You are on page 1of 6


Heart rate and cardiovascular mortality: The

Framingham Study
William B. Kannel, M.D., MPH, Craig Kannel, Ralph S. Paffenbarger, Jr., M.D.,
Dr. P.H., and L. Adrienne Cupples, Ph.D. Boston, Muss., Rochester, N. Y.,
and Stanford, Calif.

A number of epidemiologic studies have reported a a cohort of 5209 men and women for 30 years with
relationship between heart rate and coronary heart biennial examinations at the Framingham Study
disease.‘-* This report extends the investigation to clinic. This report focuses on those 5070 who were
other cardiovascular disease end points and to free of cardiovascular disease at entry into the study.
women. It also examines the relationship to noncar- Each examination included an extensive cardiovas-
diovascular death in an attempt to determine cular history and examination, an ECG, measure-
whether the relationship is unique to cardiovascular ments of vital capacity, body weight, and skin fold
disease. Data in the Framingham Study cohort thickness, various blood chemistry and blood pres-
covering 30 years of follow-up are employed in this sure determinations, and a history of cigarette
extended analysis of the heart rate effect. smoking. Heart rates were determined from resting
The specific objectives of this report are: (1) to ECG examinations with subjects supine. At each
examine the relationship of the resting heart rate to examination, newly developed cardiovascular events
the occurrence of overall, cardiovascular, and coro- were recorded. Information about cardiovascular
nary mortality rates in the Framingham Study end points and deaths was also obtained by daily
cohort; (2) to determine whether the heart rate is surveillance of hospital admissions and discharges,
specific to cardiovascular death; (3) to examine the and mortality rates were continuously monitored by
heart rate relationship to surviving coronary attacks; periodic contact with the Department of Vital Sta-
(4) to determine the net effect of heart rate on tistics for death certification, and communication
cardiovascular death taking other risk factors into with family physicians and relatives. All cardiovas-
account; and (5) to examine these relationships in cular events, including deaths, were reviewed by a
each sex and in young (under age 65 years) and old panel of investigators using a set of established
(over age 65 years) subjects. criteria.5 Cardiovascular death events include coro-
nary deaths, sudden coronary deaths, stroke deaths,
METHODOLOGY and deaths from cardiac failure. Sudden coronary
The Framingham Study, which began in 1948, is a death was defined as collapse and death within
long-term prospective study of cardiovascular dis- minutes (under 1 hour) with no other apparent
ease. The sampling response rate, follow-up, and cause.
examination procedures have been reported else- For statistical analysis, subjects were placed at
where.5 The Framingham Heart Study has followed risk at each biennial examination according to heart
rate and age in each sex. The population at risk
From the Section of Preventive Medicine and Epidemiology, Boston among the 5070 subjects free of cardiovascular dis-
University School of Medicine, Evans Research Foundation, the University ease at entry was all persons alive or those free of an
of Rochester School of Medicine, The Department of Family, Community,
and Preventive Medicine, Stanford University Medical School, and the event under consideration as specified. Rates of
School of Public Health, Boston University School of Medicine. cardiovascular deaths for each sex in the age groups
Supported in part by contracts NIH-NO. l-HV-92922 and NIH-NO. 35 to 64 years and 65 to 94 years were age adjusted
l-HV-52971. by the direct method. The net effect of heart rate,
Received for publication Dec. 11, 1986; accepted Jan. 16, 1987.
taking age and other cardiovascular risk factors into
Reprint requests: William B. Kannel, M.D., MPH, Section of Preventive
Medicine/Epidemiology, Boston University School of Medicine, Doctor’s account, was determined by multiple logistic regres-
Office Building, Suite 1105, 720 Harrison Ave., Boston, MA 02118. sion analysis,6 predicting the risk of disease in the

June 1907
1490 Kannel et al. American Heart Journal

Mean H. R.
Age Group Male Female
o---Q 35-44 74.63 77.23
- 45-54 74.66 77.06
h*-.* 55-64 74.14 76.69
- 65-74 76.16 79.05

Fig. 1. Distribution of pulse rate: Percentages for examinations 4 and 13 combined.

Table I. Number of cardiovascular or overall deaths: men by almost 3 bpm according to data from
30-year follow-up Framingham Study examinations 4 and 13 to obtain a wider age range.
Age 35-64yr Age 65-95 yr
The sex difference tends to increase with advancing
Cause age. About 5% of heart rates were above 95 and
of death Men Women Men Women below 66 (Fig. 1). The distribution in women is
slightly narrower than in men, with fewer at the
Sudden deaths 90 20 70 53
Coronary heart 175 54 175 142 extremes. The distribution does not vary significant-
disease deaths ly from one examination to another.
Cardiovascular deaths 240 106 275 273 Overall mortakity rate. In both sexes overall mortal-
Total deaths 452 314 567 543 ity rates increased progressively with resting heart
Population at risk = persons free of cardiovascular disease at entry into the rate, with no indication of any critical values or
study. threshold which could be labeled as safe or haxard-
ous (Table II). In both young and old subjects, the
next 2 years given the risk profile at the beginning of number of deaths at any heart rate is greater in men
a 2-year examination cycle. than in women. Also, mortality risk gradients in
relation to heart rate are substantially steeper in
OBSERVATIONS men than in women at any age. Mortality trends in
Over 30 years of follow-up there were 766 deaths relation to heart rates are highly significant
among subjects under age65 and 1110 deaths among 0, < 0.01) for young and old subjects of both
those over age 65 years. One half of these deaths sexes.
(894) were cardiovascular deaths. Sixty-eight per- Cardiovascular mortality rate. Cardiovascular death
cent of the cardiovascular deaths in men and 52 % in in general is also related to heart rate in both sexes,
women were attributed to coronary heart disease at all ages (Table III). As for the overall mortality
(Table I). A substantial number of fatal cardiovas- rate, no threshold heart rates are apparent, and the
cular events are available for analysis in relation to impact in men appears to be somewhat greater than
heart rate in each sex, above and below age 65. in women. Again, the cardiovascular-related mortal-
Heart rates tended to increase with age in both ity rate is higher at any heart rate in men than in
sexes with the rates in women exceeding those in women. The relationship of heart rate to cardiovas-
Volume 113
Number 6 Heart rate and cardiovascular mortality 1491

Table II. Overall deaths by resting heart rate according to ageand sex: 30-year follow-up Framingham Study
Population at risk Age-adjusted annual
(Person-years) rate/1000

Resting Men Women Men Women

heart rate
(bpm) 35-64 65-94 35-64 65-94 35-64 65-94 35-64 65-94

30-67 11,858 3660 10,920 3560 6 35 3 22

68-75 13,726 3646 17,452 6068 8 43 4 28
76-83 7506 2172 10,862 4208 11 46 6 25
84-91 5884 1572 8894 3270 13 61 8 30
92-220 4338 1408 7460 3264 24 64 9 35
Population at risk = all persons alive.
All trends significant at p < 0.01.

Table III. Cardiovascular deaths by heart rate according to age and sex: 30-year follow-up Framingham Study

Age-adjusted annual rate11000

Persons free of
cardiovascular disease All persons alive

Resting Men Women Men Women

heart rate
(bpn-4 35-64* 6594f 35-64$ 6s94# 35-64 * 65-94* 3564* 6.5~945

30-67 2 10 1 4 3 17 1 10
68-75 2 11 1 10 4 20 1 16
76-83 3 11 1 7 6 23 2 13
84-91 4 17 2 8 7 31 3 14
92-220 7 22 2 9 13 32 2 18
Trend significance: *p < 0.001; tp < 0.01; $NS = not significant; §p < 0.05.

Table IV. Coronary heart diseasedeaths by heart rate according to age and sex: 30-year follow-up Framingham Study
Age-adjusted annual incidence/l&k?

Persons free of
coronary disease All persons alive

Resting Men* Women Men* Women

heart rate
(bpm) 3564* 65-94f 35-64# 6594# 35-64* 65&j &5-64# 6.594#

30-67 1 6 0 4 2 11 1 7
68-75 2 5 1 4 3 11 1 7
76-83 2 8 0 4 4 15 0 7
84-91 3 13 2 4 6 23 2 7
92-220 6 12 1 6 9 18 1 9
Trend significance: ‘p < 0.001; tp < 0.01; $NS = not significant.

cular death holds for deaths resulting from coronary cardiac failure, stroke, and peripheral arterial dis-
disease, stroke, and cardiac failure. For deaths from ease-heart rates in men were signi&antly related
stroke there is no clear male predominance at any to cardiovascular death (Table III). This suggests
heart rate, and heart rate risk gradients are less that the association does not depend on cardiac
impressive than for the other cardiovascular end damage producing a high heart rate.
points. Even excluding subjects with interim overt Coronary mortatlty rate. For the coronary heart
cardiovascular disease-including coronary disease, disease mortality rate in particular, there is also a
June 1987
1492 Kannel et al. American Heart Journal

Table V. Sudden death by heart rate according to age in Table Vi. Percentage of coronary attacks* as sudden
men: 30-year follow-up Framingham Study deaths by heart rate: 30-year follow-up Framingham
Age-adjusted annual incidence11000
Persons free of Age-adjusted
Resting prior coronary All persons percentage of sudden
heart disease alive Resting death?
rate heart rate
@pm) 35-64 * 65-94 * 35-645 65-94# (bpm) 35-64 65-94

30-67 1 2 1 4 30-67 16.7 11.1

68-75 1 3 1 4 68-75 12.5 20.0
76-83 1 4 2 7 76-83 11.1 19.0
84-91 2 7 3 12 84-91 18.2 26.9
92-220 3 6 5 6 92-220 21.4 22.2

Trend significance: *p < 0.01; tp < 0.001; tp < 0.05. *Coronary events other than angina.
tPopulation at risk = persons free of prior coronary heart disease.

relationship to heart rate, but this is confined to

men (Table IV). Again, the association holds, even
excluding men with interim overt clinical manifesta- index of general health (Table VII). An examination
tions of coronary disease. of the fraction of all deaths resulting from cardiovas-
Case fatality rates for those sustaining a coronary cular disease in relation to resting heart rate shows
attack aIso increased in relation to antecedent rest- little relation at any age in either sex (Table VIII).
ing heart rates, suggesting a diminished capacity to This suggests that heart rate may be a nonspecific
survive coronary attacks in those who have higher measure of health and mortality rates. The finding
heart rates. This is best illustrated for sudden of a relationship between heart rate and noncardio-
coronary deaths which were distinctly related to vascular death at all ages in both sexes supports this
heart rate in men (Table V). A sharp upward trend interpretation (Table IX).
was noted in both young and old subjects, except
perhaps in those over age 65, with heart rates over 90
bpm. Significantly, in both young and old men, the The finding of an association between a high
fraction of coronary heart disease attacks occuring resting heart rate and cardiovascular morbidity and
as sudden deaths increased substantially with heart mortality is not unique to the Framingham Study. A
rate, suggesting a specific relationship of rapid heart number of prospective studies have reported an
rates to occurrence of sudden death (Table VI). association between heart rate and coronary death,
Independent relationship. The occurrence of cardio- sudden death, and myocardial infarction.1-8 This
vascular death is related to a number of cardiovas- report confirms and extends these findings but
cular risk factors including systolic blood pressure, indicates that the association may be nonspecific to
serum cholesterol, blood glucose, cigarette smoking, cardiovascular death, with the exception of sudden
and EGG-left ventricular hypertrophy. Blood pres- coronary death, in men.
sure in particular is also correlated with heart rate. The pathogenesis of the connection between rapid
Taking all these factors and age into account, the heart rates and cardiovascular death is obscure, but
Z-year incidence of overall deaths, coronary deaths a number of mechanisms may be postulated. High
in general, and sudden deaths in particular are all heart rates may indicate poor physical fitness, which
independently related to heart rate (Table VII). For has been shown to be associated with increased
the overah mortality rate a highly significant rela- cardiovascular and coronary mortality rates.8 A low-
tionship with resting heart rate persists in both ered heart rate has been shown to have a retarding
sexes, adjusting for age and for all of the coexistent effect on coronary atherosclerosis in primates.g
cardiovascular risk factors. For coronary death in Beere et al9 reported that lowering the heart rate by
general and sudden death in particular, the indepen- sinoatrial node ablation retarded high-cholesterol,
dent relationship holds only for men, at all ages. diet-induced atherogenesis in primates. The average
Noncardiovascular mortality rate. Since the overall lesion area in the low heart rate subgroup was only
mortality rate is as strongly related to heart rate as one third that of the high heart rate subgroup
cardiovascular death (comparing regression coeffi- despite no difference in blood pressure, serum lipids,
cients), it is possible that rapid heart rates are an or body weight. Local hemodynamic effects includ-
Volume 113
Number 6 Heart rate and cardiovascular mortality 1493

Table VII. Net effect of heart rate on 2-year overall and cardiovascular mortality rates by age and sex: 30-year
follow-up Framingham Study
Standardized Age 35-64 yr Age 65-94 yr
regression coefficients Men Women Men Women

Ouerall mortality rate

Univariate 0.398* 0.296* 0.184* 0.141*
Bivariate 0.395* 0.292s 0.184* 0.134
Multivariate 0.314 0.211 0.156 0.1027
Sudden death
Univariate 0.449* 0.185$ 0.254t 0.099%
Bivariate 0.438* 0.175$ 0.252t 0.090%
Multivariate 0.342* 0.125t 0.238t 0.006 i
Coronary death
Univariate 0.419* 0.232% 0.209~ 0.062%
Bivariate 0.411* 0.222% 0.2068 0.054%
Multivariate 0.309* 0.070$ 0.175t -0.032%
Cardiovascular disease mortality rate
Univariate 0.407* 0.292* 0.189* 0.144t
Bivariate 0.401* 0.285* 0.186* 0.136f
Multivariate 0.288* 0.145% 0.147t 0.083%

Bivariate = heart rate and age.

Multivariate = heart rate, systolic blood pressure, serum cholesterol, glucose intolerance, number of cigarettes/day, ECG-left ventricular hypertropby, and
Population at risk = all persons alive and free of cardiovascular disease at entry into study.
*p < 0.001; tp < 0.05; tNS = not significant; $p < 0.01.

Table VIII. Proportion of deaths due to cardiovascular Table IX. Noncardiovascular deaths by heart rate accord-
diseaseby heart rate according to age and sex: 30-year ing to age and sex: 30-year follow-up Framingham
follow-up Study
Percent Age-adjusted annual rate/1000

Heart Men Women Heart Men Women

rate rate
@pm) 35-64 65-94 35-64 65-94 (bpm) 35-64* 65-94 * 3564* 65-94f

30-67 48.6 50.0 33.3 45.5 30-67 3 17 2 12

68-75 50.0 46.5 25.0 57.1 68-75 4 23 3 12
76-83 54.5 50.0 33.3 52.0 76-83 5 23 4 12
84-91 53.8 50.8 37.2 46.7 84-91 6 30 5 16
92-220 50.0 54.1 22.2 51.2 92-220 11 32 7 17

Population at risk = all persons alive. Population at risk = all persons alive.
*p < 0.001; tp < 0.05.

ing pulsatile flow have been shown to influence the

distribution of lesions in the arterial circulation.
Tachycardia may indicate subclinical loss of cardiac tolerance. Nevertheless, rapid heart rates may signi-
reserve as a result of impaired myocardial function, fy inappropriate autonomic arou~al.‘~~ l6 An elevated
leading to the observed higher case fatality rates in heart rate is often observed in hypertension and
coronary attacks and propensity to sudden deaths. alleged to indicate an abnormality in circulatory
High heart rates may indicate autonomic imbalance control.” It has been proposed that the higher heart
or pathology, which could affect the heart rate and rate may derive from an increase in norepinephrine
the cardiovascular system. or hypersensitivity to it,12 or perhaps to a hyperbeta-
Higher heart rates have been reported in diabetic adrenergic state. l1 Cigarette smoking is known to
patients and attributed to cardiac parasympathetic transiently elevate catecholamines and to stimulate
damage alone or combined with sympathetic dam- an increase in heart rate. A high heart rate increases
age.lO The heart rate mortality relationships report- myocardial oxygen requirements which may poten-
ed here are not accounted for by impaired glucose tiate myocardial ischemia at a given level of coro-
June 1997
1494 Kannel et al. American Heart Journal

nary atherosclerosis. l3 It has been shown that when
subjects with angina stop smoking there is a 1. Dyer AR, Persky V, Stamler J, et al. Heart rate as a
prognostic factor for coronary heart disease and mortality:
decrease in heart rate and an improvement in ST findings in three Chicago epidemiologic studies. Am J Epi-
segment changes provoked by exercise.14 In this demiol 1980;112:736-749.
study of the Framingham cohort the association of 2. S&o11 M, Hangerup LM. Risk factors of myocardial infarc-
tion and death in men aged 50 at entry. A ten-year prospec-
rapid heart rates with death persists, taking ciga- tive study from the Glostrup Population Studies. Dan Med
rette smoking and blood pressure into account (Ta- Bull 1977;24:252-255.
ble VII). 3. Medalie JH, Kahn HA, Neufeld HN, et al. Five-year myocar-
dial infarction incidence. II. Associations of singie variables
There is a distinct possibility that high heart rates to age and birthplace. J Chronic Dis 1973;26:329-349.
are an index of general poor health and lack of vigor 4. Friedman GD, Klatsky AL, Siegelaub AB. Predictors of
and, hence, the nonspecific relationship to all types sudden cardiac death. Circulation 1975;51(suppl 111):164-
of death. However, all of the foregoing mechanisms 5. Shurtleff D. Some characteristics related to the incidence of
may be operative. It remains to be determined cardiovascular disease and death: Framingham Study 18 year
whether reducing the heart rate by physical condi- follow-up. In: Kannel WB, Gordon T, eds. The Framingham
Study: an epidemiological investigation of cardiovascular
tioning or administration of beta blockers can diseases. Washington, DC, United States Department of
improve the cardiovascular mortality rate in general Health, Education, and Welfare: 1974: section 30.
or the sudden death rate in particular. 6. Erikssen J, Rodahl K. Resting heart rate in apparently
healthy middle-aged men. Eur J Appl Physiol 1979;42:61-
7. Astrand PO, Rodahl K. Textbook of work physiology: Physi-
The relation of resting heart rate on biennial ECG ological bases of exercise. 2nd ed. New York: McGraw-Hill
Book Company, Inc., 1977.
examinations to mortality rates over 30 years of 8. Kannel WB, Wilson P, Blaire SN. Epidemiologic assessment
follow-up of the Framingham cohort was examined of the role of physical activity and fitness in development of
based on 1876 total deaths and 894 cardiovascular cardiovascular disease. AM HEART J 1985;109:876-885.
9. Beere PA, Glagov S, Zarius CK. Retarding effect of lowered
deaths, evolving out of 5070 subjects free of cardio- heart rate on coronary atherosclerosis. Science 1984;226:180-
vascular disease at entry into the study. In both 182.
sexes, at all ages, all-cause, cardiovascular, and 10. Ewing DJ, Campbell IW, Clarke BF. Heart rate changes in
diabetes mellitus. Lancet 1981;1:183-185.
coronary mortality rates increased progressively in 11. Conway J. Hemodynamic aspects of essential hypertension in
relation to antecedent heart rates determined bien- humans. Physiol Rev 1984;64:617-660.
nially. A more impressive association to cardiovas- 12. Holtzman E, Goldbourt U, Rosenthal T, Yaaris, Neufeld HN.
Hypertension in middle-aged men. Associated factors and
cular disease was observed in men than in women, mortality experience. Isr J Med Sci 1983;19:25-33.
which was independent of associated cardiovascular 13. Aronow WS. Effect of passive smoking on angina pectoris. N
risk factors. Engl J Med 197829921-24.
14. McHenry PL, Farris JV, Jordan JW, Morris SN. Compara-
Case fatality rates following coronary events also tive study of cardiovascular function and ventricular prema-
increased with antecedent heart rate and the frac- ture complexes in smokers and non-smokers during maximal
tion of coronary deaths as sudden death increased treadmill exercise. Am J Cardiol 1977;39:493-498.
15. Eliot R, Buell JC. Role of emotions and stress in the genesis
strikingly with heart rate in men 35 to 64 years of of sudden death. J Am Co11 Cardiol 1985;5:956-986.
age. There was also a substantial excess of noncar- 16. McKinney ME, Miner MH, Ruddel G, et al. The standard-
diovascular deaths at high heart rates, and the ized mental test protocol, test-retest reliability and compari-
son with ambulatory blood pressure monitoring. J Psycho-
proportion of all deaths resulting from cardiovascu- physiol 1985;22:453-463.
lar disease did not increase with heart rate. The
excess cardiovascular deaths with more rapid heart
rates were also noted, excluding those with interim
overt cardiovascular disease, suggesting an effect
independent of preexisting cardiac damage.