Influences of Cardiorespiratory Fitness

and Other Precursors on

Cardiovascular Disease and All-Cause
Mortality in Men and Women
Steven N. Blair, PED; James B. Kampert, PhD; Harold W. Kohl III, PhD; Carolyn E. Barlow, MS; Caroline A. Macera, PhD;
Ralph S. Paffenbarger, Jr, MD, DrPH; Larry W. Gibbons, MD, MPH

Objective.\p=m-\Toquantify the relation of cardiorespiratory fitness to cardiovascu- THE RELATION of physical inactivity
lar disease (CVD) mortality and to all-cause mortality within strata of other personal to various health problems is well es¬
characteristics that predispose to early mortality. tablished.1 Biological mechanisms that
contribute to the lower risk associated
Design.\p=m-\Observationalcohort study. We calculated CVD and all-cause death with activity include improved lipopro¬
rates for low (least fit 20%), moderate (next 40%), and high (most fit 40%) fitness
tein profile2 and carbohydrate metabo¬
categories by strata of smoking habit, cholesterol level, blood pressure, and health lism,3 lower blood pressure,4 and weight
status. loss.5 The inverse association between
Setting.\p=m-\Preventivemedicine clinic. physical activity or cardiorespiratory fit¬
Study Participants.\p=m-\Participantswere 25 341 men and 7080 women who ness and disease remains after statisti¬
completed preventive medical examinations, including a maximal exercise test. cal adjustment for these and other po¬
Main Outcome Measures.\p=m-\Cardiovasculardisease and all-cause mortality. tentially confounding variables.6-10
Results.\p=m-\Therewere 601 deaths during 211 996 man-years of follow-up, and Lower death rates associated with
89 deaths during 52 982 woman-years of follow-up. Independent predictors of regular physical activity are consistent
in different populations.6-8·11·12 However,
mortality among men, with adjusted relative risks (RRs) and 95% confidence inter- these studies were not specifically de¬
vals (Cls), were low fitness (RR, 1.52; 95% CI, 1.28-1.82), smoking (RR, 1.65; 95%
CI, 1.39-1.97), abnormal electrocardiogram (RR, 1.64; 95% CI, 1.34-2.01), chronic signed to investigate thoroughly the re¬
lation of activity or fitness to mortality
illness (RR, 1.63; 95% CI, 1.37-1.95), increased cholesterol level (RR, 1.34; 95% within various risk groups, nor to com¬
CI, 1.13-1.59), and elevated systolic blood pressure (RR, 1.34; 95% CI, 1.13-1.59). pare systematically the strength of as¬
The only statistically significant independent predictors of mortality in women were sociation of low activity or fitness with
low fitness (RR, 2.10; 95% CI, 1.36-3.21) and smoking (RR, 1.99; 95% CI, 1.25\x=req-\ the strength of other mortality predic¬
3.17). Inverse gradients were seen for mortality across fitness categories within tors. Examination of such relations will
strata of other mortality predictors for both sexes. Fit persons with any combination allow assessment of possible effect modi¬
of smoking, elevated blood pressure, or elevated cholesterol level had lower fication offitness and other mortality pre¬
dictors. We extended observations in the
adjusted death rates than low-fit persons with none of these characteristics. Aerobics Center Longitudinal Study by
Conclusions.\p=m-\Lowfitness is an important precursor of mortality. The protec-
tive effect of fitness held for smokers and nonsmokers, those with and without el- quantifying the relation of fitness to risk
of cardiovascular disease (CVD) and all-
evated cholesterol levels or elevated blood pressure, and unhealthy and healthy cause mortality within strata ofother pre¬
persons. Moderate fitness seems to protect against the influence of these other dictors of early mortality and compared
predictors on mortality. Physicians should encourage sedentary patients to become the strengths of associations between fit¬
physically active and thereby reduce the risk of premature mortality. ness and other predictors of mortality.
JAMA. 1996;276:205-210
SUBJECTS AND METHODS
Study Participants
From the Cooper Institute for Aerobics Research Presented in part at the National Institutes of Health
(Drs Blair and Kampert and Ms Barlow) and the Consensus Development Conference Physical Ac-
on Study participants were 25 341 men and
Cooper Clinic (Dr Gibbons), Dallas, Tex; Baylor Col- tivity and Cardiovascular Health, Bethesda, Md, De- 7080 women who ranged in age from 20
lege of Medicine, Houston, Tex (Dr Kohl); School of cember 18-20, 1995. to 88 years at baseline and completed a
Public Health, University of South Carolina, Columbia Reprints: Steven N. Blair, PED, Cooper Institute for
(Dr Macera); and Stanford University School of Medi- Aerobics Research, 12330 Preston Rd, Dallas, TX preventive medical examination between
cine, Palo Alto, Calif (Dr Paffenbarger). 75230. (e-mail: steve0704@aol.com). December6,1970, and December 31,1989.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/13/2018
 All study participants were residents of Table 1.—Baseline Characteristics of Survivors and Decedents, Men and Women, Aerobics Center Longi¬
the United States, had complete data from tudinal Study, 1970 Through 1989*
the medical examination, and achieved at Men Women
least 85% of their age-predicted maximal
heart rate (220 minus age in years) dur¬ Survivors Decedents Survivors Decedents
Characteristic (n=24 740) (n=601) (n=6991) (n=89)
ing the treadmill test. Most study par¬ Age, y 42.7 (9.7) 52.1 (11.4) 42.6(10.9)
ticipants were apparently healthy at base¬ Follow-up, y 8.4 (4.9) 7.1 (4.2) 7.5 (4.8) 7.6 (4.7)
line, although 1866 men and 350 women Body mass index, kg/m2 26.0 (3.6) 26.3 (3.5) 22.6 (3.9) 23.7 (4.5)
had an abnormal resting or exercise elec¬
Blood pressure, mm Hg
trocardiogram (ECG), and 4802 men and Systolic 121.1 (13.5) 130.4(19.1) 112.6(14.8)
958 women reported a history of 1 or Diastolic 80.7 75.4 (9.6) 79.9
(9.5) 84.8(11.8) (9.2)
more of the following chronic illnesses at
Total cholesterol
baseline: myocardial infarction, stroke, hy¬ mmol/L 5.5(1.1) 5.9(1.2) 5.2(1.0) 5.9(1.1)
pertension, diabetes mellitus, or cancer. mg/dL 213.1 (40.6) 228.9 (45.4) 202.7 (40.5)
Clinical Examination Fasting glucose
mmol/L 5.6 (0.9) 6.0(1.8) 5.2 (0.8) 5.6 (1.4)
The baseline evaluation was per¬ mg/dL 100.4(16.3) 108.1 (32.0) 94.4(14.5)
formed after participants gave their in¬ Fitness, %
formed written consent for the medical Low 20.1 18.8 44.9
examination and subsequent registra¬ Moderate 42.0 39.1 33.7
tion in the follow-up study. Examina¬ High 37.9 19.3 40.6 21.3
tions followed an overnight fast of at Current or recent smoker, % 26.3 36.9 30.3
least 12 hours and included a personal Family history of coronary
and family health history, a physical ex¬ heart disease, % 25.4 33.8 25.2 27.0

amination, a questionnaire on demo¬ Healthy, % 77.3 48.1 83.2 66.3

graphic characteristics and health hab¬
its, anthropometry, resting ECG, blood
chemistry analyses, blood pressure mea¬
surement, and a standardized maximal
exercise test on a motor-driven tread¬
mill. All procedures were administered
by technicians who followed a standard
manual of operations.
Height and weight were measured on
a standard physician's scale and stadiom-
eter, and body mass index was calculated
as weight in kilograms divided by the
square of height in meters. Blood pres¬
sures were measured by auscultatory
methods with a mercury sphygmomanom-
eter. Serum samples were analyzed by
automated techniques in a laboratory par¬
ticipating in the Centers for Disease Con¬
trol and Prevention Lipid Standardiza¬
tion Program. Current smokers or those
who quit smoking within 2 years of the
examination were classified as smokers,
and those who had never smoked or had
quit smoking more than 2 years before
the examination were classified as non-
smokers. We adopted this conservative
definition of smoking status because some
smokers may have quit temporarily in
preparation for the examination, and mor¬
tality risk for recent quitters is similar to
that of continuing smokers.
We measured participants' cardiore¬
spiratory fitness with a maximal exercise
test following a standard protocol.13 Spe¬
cific details of treadmill speed and eleva¬
tion have been described.8 Exercise test
performance with this protocol correlates
highly (r>0.92) with measured maximal
oxygen uptake.14·15 The advantage in us¬
ing fitness in this study is that, although
there is a genetic component, fitness is
determined primarily by exercise hab¬
its16 and is measured more objectively
*Data are given as means (SDs) except where noted. Low fitness indicates least fit 20%; moderate fitness, next
40%; high fitness, most fit 40%; family history of coronary heart disease, either parent dead of coronary heart disease;
chronic illness and abnormal ECG. Recent smokers had
healthy, no no
than self-report of physical activity, thus
reducing misclassification bias.
Mortality Surveillance
We followed up study participants for
mortality from the date of their baseline
examination to the date of death for de¬
cedents, or to December 31, 1989, for
survivors. We used the National Death
Index to identify possible decedents and
then retrieved official death certificates
from 44 states for these study partici¬
pants. The underlying cause and up to 4
contributing causes of death were coded
by a nosologist according to the Inter¬
national Classification of Diseases,
Ninth Edition, Revised, with CVD mor¬
tality defined as codes 390 to 449.9.
Data Analysis
Treadmill time at the baseline exami¬
nation was used to group study partici¬
pants into fitness categories based on age-
and sex-specific cutpoints. The least fit
20% of participants in each age-sex group
were classified as low fit, the next 40% as
moderately fit, and the remaining 40% as
high fit. Participants were classified as
healthy unless they reported chronic ill¬
nesses or had an abnormal ECG at base¬
line. Indicators of high- and low-risk cat¬
egories of other variables were based on
standard cutpoints.
Log-linear proportional hazards mod¬
els were used to estimate relative risks
(RRs) of CVD and all-cause mortality
separately for men and women. Cox par¬
tial likelihood methods17 were used to fit
all models, providing point estimates, 95%
within 2 years of baseline.
quit smoking
confidence intervals (CIs), and hypoth¬
eses tests. Survival time in days was
counted from the day of examination un¬
til death or end of observation on De¬
cember 31, 1989. Mortality rates per 10 000
person-years of observation are reported
as adjusted by Cox regression. All re¬
ported values are 2-sided.
RESULTS
The average follow-up interval from
baseline examination to date of death or
to December 31, 1989, was 8.4 years
(range, 0.1-19.1 years) in men and 7.5 years
(range, 0.1-18.9 years) in women. In men,
601 deaths occurred during 211996 man-
years of observation, with 226 deaths from
CVD. In women, 89 deaths occurred dur¬
ing 52 982 woman-years of follow-up, with
21 deaths from CVD. Baseline descrip¬
tive characteristics of the study group
are shown in Table 1.
Adjusted death rates and RRs for CVD
mortality and all-cause mortality among
men and women are shown in Table 2 and
Table 3, respectively. Data are given for
low cardiorespiratory fitness and 8 other
predictors of early death. Low fitness,
cigarette smoking, elevated systolic blood
pressure, elevated serum cholesterol level,
and poor health status (abnormal ECG or
chronic illness) were associated signifi¬
cantly with CVD mortality and all-cause
mortality in men. Low fitness and smok¬
ing were related significantly to all-cause
mortality in women. Only elevated fast¬
ingglucose level and abnormal ECG were
associated with CVD mortality in wom¬
en, although low fitness was ofborderline

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/13/2018
 Table 2.—Cardiovascular Disease Mortality and All-Cause Mortality Risk Analyses for Selected Mortality Predictors, Men, Aerobics Center Longitudinal Study,
1970 Through 1989*
Cardiovascular Disease All Causes
1
Relative Risk Relative Risk
Person-Years Death Rate/ I -1 Death Rate/ I- —I
of Follow-up 10 000 Adjusted* (95% 10 000 Adjusted* (95%
No. of (% of Person- No. of Person- Confidence No. of Person- Confidence
Mortality Predictor Subjects Years) Deaths Years* Adjusted! Interval) Deaths Yearsf Adjusted* Interval)
Low fitness
(20% least fit) 5223 54729(26) 111 20.0 2.69 1.70(1.28-2.25) 250 45.5 2.03 1.52(1.28-1.82)
Current or
recent smoker 6730 60829(29) 82 16.6 2.01 1.57(1.18-2.10) 222 42.7 1.89 1.65(1.39-1.97)
Systolic blood pressure
2140 mm Hg 2759 26 398 ( 12) 87 19.5 2.07 1.34(1.00-1.80) 184 43.6 1.67 1.30(1.08-1.58)
Cholesterol 26.2 mmol/L
(2240 mg/dL) 6025 51262(24) 106 1.65(1.26-2.15) 229 37.0 1.34(1.13-1.59)
Either parent dead of
coronary heart disease 6499 53440 (25) 84 14.3 1.51 1.18(0.89-1.57) 203 1.07(0.90-1.29)
Body mass Index ==27kg/m2 8198 65 534(31) 96 14.9 1.70 1.20(0.91-1.58) 223 34.3 1.33 1.02(0.86-1.22)
Fasting glucose
26.7 mmol/L
(a 120 mg/dL) 1396 13229(6) 36 15.4 1.49 0.95(0.66-1.37) 92 44.3 1.63 1.24 (0.98-1.56)
Abnormal
electrocardiogram 1866 15680(7) 99 36.7 4.28 3.01 (2.24-4.04) 158 54.0 2.05 1.64(1.34-2.01)
Chronic illness 4802 41016(19) 124 26.3 3.80 2.52(1.89-3.36) 242 1.63(1.37-1.95)
Totals 25 341 211996(100) 226 10.7§ 601 28.3§
*AII comparisons are dichotomies, with the referent category being the low-risk group (relative risk =1 ), and the high-risk group data shown In the table. Data for the reference
categories are not included, but can be estimated for each predictor by subtracting the values in the table for the high-risk group from the totals (25 341 men, 211 996 man-years,
601 deaths from all causes, 226 deaths from cardiovascular disease; 7080 women, 52 982 woman-years, 89 deaths from all causes, 21 from cardiovascular disease). Ellipses
indicate not
applicable.
tAdjusted for age and examination year.
^Adjustedrate.for age, examination year, and each of the other variables in the table.
§Crude

Table 3.—Cardiovascular Disease Mortality and All-Cause Mortality Risk Analyses for Selected Mortality Predictors, Women, Aerobics Center Longitudinal Study,
1970 Through 1989*
Cardiovascular Disease All Causes
I
Relative Risk Relative Risk
Person-Years Death Rate/ -1 Death Rate/ I— —I
of Follow-up 10 000 Adjusted* (95% 10 000 Adjusted* (95%
No. of (% of Person- No. of Person- Confidence No. of Person- Confidence
Mortality Predictor Subjects Years) Deaths Years* Adjusted* Interval) Deaths Years* Adjusted* Interval)
Low fitness
(20% least fit) 1352 13086(25) 11 7.7 2.79 2.42 (0.99-5.92) 40 28.8 2.23 2.10(1.36-3.26)
Current or
recent smoker 1321 10811(20) 6.0 1.70(0.58-4.97) 27 2.12 1.99(1.25-3.17)
Systolic blood pressure
2140 mm Hg 3959 (7) 2.06 1.47 (0.55-3.93) 15 15.1 0.89 0.76(0.41-1.40)
Cholesterol 26.2 mmol/L
(2240 mg/dL) 1223 9034(17) 3.9 0.99 0.74 (0.28-1.95) 31 18.9 1.1É 1.09 (0.68-1.74)
Eitherparent dead of
coronary heart disease 1788 13474(25) 3.2 0.76 0.58 (0.20-1.72) 24 12.9 0.71 0.70(0.43-1.16)
Body mass index 227 kg/m2 5486(10) 0.48 0.28(0.06-1.26) 15 19.5 1.18 0.94(0.52-1.69)
Fasting glucose
26.7 mmol/L
(2120 mg/dL) 148 1202(2) 14.2 3.80 4.10(1.11-15.2) 33.3 2.03 1.79(0.80-4.00)
Abnormal
electrocardiogram 350 2816(5) 5.38 5.02(1.90-13.3) 16 26.2 1.55(0.87-2.77)
Chronic illness 958 7085(13) 6.9 1.66(0.59-4.64) 18 18.1 1.09 1.05(0.61-1.82)
Totals 7080 52982(100) 4.0§ 89 16.8§
*AII comparisons are dichotomies, with the referent category being the low-risk group (relative risk =1 ), and the high-risk group data shown in the table. Data for the reference
categories are not included, but can be estimated for each predictor by subtracting the values in the table for the high-risk group from the totals (25 341 men, 211 996 man-years,
601 deaths from all causes, 226 deaths from cardiovascular disease; 7080 women, 52 982 woman-years, 89 deaths from all causes, 21 from cardiovascular disease). Ellipses
indicate not
applicable.
*Adjusted for age and examination year.
*Adjusted for age, examination year, and each of the other variables in the table.
§Crude rate.

statistical significance (P=.05). Among
women, smoking and elevated blood pres¬
sure have RRs for CVD mortality that
are similar in magnitude to those for men.
Failure to observe significant RRs for
these variables may be due to low statis¬
tical power resulting from the small num-
ber of CVD deaths in women. The in¬
creased mortality risk in low-fit men and
women is one of the strongest anteced¬
ents of mortality seen in these analyses.
We next calculated death rates for
all-cause mortality in cross-tabulation
analyses, using 3 levels of cardiorespi-
ratory fitness and 2 categories of the
other mortality predictors. Results for
men are shown in Figure 1 for fitness
categories by smoking, systolic blood
pressure, serum cholesterol level, and
health status. An inverse gradient of
risk was seen across fitness groups

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/13/2018
 Figure 1.—Cardiorespiratory fitness and all-cause mortality by levels of cigarette smoking, systolic blood pressure, total cholesterol level, and personal health sta¬
tus in 25341 men in the Aerobics Center Longitudinal Study. Height of the bars shows the all-cause death rates adjusted for age, year of baseline examination,
chronic illness (myocardial infarction, stroke, diabetes, hypertension, or cancer), abnormal electrocardiogram, parental history of death from coronary heart disease,
fasting blood glucose level, body mass Index, and each of thè other risk factors in the figure. Numbers in the bars represent number of deaths. The values are
for a test of trend across cardiorespiratory fitness categories (low fitness indicates least fit 20%; moderate fitness, next 40%; and high fitness, most fit 40%).

within each stratum of the other pre¬
dictors, a lower death rate in both strata
of each of the other predictors for mod¬
erately fit men when compared with low-
fit men, and an even lower rate for high-
fit men. The reduced risk for all-cause
mortality in the high-fit men compared
with the low-fit men ranged from 32%
for elevated systolic blood pressure to
50% for elevated cholesterol level and
poor health status. We performed simi¬
lar analyses for fitness stratified by fast¬
ing glucose level, family history, and
body mass index. Results were similar
to those shown in Figure 1, ie, inverse
gradients of risk were seen across fit¬
ness categories in both strata of each of
these other predictors (data not shown).
The association of fitness to mortality
was similar and was statistically signifi¬
cant in both strata of each of these vari¬
ables.
We conducted similar cross-tabulation
analyses in men for fitness and other
mortality predictors with CVD as the
outcome. Results (data not shown) were
similar to those for the all-cause mor¬
tality analyses, except that fitness was
not associated with CVD mortality in
men with systolic blood pressure 140
mm Hg or higher. The trends across
fitness categories were significantly dif¬
ferent (P=.005) in men who had normal
blood pressure compared with men who
had elevated systolic blood pressure.
Figure 2 shows cross-tabulations of
fitness to other risk characteristics in
relation to all-cause mortality in wom¬
en, with results that are consistent with
those observed in men. For each of the
predictors considered, the moderate- and
high-fit women in each risk stratum had
lower death rates than the low-fit wom¬
en, although the rates were more un¬
stable due to the smaller number of
deaths. High-fit women with elevated
cholesterol levels had a 23% lower risk
of death than low-fit women, which was
the smallest difference in risk between
high- and low-fit women. The largest
difference in risk for the high-fit women
was in those with systolic blood pres¬
sure of 140 mm Hg or greater, who had
an 81% lower risk of all-cause mortality
when compared with low-fit women. Dif¬
ferences in risk for other subgroups were
between these extremes. Too few CVD
deaths occurred in women to reliably
perform these analyses for the end point
of CVD death.
We calculated death rates for cross-
tabulations of fitness by presence or ab¬
sence of nonspecific combinations of the
3 other major CVD mortality predictors:
cigarette smoking, high blood pressure,
and high cholesterol level. Standard cut-
points were used to identify high risk
status for the other predictors (current
or recent smoking, systolic blood pres¬
sure of > 140 mm Hg, and cholesterol level
of >6.2 mmol/L [240 mg/dL). Participants
were grouped in 3 risk categories based
on the total number of the other predic¬
tors that were present (0, any 1, or any 2
or all 3), which then were cross-tabulated
with the 3 fitness categories.
Figure 3 shows the results for all-cause
mortality, which were consistent for men
and women. The analyses for women are
shown for 2 cardiorespiratory fitness cat¬
egories (moderate- and high-fitness cat¬
egories combined) because of the small
number of deaths. A higher cumulative
total of other predictors was associated
with increased risk in each fitness cat¬
egory. A graded inverse trend of death
rates was seen from low to high fitness
within strata of the number of other risk
factors. We observed a large difference
in risk between the highest and lowest

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/13/2018
 Figure 2.—Cardiorespiratory fitness and all-cause mortality by levels of cigarette smoking, systolic blood pressure, total cholesterol level, and personal health sta¬
tus In 7080 women in the Aerobics Center Longitudinal Study. Height of the bars shows the all-cause death rates adjusted for age, year of baseline examination,
chronic illness (myocardial Infarction, stroke, diabetes, hypertension, or cancer), abnormal electrocardiogram, parental history of coronary heart disease, fasting
blood glucose level, body mass Index, and each of the other risk factors in the figure. Numbers in the bars represent number of deaths. The values are for a
test of trend across cardiorespiratory fitness categories (low fitness indicates least fit 20%; moderate fitness, next 40%; and high fitness, most fit 40%).

risk groups. For example, in men, mul¬
tivariable adjusted death rates were 57.3
per 10 000 man-years of observation in
low-fit men with 2 or 3 risk factors, and
17.6 per 10 000 man-years in high-fit men
with none of the other risk factors. The
RR for these 2 contrasting groups was
3.25 (95% CI, 2.26-4.69), and the RR for
the same groups of women was 4.18 (95%
CI, 1.78-9.78). High-fit men with 2 or 3 of
the other predictors had a 15% lower
death rate than low-fit men with none of
the other predictors. Similar data for
women were even more striking: the ad¬
justed all-cause death rate was 31.6 per
10 000 woman-years of observation in the
low-fit women with no other predictors,
whereas fit women with 2 or 3 predictors
had a death rate of 16.1 per 10000, ie,
almost 50% lower.
COMMENT
Three principal observations from the
analyses presented herein from this group
ofwell-educated men and women deserve
notice, although the extent to which the
results can be extended to other popula¬
tions is unknown. First, the inverse gra¬
dient of mortality rates across fitness cat¬
egories was consistent in various strata
ofthe other mortality predictors. The gen¬
eral pattern was that of a considerably
lower death rate in moderately fit men
and women compared with those in the
low-fitness group. Adjusted all-cause
death rates were from 17% to 39% lower
in moderately fit mean compared with
low-fit men who smoked cigarettes, had
elevated blood pressure, elevated choles¬
terol levels, or were unhealthy. The lower
mortality rates for similar analyses for
women ranged from 48% to 67%. For most
analyses, we found an even lower risk for
high-fit participants compared with mod¬
erately fit participants.
Men and women with normal levels of
the other mortality predictors also ben¬
efited from being in the moderate fit¬
ness category compared with the low-fit
category. Moderately fit men who were
nonsmokers had a 41% lower all-cause
death rate than those who were in the
corresponding low-fit category. Similar
data for the other predictors showed
lower death rates of 39% in those with
normal blood pressure, 27% in those with
normal cholesterol level, and 33% in the
healthy group. Compared with low-fit
women, moderately fit women had lower
all-cause death rates of 55% in nonsmok-
ers, 54% in normotensives, 56% in those
with normal cholesterol levels, and 46%
in the healthy group.
The second major finding of our analy¬
ses was the substantial strength and in¬
dependence of low cardiorespiratory fit¬
ness as a precursor for CVD and all-cause
mortality. The RRs for low fitness were
among the highest seen in our analyses.
The adjusted RRs of all-cause mortality
due to low fitness (1.52) and cigarette
smoking (1.65) were similar in men. The
RR for these 2 characteristics was ap¬
proximately 2 in women. These results
are consistent with our earlier finding
that becoming fit has a similar effect on
reduction in mortality to stopping smok¬
ing.10 In comparison, other significant pre¬
cursors of all-cause mortality in men, el¬
evated blood pressure and cholesterol
levels, had RRs of about 1.3. In men, low
fitness had a strong association with CVD
mortality (RR, 1.70), and was similar to
elevated cholesterol level (RR, 1.65).
The third major finding was that mod¬
erate and high levels of cardiorespira¬
tory fitness seem to provide protection
against the force of combinations of other
mortality predictors on deaths (Figure
3). Death rates were higher in persons

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/13/2018
 ter Longitudinal Study; the Cooper Clinic physi¬
cians and technicians for collecting the baseline
data; Melba Morrow, MA, for editorial assistance;
Stephanie Parker for secretarial support; Roberta
Bannister and her staff for data entry; and Marilù
Meredith, EdD, and Lee Andrus for computer ser¬
vices support.

References
1. Bouchard C, Shephard RJ, Stephens T, eds.
Physical Activity, Fitness, and Health: Interna-
tional Proceedings and Consensus Statement.
Champaign, Ill: Human Kinetics Publishers Inc;
1994:569-915.
2. Stefanick ML, Wood PD. Physical activity, lipid
and lipoprotein metabolism, and lipid transport. In:
Bouchard C, Shephard RJ, Stephens T, eds. Physi-
cal Activity, Fitness, and Health: International
Proceedings and Consensus Statement. Champaign,
Ill: Human Kinetics Publishers Inc; 1994:417-431.
3. Kriska AM, Blair SN, Pereira MA. The potential
role of physical activity in the prevention of non\x=req-\
insulin-dependent diabetes mellitus. In: Holloszy
JO, ed. Exercise and Sport Sciences Reviews. Bal-
timore, Md: Williams & Wilkins; 1994:121-143.
4. American College of Sports Medicine. Physical
activity, physical fitness, and hypertension: posi-
tion stand. Med Sci Sports Exerc. 1993;25:1-10.
5. Blair SN. Evidence for success ofexercise in weight
loss and control. Ann Intern Med. 1993;119:702-706.
6. Paffenbarger RS Jr, Hyde RT, Wing AL, Hsieh
CC. Physical activity, all-cause mortality, and lon-$
gevity of college alumni. N Engl J Med. 1986;314:
605-613.
7. Morris JN, Clayton DG, Everitt MG, Semmence
AM, Burgess EH. Exercise in leisure time: coronary
attack and death rates. Br Heart J. 1990;63:325-334.
8. Blair SN, Kohl HW III, Paffenbarger RS Jr, Clark
DG, Cooper KH, Gibbons LW. Physical fitness and
all-cause mortality: a prospective study of healthy
men and women. JAMA. 1989;262:2395-2401.
Figure 3.—Cardiorespiratory fitness and all-cause mortality by the number of multiple risk factors in 25 341 9. Paffenbarger RS Jr, Hyde RT, Wing AL, Lee I,
men (top) and 7080 women (bottom) in the Aerobics Center Longitudinal Study. Height of the bars shows Jung DL, Kampert JB. The association of changes
all-cause death rates adjusted for age, year of baseline examination, chronic illness (myocardial infarction, in physical-activity level and other lifestyle char-
stroke, diabetes, hypertension, or cancer), abnormal electrocardiogram, parental history of death of coro¬ acteristics with mortality among men. N Engl J
nary heart disease, fasting blood glucose level, and body mass index. Numbers in the bars represent the Med. 1993;328:538-545.
number of deaths. Cardiorespiratory fitness categories for men are low fitness, least fit 20%; moderate fit¬ 10. Blair SN, Kohl HW III, Barlow CE, Paffen-
ness, next 40%; and high fitness, most fit 40%; for women, the moderate and high fitness categories are barger RS Jr, Gibbons LW, Macera CA. Changes
combined. The 3 categories of multiple risk factors are none, any 1, and any 2 or all 3 among current or re¬ in physical fitness and all-cause mortality. JAMA.
cent smoking, total cholesterol level of 6.2 mmol/L (240 mg/dL) or higher, and systolic blood pressure of 140 1995;273:1093-1098.
mm Hg or higher. 11. Peters RK, Cady LD Jr, Bischoff DP, Bern-
stein L, Pike MC. Physical fitness and subsequent
myocardial infarction in healthy workers. JAMA.
with multiple risk factors, but fitness tion of fitness with CVD and all-cause 1983;249:3052-3056.
12. Helmrich SP, Ragland DR, Leung RW, Paffen-
was inversely associated with death mortality. Whether our subjects were un¬ barger RS Jr. Physical activity and reduced occur-
rates in persons who had no other pre¬ healthy or healthy, smokers or nonsmok¬ rence of non\p=m-\insulin-dependent diabetes mellitus.
dictors or had any 1, and in those with ers, had elevated blood pressure or in¬ N Engl J Med. 1991;325:147-152.
creased cholesterol levels or normal values 13. Balke B, Ware RW. An experimental study of
any 2 or all 3 predictors. High-fit per¬ physical fitness in Air Force personnel. U S Armed
sons with multiple predictors had lower for these variables, were obese or had Forces Med J. 1959;10:675-688.
death rates than low-fit persons who normal weights, did or did not have a 14. Pollock ML, Bohannon RL, Cooper KH, et al. A
had no other predictors. family history of coronary heart disease, comparative analysis of four protocols for maximal
Previous studies show that regular ex¬ or had combinations of other mortality
treadmill stress testing. Am Heart J. 1976;92:39-46.
15. Pollock ML, Foster C, Schmidt D, Hellman C,
ercise improves cardiorespiratory fitness16 predictors or were at low risk by stan¬ Linnerud AC, Ward A. Comparative analysis of
and has a beneficial effect on blood pres¬ dard criteria, all seemed to benefit from physiologic responses to three different maximal
sure,18 lipoprotein profile,2 and glucose being moderately or highly fit compared graded exercise test protocols in healthy women.
Am Heart J. 1982;103:363-373.
tolerance,3 and thus should be considered with low-fit men and women of like risk
16. American College of Sports Medicine. The rec-
therapeutic in persons with abnormali¬ profile. We believe that physicians should ommended quantity and quality of exercise for de-
ties of these risk factors. Physicians and counsel all of their sedentary patients to veloping and maintaining cardiorespiratory and
patients sometimes become discouraged become more physically active and im¬ muscular fitness in healthy adults. Med Sci Sports
Exerc. 1990;22:265-274.
when a patient begins an exercise pro¬ prove their cardiorespiratory fitness. 17. Cox DR. Regression models and life tables. J R
gram in an attempt to reduce weight, Stat Soc. 1972;34:187-220.
lower blood pressure, or reduce choles¬ 18. Fagard RH, Tipton CM. Physical activity, fit-
terol levels and, despite good adherence This study was supported in part by US Public ness, and hypertension. In: Bouchard C, Shephard
to exercise, the clinical variable fails to Health Service research grant AG06945 from the RJ, Stephens T, eds. Physical Activity, Fitness,
National Institute on Aging, Bethesda, Md. and Health: International Proceedings and Con-
improve. Our study underscores the We thank our many study participants; Kenneth sensus Statement. Champaign, Ill: Human Kinetics
strong, graded, and independent associa- H. Cooper, MD, for establishing the Aerobics Cen- Publishers Inc; 1994:633-655.

Downloaded From: by a Tufts Univ. Hirsh Health Sciences Library User on 07/13/2018