigorous antihypertensive
therapy over these decades. This may be attributable
to a threefold increase in diabetes and the salvage of
many coronary cases by heart surgery and improved
medical therapy so that they eventually culminate in
heart failure.
MULTIVARIATE RISKS
Risks of developing heart failure can be estimated
from a multivariate risk profile using the standard
cardiovascular risk factors of systolic blood pressure,
lipid levels, glucose levels, cigarette smoking, and
 March 1991
956 Kannel and Belanger American Heart Journal

40
1 SYSTOLIC BP: n cl40
MEN
30
Age- /
adjusted
biennial 20
rate per
1000

65-94
AGE
Fia. Risk of heart failure by systolic blood pressure
4. at systolic pressure < 90 mm Hg: 34-year follow-up
from the Framingham Study. -

Table VII. Risk of heart failure by blood pressure param- Table IX. Risk of congestive heart failure by cholesterol
eters: 34-year follow-up from the Framingham Study profile on examinations 10 to 12: 34-year follow-up, sub-
jects 49 to 81 years old
Qs/Ql risk ratio
Age-adjusted biennial rate per 1000
Men Women TotallHDL
cholesterol ratio Men Women
Components of 35-64 65-94 35-64 65-94
blood pressure yr yr yr yr < 5.0 8.4 6.5
5-9.9 12.5 9.4
Systolic blood pressure 3.11 1.8$ 4.41 2.41 210 41.6 52.9
Diastolic blood pressure 1.4* 1.6f 2.9t 1.2*
Mean arterial blood pressure 2.61 2.1% 3.91: 1.6t Trend significant at p < 0.01.
Pulse pressure 3.4% 2.3% 3.01 2.2%

*p < 0.05. upper quintile of multivariate risks. These risk pro-

tp < 0.01.
lp < 0.001. files distribute the risks over a wide range and segre-
gate those at lower risk from high risk quite well. In
persons with either hypertension or coronary disease,
Table VIII. Risk of congestive heart failure by glucose risk of heart failure varies widely depending on the
intolerance status: 34-year follow-up from the Framing- level of these other risk factors. These risk factors al-
ham Study low estimates of the risk of heart failure using only
Age-adjusted biennial ordinary office procedures.
rate per 1000
PREVENTATIVE IMPLICATIONS
Men Women Chronic heart failure appears to be an irremedial
Glucose
intolerance 35-64 65-94 35-64 65-94 terminal event, and it appears unlikely that pharma-
cologic therapy can significantly increase longevity in
Absent 5.1 19.1 2.7 14.5 those who have overt heart fai1ure.l The inability of
Present 14.6* 38.5* 15.1* 42.0*
the myocardium to pump sufficient blood to meet
*p < 0.001. metabolic demands of the tissues is often aggravated
by reflex increases in peripheral vasoconstriction,
which is induced by the renin-angiotensin-aldoster-
ECG-LVH.6 A more specific profile for heart failure one system and the sympathoadrenal axis. The best
should include vital capacity, heart rate, and heart that therapy can do appears to be to improve the
enlargement. Surprisingly, this results in only a contractile state of the heart and reduce arterial and
modest improvement in predicting failure over that venous constriction to improve symptoms and the
using the standard cardiovascular risk factors for quality of life. Inotropic drugs increase contractility,
coronary heart disease. Using such profiles, about and drugs are available that act directly on arteries
53% of future heart failure can be detected in the and veins to improve cardiac function by reducing
Volume 121
Number 3, Part 1 Epidemiology of heart failure 957

both afterload and preload. Angiotensin-converting dial damage by instituting effective therapy with
enzyme inhibitors can be used to suppress angioten- agents that do not potentiate or induce arrhythmias
sin II formation, thereby reducing its vasoconstric- to which such persons are prone. Although vasodila-
tive action and its enhancements of these steroid se- tor therapy improves short-term survival when added
cretions and consequent salt retention and plasma to treatment with glycosides and diuretics, the mor-
volume expansion. These agents produce a short- tality rate remains extremely high once acute failure
term prolongation of life. However, a review’ of tri- ensues. The prevention of extensive myocardial dam-
als indicates that heart failure has a high mortality age caused by coronary disease and the control of
rate, and in its chronic form, produces much suffering hypertension are of paramount importance. The
and a large economic burden for the patient, family, prevention of left ventricular dilatation and remod-
and society. elling before myocardial decompensation appears to
It appears that intervention must be started earlier be the key to delay of failure.
in the course of heart failure when there is only early
evidence of impaired left ventricular function that REFERENCES
can be sought in high-risk candidates for heart fail- 1. Firth BG, Yancy CW. Survival in congestive heart failure: have
ure by noninvasive testing. At this point, the under- we made a difference? Am J Med 1990;88:1-3N-l-7N.
2. McKee PA, Castelli WP, McNamara PM, et al. The natural
lying cause should be corrected and the myocardium history of congestive heart failure: the Framingham Study. N
supported and compensatory mechanisms available Engl J Med 1971;26:1441-6.
to the heart enhanced. Because of the lethal outlook, 3. Cupples LA, D’Agostino RB. Framingham Study Monograph
Section 34. An epidemiological investigation of cardiovascular
symptomatic relief alone is an inadequate treatment disease. Some risk factors related to the annual incidence of
goal. The optimal approach to the problem appears cardiovascular disease and death using pool of repeated bien-
to require the early detection and correction of pre- nial measurements: 30-year follow up. In: Kannel WB, Wolf
PA, Garrison RJ, eds. NIH Publication No. 87-2703, U.S.
disposing factors such as hypertension, LVH, im- Dept. of Commerce, National Technical Information Service,
paired glucose tolerance, and coronary disease before Springfield Va.
extensive myocardial damage ensues. Using simple 4. Yancy CW, Firth BG. Congestive heart failure. Dis Mon
1988;34:467-536.
office procedures and laboratory tests, it is possible to 5. Kannel WB, Plehu JF, Cupples LA. Cardiac failure and sud-
identify candidates for heart failure early in its den death in the Framingham Study. AM HEART J 1988;
course through a noninvasive study to detect persons 115:869-75.
6. Kannel WB, Cupples LA. Epidemiology and risk profile of
who have already sustained significant impairment of cardiac failure. Cardiovasc Drugs Ther 1988,2:387-95.
systolic and diastolic left ventricular function. At this
stage it may be possible to avoid irreversible myocar-