Lee I, Rexrode KM, Cook NR, Manson JE, Buring JE. Physical Activity and Coronary Heart Disease in WomenIs "No Pain, No Gain" Passé?. JAMA. 2001;285(11):1447-1454. doi:10.1001/jama.285.11.1447
Author Affiliations: Division of Preventive Medicine (Drs Lee, Rexrode, Cook, Manson, and Buring) and Channing Laboratory (Dr Manson), Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Department of Epidemiology, Harvard School of Public Health (Drs Lee, Manson, and Buring), and Department of Ambulatory Care and Prevention, Harvard Medical School (Dr Buring), Boston, Mass.
Context Physically active women have lower coronary heart disease (CHD) rates
than inactive women. However, whether the association differs by intensity
of activity or in women at high risk for CHD is unclear.
Objective To examine the relation between physical activity, specifically investigating
walking (a light-to-moderate activity depending on pace), and CHD among women,
including those at high risk for CHD.
Design, Setting, and Participants Cohort study of 39 372 healthy female health professionals aged
45 years or older, enrolled throughout the United States between September
1992 and May 1995, with follow-up to March 1999. Recreational activities,
including walking and stair climbing, were reported at study entry.
Main Outcome Measure Correlation of CHD with energy expended on all activities, vigorous
activities, and walking.
Results A total of 244 cases of CHD occurred. Adjusting for potential confounders,
the relative risks (RRs) of CHD for less than 200, 200-599, 600-1499, and
1500 or more kcal/wk expended on all activities were 1.00 (referent), 0.79
(95% confidence interval [CI], 0.56-1.12), 0.55 (95% CI, 0.37-0.82), and 0.75
(95% CI, 0.50-1.12), respectively (P for linear trend
= .03). Vigorous activities were associated with lower risk (RR, 0.63; 95%
CI, 0.38-1.04 comparing highest and lowest categories). Walking also predicted
lower risk among women without vigorous activities. Among these women, the
multivariate RRs for walking 1 to 59 min/wk, 1.0 to 1.5 h/wk, and 2 or more
h/wk, compared with no regular walking, were 0.86 (95% CI, 0.57-1.29), 0.49
(95% CI, 0.28-0.86), and 0.48 (95% CI, 0.29-0.78), respectively. For walking
paces of less than 3.2 km/h (2.0 mph), 3.2 to 4.7 km/h (2.0-2.9 mph), and
4.8 km/h (3.0 mph) or more, compared with no regular walking, RRs were 0.56
(95% CI, 0.32-0.97), 0.71 (95% CI, 0.47-1.05), and 0.52 (95% CI, 0.30-0.90),
respectively. When analyzed simultaneously, time spent walking (P for linear trend = .01) but not walking pace (P for linear trend = .55) predicted lower risk. The inverse association
between physical activity and CHD risk did not differ by weight or cholesterol
levels (P for interaction = .95 and .71, respectively),
but there were significant interactions by smoking and hypertension status.
Physical activity was inversely related to risk in current smokers but not
hypertensive women (P for interaction = .01 and .001,
Conclusions These data indicate that even light-to-moderate activity is associated
with lower CHD rates in women. At least 1 hour of walking per week predicted
lower risk. The inverse association with physical activity was also present
in women at high risk for CHD, including those who were overweight, had increased
cholesterol levels, or were smokers.
Coronary heart disease (CHD) is the leading cause of mortality among
women in the United States.1 Physical inactivity
is among the risk factors for this disease. A 1990 meta-analysis concluded
that physically active individuals had about half the CHD rates of those who
were sedentary.2 However, less than one fifth
of the studies in the meta-analysis included women. Since then, additional
studies have been conducted in women, and the available evidence clearly indicates
that active women experience lower CHD rates than inactive women.3- 26
What is less clear are the kinds and intensity of physical activities
that are associated with lower risk. It is important to clarify this issue
in light of a recent physical activity recommendation that calls for at least
30 minutes of moderate-intensity physical activity (eg, brisk walking at 4.8-6.4
km/h [3.0-4.0 mph]) most days of the week.27,28
This contrasts with previous recommendations that advocated vigorous-intensity
exercise (eg, jogging, running) for at least 20 minutes continuously, 3 times
per week.29 A major difference is the current
emphasis on moderate instead of vigorous activity. This concession was made
partly to encourage physical activity among sedentary individuals because
the previous, more difficult prescription was believed to pose a barrier.28
Another issue on which few data exist is whether physical activity is
inversely related to risk among healthy women at high risk for CHD (eg, smokers).
This has important clinical implications because if an inverse relationship
exists, physicians should strongly promote physical activity in these women.
We therefore investigated the relationship between physical activity
and CHD risk among women, including those at high risk, specifically exploring
the association with walking.
Subjects were selected from the Women's Health Study, a randomized,
double-blind, placebo-controlled trial of low-dosage aspirin and vitamin E
for primary prevention of cardiovascular disease and cancer.30- 32
Between September 1992 and May 1995, female health professionals throughout
the United States and Puerto Rico were invited to participate. Women completed
a mailed baseline questionnaire on sociodemographic characteristics, health
habits, and medical history. Those who were eligible and willing to be in
the trial were enrolled into a 3-month run-in phase during which women took
their study pills (all placebos). At the end of the run-in phase, women completed
the run-in questionnaire that ascertained compliance with pill taking, health
habits, and recent medical history. Women with good compliance who were still
eligible and willing to participate were then randomized into the trial and
started taking their randomized pill assignment.
A total of 39 876 women aged 45 years or older who were free of
self-reported coronary heart disease, cerebrovascular disease, and cancer
(other than nonmelanoma skin cancer) were randomly assigned to the agents
tested. For this study, we excluded 504 women with missing information on
physical activity or weight or who provided postrandomization reports of CHD
occurring before randomization, leaving 39 372 women.
On the run-in questionnaire, we asked women to estimate the average
time (0, 1-19 min/wk, 20-59 min/wk, 1 h/wk, 1.5 h/wk, 2-3 h/wk, 4-6 h/wk,
or ≥7 h/wk) spent on 8 groups of recreational activities during the past
year: walking or hiking; jogging (slower than 10-minute miles); running (10-minute
miles or faster); bicycling, including use of stationary machines; aerobic
exercise, aerobic dance, or use of exercise machines; lower-intensity exercise,
including yoga, stretching, or toning; tennis, squash, or racquetball; and
lap swimming. We also inquired about the usual pace of walking (do not walk
regularly; <3.2 km/h [2.0 mph; easy, casual pace], 3.2-4.7 km/h [2.0-2.9
mph; normal, average pace], 4.8-6.3 km/h [3.0-3.9 mph; brisk pace], or ≥6.4
km/h [4.0 mph; very brisk/striding pace]) and the number of flights of stairs
climbed daily (0, 1-2, 3-4, 5-9, 10-14, or ≥15). Based on the energy cost
of these activities, we assigned a multiple of resting metabolic rate (MET
score) to each group of activities and stair climbing.33
Since resting metabolic rate (1 MET) is approximately 1 kcal/kg of body weight
per hour, we estimated energy expenditure by multiplying the assigned MET
score by body weight and hours per week of participation using the midpoint
of time categories.34 We summed kilocalories
per week from the 8 groups of recreational activities and stair climbing to
estimate weekly energy expenditure.
This assessment of physical activity is reliable and valid.35 In a random sample of nurses, the test-retest correlation
coefficient over 2 years was 0.59. Questionnaire estimates of physical activity,
compared with 4 past-week recalls of physical activity collected during the
year prior to questionnaire administration, had a correlation of 0.79; compared
with activity diaries kept for 4 separate weeks during the same year, the
correlation was 0.62.
Information on variables that could potentially influence the association
between physical activity and CHD risk was ascertained from the baseline and
run-in questionnaires. These variables were age, weight, height, cigarette
smoking status, diet (including alcohol consumption), history of hypertension,
history of elevated cholesterol level, history of diabetes mellitus, menopausal
status, use of postmenopausal hormones, and parental history of CHD. We considered
a history of hypertension to be positive if women reported this diagnosis
or blood pressure of higher than 140/90 mm Hg. Women were classified as having
an elevated cholesterol level if they reported this diagnosis or cholesterol
levels of more than 240 mg/dL (6.21 mmol/L). A positive history of diabetes
mellitus was based on self-report.
Every 6 months during the first year and then annually, women completed
brief mailed questionnaires that inquired about compliance to their assigned
treatment, serious adverse effects of study agents, risk factors, and end
points of interest to the trial. Women reported a diagnosis of myocardial
infarction (MI) or coronary revascularization procedures (coronary artery
bypass grafting [CABG] or percutaneous transluminal coronary angioplasty [PTCA])
on these questionnaires, or wrote or telephoned the study staff. Deaths were
reported by family members or postal authorities. Follow-up in the trial is
high: at 36 months, the latest follow-up point attained by all participants,
morbidity/mortality follow-up was more than 99% complete.
We sought medical records and other relevant information, including
death certificates and autopsy reports, for women reporting MI, CABG, or PTCA,
and for decedents. Reported diagnoses of CHD or death from CHD were considered
confirmed only after examination of all available information by an end-points
committee of physicians. Myocardial infarction was confirmed using World Health
Organization criteria (ie, symptoms plus either typical electrocardiographic
changes or elevated cardiac enzyme levels).36
Fatal CHD was documented from convincing evidence of a cardiovascular mechanism
from all available sources, including death certificates, hospital records,
and, for deaths occurring outside the hospital, observers' impressions. CABG
and PTCA were confirmed by hospital records. This report includes data as
of March 1999.
The following dimensions of physical activity were considered: (1) energy
expended on all activities assessed, (2) energy expended on vigorous recreational
activities, and (3) walking. Recreational activities requiring at least 6
METs (ie, jogging, running, aerobic exercise, aerobic dance, use of exercise
machines, tennis, squash, racquetball, and lap swimming) were categorized
Women were first categorized into approximate quartiles of energy expended
on all activities: less than 200, 200 to 599, 600 to 1499, and 1500 or more
kcal/wk. Proportional hazards regression37
was used to estimate hazard ratios (relative risks [RRs]) of CHD as a function
of these 4 categories of physical activity, and both age (in years) and randomization
to treatment assignment were controlled for. Post hoc power calculations revealed
80% power to detect an RR of between 0.6 and 0.7, comparing highest and lowest
categories. To test for a linear trend across categories of physical activity,
the 4 categories of physical activity were treated as a single ordinal variable.
A multivariable model was then used that additionally adjusted for other potential
confounders, including smoking status (never, past, or current smoker of <15
or ≥15 cigarettes/d); alcohol consumption (rarely, 1-3 drinks/mo, 1-6 drinks/wk,
or ≥1 drink/d); saturated fat intake, fiber intake, and consumption of
fruits and vegetables (in quintiles); menopausal status (premenopausal or
postmenopausal); postmenopausal hormone use (never, past, or current use);
and parental history of MI before age 60 years (no or yes). In the main analyses,
no adjustment was made for body mass index or history of hypertension, elevated
cholesterol level, or diabetes mellitus because these are biological intermediates
in the pathway between physical activity and decreased CHD risk. In secondary
analyses, adjustments were made for these intermediates.
For energy expended on vigorous recreational activities of at least
6 METs, women were categorized into 5 groups: no vigorous recreational activity
plus less than 200 kcal/wk expended on other activities, no vigorous recreational
activity plus at least 200 kcal/wk expended on other activities, and 1 to
199, 200 to 499, and 500 or more kcal/wk expended on vigorous recreational
activities. The cut points were chosen so that women who participated in any
vigorous recreational activities were divided into approximate thirds. Relative
risks for CHD were estimated in parallel fashion to that described herein.
Because walking is the most popular leisure activity among women,28 this activity was specifically examined in relation
to CHD risk. To prevent confounding by vigorous activities, analyses of walking
were conducted among the 22 865 women (58%) who reported no vigorous
recreational activity. A priori, some response categories were combined on
the run-in questionnaire for a more even distribution of women by time spent
walking (no regular walking, 1-59 min/wk, 1.0-1.5 h/wk, or ≥2.0 h/wk) and
usual walking pace (no regular walking, <3.2 km/h [2.0 mph], 3.2-4.7 km/h
[2.0-2.9 mph], or ≥4.8 km/h [3.0 mph]). Relative risks of CHD were estimated
in separate models for time spent walking and walking pace. To ascertain which
was more predictive, terms for both were then simultaneously included in a
Finally, we investigated whether the association of physical activity
with CHD rates differed among women without and with the following CHD risk
factors: body mass index (<25 kg/m2 or ≥25 kg/m2),
cigarette smoking (never, past, or current), and history of hypertension or
elevated cholesterol level (no or yes). To test whether the association differed,
an interaction term between physical activity (energy expended on all activities)
and each risk factor was added in 4 separate multivariable models.
During the study, participants were observed for an average of 5 years,
and 244 confirmed incidents of CHD occurred (nonfatal MI or fatal CHD, n =
125; CABG or PTCA, n = 199; women could have had more than 1 event). Table 1 shows the characteristics of the
participants. More active women had a lower mean body mass index than less
active women. Women with higher levels of physical activity were less likely
to smoke cigarettes but more likely to consume alcohol. They also had a healthier
diet, consuming less saturated fat, more fiber, and more fruits and vegetables.
More active women were more likely to use postmenopausal hormones. At higher
levels of physical activity, prevalences of hypertension, elevated cholesterol
level, and diabetes mellitus were lower. The least active women were somewhat
more likely to have had a parent with MI prior to age 60 years.
In analyses that were adjusted for age and randomized treatment assignment,
there was a strong inverse association (P for linear
trend <.001) with energy expended on all activities in relation to CHD
rates (Table 2). The inverse association
persisted after further adjustment for smoking status, diet, alcohol use,
menopausal status, postmenopausal hormone use, and parental history of MI
before age 60 years (P for linear trend = .03). Women
who expended 600 to 1499 kcal/wk were at significantly lower risk of subsequently
developing CHD than less active women (RR, 0.55; 95% confidence interval [CI],
0.37-0.82). At higher levels of energy expenditure, no additional risk reduction
When biological intermediates (body mass index, hypertension, elevated
cholesterol level, and diabetes mellitus) were controlled for in secondary
analyses, the inverse relationship was attenuated. The RRs associated with
the 4 categories of energy expended on all activities were 1.00 (referent),
0.79, 0.62, and 0.84, respectively (P for linear
trend = .14).
Nonfatal MI or fatal CHD and CABG or PTCA were then investigated separately.
Significant inverse associations were observed for each end point. The multivariable
RRs for nonfatal MI or fatal CHD associated with the 4 categories of energy
expended on all activities were 1.00 (referent), 0.68, 0.57, and 0.65, respectively
(P for linear trend = .05). For coronary revascularization
procedures, they were 1.00 (referent), 0.78, 0.49, and 0.74, respectively
(P linear trend = .03).
Next, the energy expended on vigorous recreational activities of at
least 6 METs was examined. When highest and lowest categories of women were
compared, there was a lower risk of CHD among the most active women that was
of borderline significance (RR, 0.63; 95% CI, 0.38-1.04).
Whether walking is inversely related to risk of subsequent CHD among
women who did not engage in vigorous activities was also assessed. Among participants,
22 865 women (58%) reported no vigorous recreational activity; any activity
these women (75% of the 22 865) carried out consisted primarily of walking.
In this subgroup, both time spent walking and usual walking pace were inversely
related to CHD risk when analyzed separately (Table 3). Women who walked at least 1 hour per week or whose usual
walking pace was at least 4.8 km/h (3.0 mph) experienced about half the CHD
risk of women who did not walk regularly. To ascertain which of the 2 walking
parameters was more important, variables for both were entered in a single
multivariable model. In doing so, time spent walking (P for linear trend = .01) but not usual pace of walking (P for linear trend = .55) was significantly related to lower CHD rates,
indicating that time but not pace independently predicted lower risk.
Finally, whether the association of physical activity with CHD risk
differed among women without and with CHD risk factors was examined (Figure 1). There was no evidence that the
inverse association differed between women of normal weight and overweight
women (P for interaction = .95). There also was no
evidence that the inverse relationship differed between women without and
with elevated cholesterol level (P for interaction
= .71). However, there were significant interactions by smoking status and
history of hypertension (P for interaction = .01
and .001, respectively). Physical activity was inversely associated with CHD
rates in current and past smokers, but not in women who had never smoked (P for linear trend = .005, .10, and .75, respectively).
Among nonhypertensive women, an inverse association with physical activity
(P for linear trend = .001) was observed; among hypertensive
women, a u-shaped relationship (P for quadratic trend
= .07) was observed.
This study clearly indicates that physically active women have lower
CHD rates. It is encouraging to observe that vigorous activities were not
necessary for lower CHD rates. Among women who did not engage in vigorous
activities, walking (a light- to moderate-intensity activity, depending on
pace33) was associated with lower risk. These
data suggest that walking need not be fast-paced for benefit; time spent walking
was more important than walking pace. Additionally, we observed inverse associations
between physical activity and CHD risk among those who were overweight, smokers,
and women with elevated cholesterol levels. Among hypertensive women, there
was a u-shaped association between physical activity and CHD risk.
These findings support recent guidelines recommending moderate-intensity
physical activity for at least 30 minutes most days of the week (generating
energy expenditure of about 1000 kcal/wk).27,28
Moreover, they raise the possibility that even lesser degrees of activity
may decrease CHD risk. We found that time spent walking but not walking pace
independently predicted lower risk, implying that walking slower than 4.8
to 6.4 km/h (3-4 mph; ie, a light-intensity activity) may be beneficial. Women
who walked at least 1 h/wk had about half the CHD rates of women who did not
walk regularly. Assuming that walking 5 d/wk satisfies the definition of "most
days of the week," the recent guidelines recommend brisk walking for at least
Limitations in the assessment of walking should be considered when interpreting
our findings. Based on data from another study, women probably reported walking
pace validly; when asked to walk at a pace they gauged to be of at least moderate
intensity, both unfit and fit participants correctly walked more than 4.8
km/h (3 mph; Isabelle M. T. Bohlmann, MSc, written communication, August 2000).
Time spent walking may be less validly reported. The questionnaire used probably
measured purposeful walking (for transportation or exercise) only, rather
than all walking (eg, walking around the home). A recent study compared walking
reported on questionnaires and measured using pedometers.38
Reported walking was only 0.34 times the distance measured by pedometers.
It is unclear whether the recent activity guidelines refer to time spent on
all walking or purposeful walking. If the former were intended, our findings
probably are in accord with the recommended time. Women who reported walking
at least 1 h/wk were at significantly reduced risk. If only purposeful walking
was measured and was 0.34 of all walking, these women would actually have
spent 2.9 h/wk on all walking.
Walking was specifically investigated in 3 previous studies of women.
In the Nurses' Health Study, among women with no vigorous activities, those
who expended at least 3.9 MET-hours/wk walking (approximately 1 h/wk at brisk
pace) had lower CHD risk compared with those who expended no more than 0.5
MET-hours/wk.25 In the College Alumni Health
Study, walking at least 10 blocks/d (approximately 2 h/wk at brisk pace) lowered
risk of cardiovascular disease in women compared with walking less than 4
blocks/d.26 Among men and women aged 65 years
or older, walking more than 4 h/wk reduced risk of hospitalization due to
cardiovascular disease compared with walking less than 1 h/wk.17
The differences in the time required may be partly due to comparison with
reference groups who were more active than the referent in this study (no
vigorous activity and no regular walking); the reference group in the Nurses'
Health Study was most similar.
Few data, either in men or women, are available regarding whether walking
time or pace is more predictive of lower risk. In the Nurses' Health Study,
walking pace was associated with lower CHD risk after adjustment for MET-hours
of walking (a combined measure of walking pace and time).25
Our finding that time spent walking is more important—implying that
total energy expenditure is the relevant parameter—requires confirmation.
Some supporting evidence comes from an experiment in which sedentary women
were randomized to a control group or 1 of 3 exercise groups that walked the
same distances (4.8 km/d [3 miles/d], 5 d/wk) at 4.8, 6.4, or 8.0 km/h (3,
4, or 5 mph) for 24 weeks.39 All 3 exercise
groups improved in fitness (measured by maximal oxygen uptake) compared with
controls, with fitness improving in a dose-response fashion among exercisers.
High-density lipoprotein cholesterol levels, however, increased significantly,
to the same extent, among the 3 groups.
How can our findings be reconciled with those from other studies, primarily
of men, that observed only vigorous-intensity activity to be associated with
decreased CHD risk?40 We believe the data represent
a spectrum of responses to physical activity. Among persons with little activity,
institution of even light-to-moderate activity is associated with benefit.
Among persons who are more active and fit, vigorous activity is needed for
additional health benefits. Our study participants, in whom light-to-moderate
activity was associated with decreased CHD risk, were relatively inactive.
The most active group (the most active 21%; Table 1) expended at least 1500 kcal/wk in recreational activities
and stair climbing. In contrast, in a study of men in which physical activity
was assessed in similar fashion to this study, subjects were much more active;
the most active 20% expended at least 3129 kcal/wk.41
In that study, vigorous but not nonvigorous activity was associated with greater
Few data exist regarding the role of physical activity in the primary
prevention of CHD among high-risk women. As with this study, the Nurses' Health
Study also observed inverse associations among overweight women and smokers.25 Some investigators have postulated that overweight
but active individuals have lower morbidity and mortality than normal-weight
but sedentary individuals.42 In contrast, this
study and the Nurses' Health Study25 both indicate
that inactivity and overweight adversely affect CHD risk in women to a similar
extent. The observation of a u-shaped relationship among hypertensive persons
also has been reported in Finnish men.21 Two43,44 of 3 randomized trials that tested
exercise of varying intensities among hypertensive patients43- 45
suggested that higher-intensity physical activity is less effective in decreasing
systolic blood pressure than lower-intensity activity. This could partly account
for the u shape.
Strengths of this study include its large size, well-characterized participants,
and careful documentation of CHD. Furthermore, strict health criteria were
used for enrollment into the parent trial. Therefore, it is unlikely that
underlying disease could have decreased physical activity at study entry,
biasing results. It is also less likely that women who did not walk were limited
by physical disability.
Limitations include self-reports of physical activity. While the questionnaire
used is reliable and valid for large-scale studies,35
it does not offer the precision of, say, electronic devices that measure movement.
The latter, however, are impractical for large studies. Moreover, physical
activity data were gathered in this study prior to CHD occurrence; thus, any
misclassification is likely random, diluting the true association. While we
did adjust for a large number of potential confounders, confounding by unmeasured
factors is a concern in any observational study. Participants also were not
representative of the general population. The proportions of women who were
overweight or had elevated blood pressure or cholesterol levels were similar
to the general population, but fewer smoked and more used postmenopausal hormones.32 However, it is unlikely that the biologic effects
of physical activity would differ in the general population.
In conclusion, this study indicates that physical activity, easily within
the ability of almost all women, is associated with lower CHD rates. At least
1 hour of walking per week, regardless of pace, was associated with lower
CHD rates among relatively sedentary women. Because this is less than what
current guidelines suggest, confirmation of these findings is desirable. Meanwhile,
a conservative approach is to endorse current guidelines recommending moderate-intensity
physical activity for 30 min/d most days of the week. In the present study,
this level of physical activity was associated with lower CHD rates, even
among women who were overweight, smoked, or had elevated cholesterol levels.