Physical Performance in Peripheral Arterial Disease: A Slower Rate of
Decline in Patients Who Walk More
Mary McGrae McDermott, MD; Kiang Liu, PhD; Luigi Ferrucci, MD, PhD; Michael H. Criqui, MD, MPH; Philip Greenland, MD;
Jack M. Guralnik, MD, PhD; Lu Tian, ScD; Joseph R. Schneider, MD, PhD; William H. Pearce, MD; Jin Tan, MS; and Gary J. Martin, MD
Background: Exercise rehabilitation programs increase treadmill
walking performance in patients with peripheral arterial disease
(PAD) and intermittent claudication. However, it is unknown
whether patients with PAD who walk for exercise regularly have
less functional decline than those with less walking activity.
Objective: To determine whether patients with PAD who report
that they walk for exercise 3 or more times per week have less
annual functional decline than those who walk for exercise less
Design: Prospective cohort study with a median follow-up of 36
months (interquartile range, 24 to 36 months).
Setting: Academic medical center.
Participants: 417 men and women with PAD.
Measurements: Participants were classified at baseline and annu-
ally according to the number of times they reportedly walked for
exercise each week. Functional assessments (6-minute walk dis-
tance, 4-meter walking speed, summary performance score) were
measured at baseline and annually. Results were adjusted for age,
sex, ethnicity, comorbid conditions, body mass index, ankle–bra-
chial index, education, leg symptoms, cigarette use, geriatric de-
pression score, previous year’s level of functioning, and patterns of
Results: Compared with those who exercised less frequently, pa-
tients who walked for exercise 3 or more times per week had a
significantly smaller average annual decline in 6-minute walking
distance (?48.0 feet per year compared with ? 56.6 feet per year
for those who walked 1 to 2 times per week and ? 79.4 feet per
year for nonexercisers; P for trend ? 0.037). Patients who walked
for exercise at least 3 times per week experienced a smaller average
annual decline in the usual-paced 4-meter walking velocity (?0.014
m/s per year compared with ? 0.022 m/s per year for those who
walked 1 to 2 times per week and ? 0.045 m/s per year for
nonexercisers; P ? 0.005). Similar findings were observed for the
fast-paced 4-meter walk. The subset of asymptomatic patients who
walked for exercise 3 or more times per week had annual declines
in 6-minute walking performance (P ? 0.107), normal-paced walk-
ing velocity (P ? 0.065), and the summary performance score (P ?
0.115); however, these declines were smaller than those observed
in asymptomatic participants who walked fewer than 3 times per
Limitations: Because this was an observational study, associations
reported here cannot be construed as causal relationships. Sample
sizes for subgroup analyses were small, which limited statistical
Conclusion: Among patients with PAD, self-directed walking ex-
ercise performed at least 3 times weekly is associated with signifi-
cantly less functional decline during the subsequent year. Similar
trends were observed in the subset of asymptomatic patients with
PAD. These findings may be particularly important for the numer-
ous patients with PAD who do not have access to supervised
walking exercise programs.
Ann Intern Med. 2006;144:10-20.
For author affiliations, see end of text.
practices (1, 2). Most patients with the disease do not have
classical symptoms of intermittent claudication (1–3). Com-
pared with those without PAD, persons with the disease have
significantly greater functional impairment and more rapid
functional decline (2–4). Exercise rehabilitation that includes
supervised treadmill walking substantially improves treadmill
walking performance in men and women with intermittent
claudication (5). However, such barriers as cost, transporta-
tion, and program availability often limit access to exercise
rehabilitation programs for patients with PAD (6, 7).
Clinical guidelines for PAD recommend supervised
walking exercise, but evidence for the benefits of unsuper-
vised walking exercise is minimal to absent (8, 9). Specifi-
cally, it is unknown whether patients with PAD who en-
gage in regular self-directed walking exercise have less
functional decline than those who are sedentary. For per-
sons with PAD, supervised treadmill walking exercise 3 or
more times per week is more effective than less frequent
eripheral arterial disease (PAD) of the lower extremities
affects 20% to 30% of older patients in general medical
supervised walking exercise (5). Therefore, we conducted a
prospective observational study to examine whether pa-
tients with PAD who report that they walk for exercise 3 or
more times per week have less functional decline than PAD
participants who walk for exercise less frequently.
Methods for this longitudinal observational study of
men and women with and without PAD have been de-
Editors’ Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Summary for Patients. . . . . . . . . . . . . . . . . . . . . . . I-20
Conversion of figures and tables into slides
Annals of Internal Medicine
10 © 2006 American College of Physicians
scribed elsewhere (4). The protocol was approved by the
institutional review boards of Northwestern University and
Catholic Health Partners Hospital. Participants gave in-
Participants were 55 years of age and older at baseline.
Potential participants were identified consecutively from
patients who tested positive for PAD in 3 Chicago-area
noninvasive vascular laboratories. A few participants were
identified from lists of consecutive patients with recent ap-
pointments in our general internal medicine practice. This
latter group of patients was screened for PAD by calculat-
ing their ankle–brachial index at baseline; PAD was de-
fined as an index of less than 0.90 (10–12). Baseline visits
occurred between October 1998 and January 2000, and
follow-up visits were scheduled annually.
Exclusion criteria have been previously reported (4).
Patients with dementia, recent major surgery, or foot or leg
amputations were excluded. We also excluded nursing
home residents and wheelchair-bound patients. Patients
who did not speak English were excluded because the in-
vestigators were not fluent in non-English languages. Par-
ticipants who underwent leg revascularization were ex-
cluded from analyses after the procedure.
Ankle–Brachial Index Measurement
In accordance with established methods, we used a
hand-held Doppler probe (Nicolet Vascular Pocket Dop
II, Golden, Colorado) to obtain systolic pressures in the
right and left brachial, dorsalis pedis, and posterior tibial
arteries (4, 13, 14). To maximize precision, each pressure
was measured twice. We calculated the ankle–brachial in-
dex in each leg by dividing the mean of all 4 dorsalis pedis
and posterior tibial pressures by the mean of the 4 brachial
pressures (13). When calculated by this method, the index
correlates more closely with lower-extremity arterial func-
tion than when determined by alternative methods (13).
We used the average brachial pressures in the arm with
highest pressure when the measurement was higher in the
same arm in both measurement sets and the 2 brachial
pressures differed by 10 mm Hg or more in at least 1
measurement set; in such cases, subclavian stenosis was
possible (13, 14). The lowest ankle–brachial index was
used in analyses.
On the basis of a previous study (3, 4, 15), we used the
San Diego Claudication Questionnaire to classify patients
into 5 groups according to type of leg symptoms. Four
groups had exertional leg symptoms as determined by an
affirmative response to the question, “Do you get pain in
either leg or buttock on walking?” These participants were
further classified as having 1) intermittent claudication
(n ? 133), defined as exertional calf pain that does not
begin at rest, causes the participant to stop walking, and
resolves within 10 minutes of rest; 2) leg pain on exertion
and rest (n ? 76), defined as exertional leg pain that some-
times begins at rest; 3) atypical exertional leg pain/carry on
(n ? 39), defined as exertional leg symptoms that do not
begin at rest and do not stop the individual while walking;
and 4) atypical exertional leg pain/stop (n ? 89), defined
as exertional leg symptoms that do not begin at rest, stop
the individual from walking, and do not involve the calves
or resolve within 10 minutes of rest. A fifth group of pa-
tients was defined as asymptomatic (n ? 80) because they
reported no pain in either leg or buttock on walking.
We used algorithms developed for the Women’s
Health and Aging Study to document the following co-
morbid conditions: angina, diabetes mellitus, myocardial
infarction, stroke, heart failure, pulmonary disease, spinal
stenosis, disk disease, Parkinson disease, and hip fracture
(16). American College of Rheumatology criteria were used
to diagnose knee and hip osteoarthritis (17, 18).
Rationale for use of the specific functional outcome
measures used in the Walking and Leg Circulation Study
has been described previously (19).
Per a standardized protocol (20, 21), participants
walked up and down a 100-foot hallway for 6 minutes
after they were instructed to cover as much distance as
possible during the allotted time.
Patients with lower-extremity peripheral arterial disease
(PAD) benefit from supervised walking programs, but cost,
travel, and other factors often limit participation.
This prospective study shows that a self-directed program
of walking at least 3 times per week for exercise is associ-
ated with a significantly reduced functional decline during
the subsequent year in patients with PAD when compared
with those who walk less frequently.
Self-directed walking for exercise may benefit a much
larger proportion of patients with PAD than is currently
being served by supervised rehabilitation programs.
Observational studies, such as this one, cannot prove a
causal relationship between walking frequency and func-
Exercise and Peripheral Arterial Disease
3 January 2006 Annals of Internal Medicine Volume 144 • Number 1 11
Repeated Chair Rises
Participants sat in a straight-backed chair with their
arms folded across their chest and rose to a standing posi-
tion, repeating the exercise 5 consecutive times as quickly
as possible. We measured the time each patient required to
complete 5 chair rises.
Participants were asked to hold 3 increasingly difficult
standing positions for 10 seconds each: standing with feet
together and parallel (side-by-side stand); standing with
feet parallel, with the toes of 1 foot adjacent to and touch-
ing the heel of the opposite foot (semi-tandem stand); and
standing with 1 foot directly in front of the other (tandem
stand) (22, 23).
Walking velocity was measured with a 4-meter walk
performed at “usual” and “fastest” pace. For the usual-
paced walk, participants were instructed to walk at their
usual pace, “as if going down the street to the store.” For
the fastest-paced walk, participants were instructed to walk
as fast as they could. Each walk was demonstrated by the
research assistant. Participants were given the command
“ready, go”; timing began on “go.” Each walk was per-
formed twice, and the faster time in each pair was used in
analyses (22, 23).
Summary Performance Score
The summary performance score combined data from
the usual-paced 4-meter walking velocity, time to rise from
a seated position 5 times, and standing balance. Individuals
received a score of 0 for each task they were unable to
complete. Scores ranging from 1 to 4 were assigned for all
completed tasks; the scoring system was based on quartiles
of performance for over 5000 participants in the Estab-
lished Populations for the Epidemiologic Study of the El-
derly (22, 23). Scores were then summed to obtain the
summary performance score, which ranged from 0 to 12.
During each study visit, participants were classified as
current exercisers if they responded affirmatively to the
question, “During the past 2 weeks, have you gone walking
for exercise?” Frequency and duration of walking for exer-
cise were also determined at each visit. A minimum fre-
quency of 3 times per week and a minimum duration of 30
minutes per session are most optimal for supervised walk-
ing exercise programs in patients with PAD according to
published literature (5). Therefore, we defined optimal
walking frequency as 3 or more times per week and opti-
mal walking duration as 90 minutes per week.
We measured depressive symptoms annually by using
the Geriatric Depression Scale (short form), a 15-item
questionnaire (24). Possible scores for the questionnaire
ranged from 0 to 15; a score of 0 indicated no depressive
symptoms and a score of 15 indicated that all depressive
symptoms defined in the questionnaire were present.
Height and weight were measured at each visit. Body
mass index (BMI) was calculated by dividing the patient’s
body weight in kilograms by the square of his or her height
in meters. Patients annually reported cigarette use (pack-
years) and the number of blocks they walked during the
past week. The principal investigator reviewed all medica-
tion use to identify patients who used aspirin, statins, and
angiotensin-converting enzyme inhibitors.
We measured each patient’s physical activity continu-
ously over 7 days (beginning with the baseline visit) by
using the Caltrac (Muscle Dynamics Fitness Network,
Torrance, California) vertical accelerometer (25–29). This
accelerometer is the size of a pager and is worn at the waist.
All participants were eligible to receive the accelerometers,
but supply was limited; consequently, accelerometers were
only distributed if they were available at the time of the
participant’s first visit. Of all participants, 49% wore the
accelerometer at baseline. There were no significant differ-
ences in age, sex, ankle–brachial index, leg symptoms,
BMI, baseline performance, comorbid conditions, or fre-
quency of walking exercise between participants who wore
the accelerometer and those who did not.
Individuals who were unable to complete functional
measures at follow-up because of wheelchair confinement,
exhaustion, or other symptoms were classified as too dis-
abled to complete functional measures. The principal in-
vestigator made these decisions on the basis of information
contained in data collection forms that were blinded to
other participant characteristics. When no explanation was
provided for the participant’s refusal to complete func-
tional tests, the participant was considered too disabled to
walk if he or she met at least 2 of the following criteria:
The participant reported walking fewer than 5 blocks dur-
ing the previous week, the score for repeated chair rises
equaled 0 or 1, or the score for the standing balance test
equaled 0 or 1. The criteria were defined before data anal-
yses. Individuals who refused functional testing at fol-
low-up and met 2 of these criteria were assigned the min-
imum value for each test not completed. The minimum
value for each test was equivalent to the poorest perfor-
mance among those who completed testing at the corre-
On the basis of previous studies of exercise rehabilita-
tion programs in patients with PAD (5), we categorized
participants into 1 of the following groups: walking for
exercise 3 times or more per week, walking for exercise
Exercise and Peripheral Arterial Disease
12 3 January 2006 Annals of Internal Medicine Volume 144 • Number 1
fewer than 3 times per week, or no walking for exercise.
Baseline characteristics across these groups were compared
by using general linear models for continuous variables and
chi-square tests for categorical variables.
We compared change in functioning (for example,
6-minute walk distance) across different patient groups by
using a longitudinal or repeated-measure analysis of covari-
ance with a mixed-effects linear regression model (30). De-
tails of the mixed-effects regression and our strategy for
handling missing data are reported in the Appendix (avail-
able at www.annals.org). We analyzed these successive dif-
ferences, adjusting for baseline covariates (sex, age, and eth-
nicity) and a time-dependent covariate that represented
functional performance at the immediately preceding visit.
Parallel analyses were repeated with additional adjustments
for baseline comorbid conditions, leg symptoms, educa-
tion, and time-dependent covariates (BMI, ankle–brachial
index, depressive symptoms, and pack-years of smoking).
The independent variable for walking exercise was in-
cluded as a time-dependent variable, with the previous
year’s exercise status entered into the statistical model. For
example, in evaluating the relationship between exercise
and functional decline between the first and second fol-
low-up visits, the exercise status at the first visit was the
independent variable. Analyses were repeated with addi-
tional adjustments for aspirin, statin, and angiotensin-con-
verting enzyme inhibitor use (time-dependent variables).
To determine whether associations between walking
exercise and functional decline were consistent across levels
of functional performance, analyses were repeated after
stratifying participants according to tertile of the baseline
functional measure being analyzed. Analyses were also re-
peated within the 5 previously defined categories of leg
symptoms. In these subgroup analyses, functional decline
Figure 1. Description of potential study participants.
Exercise and Peripheral Arterial Disease
3 January 2006 Annals of Internal Medicine Volume 144 • Number 1 13
of participants who walked for exercise 3 or more times per
week was compared with that of a combined group of
participants who walked 0, 1, or 2 times per week; the
subset of patients who walked 1 or 2 times per week was
too small for independent analysis. Finally, associations be-
tween duration of exercise walked per week and functional
decline were determined. All analyses were performed by
using SAS statistical software, version 9.1 (SAS Institute,
Inc., Cary, North Carolina).
Role of the Funding Source
The funding source for this study played no role in the
design, conduct, or reporting of the study, or in the deci-
sion to submit the manuscript.
Figure 1 shows reasons for nonparticipation among
those potentially eligible. Characteristics of participants
with PAD are shown in Table 1 according to frequency of
walking exercise at baseline. More frequent walking exer-
cise was associated with a higher number of blocks walked
during the past week and higher physical activity as mea-
sured by the accelerometer. We observed no differences in
ankle–brachial index across the exercise categories. The
only significant differences in baseline characteristics across
the 3 walking groups were for African-American ethnicity,
leg symptom categories, and baseline performance on the
6-minute walk and usual-paced 4-meter walking test.
Figure 2 shows associations between walking exercise
frequency and average annual functional decline over 3
years of follow-up. After adjusting for age, sex, ethnicity,
previous year’s performance, ankle–brachial index, BMI,
cigarette use, leg symptoms, depressive symptoms, educa-
tion, comorbid conditions, and missing data patterns, we
observed significant, graded associations between exercise
Table 1. Baseline Characteristics of Participants according to Baseline Exercise Categories (n ? 417)
Characteristic Baseline Exercise CategoryP Value*
Walked >3 Times/Week
(n ? 143)
Walked 1 or 2 Times/Week
(n ? 83)
Little or No Walking
(n ? 191)
Age (SD), y
African-American ethnicity, %
Ankle–brachial index (SD), n
Body mass index (SD), kg/m2
Pack-years of smoking (SD), n
39.9 (35.6)34.7 (31.5)37.4 (33.3)0.52
Comorbid conditions, %
Cardiac or cerebrovascular disease†
Leg symptom categories, %
No exertional pain
Pain at rest
Education level, %
Less than high school
High school or college
Professional or graduate school
Geriatric Depression Scale score (SD)§2.75 (2.95) 3.33 (2.87)3.34 (3.33)0.21
City blocks walked in past week (SD), n
56.4 (73.1)29.5 (39.0) 18.1 (32.7)
6-minute walking distance (SD), ft
Usual-paced 4-meter walk (SD), m/s
Fastest-paced 4-meter walk (SD), m/s
Summary performance score (SD)?
* P values for comparisons across the exercise categories using analysis of variance for continuous variables and chi-square test for categorical variables.
† Included myocardial infarction, heart failure, angina, and stroke.
‡ Included knee arthritis, hip arthritis, hip fracture, spinal stenosis, and disk disease.
§ Possible score ranged from 0 to 15 with 0 being best and 15 being worst.
? Possible score ranged from 0 to 12 with 0 being worst and 12 being best.
Exercise and Peripheral Arterial Disease
14 3 January 2006 Annals of Internal Medicine Volume 144 • Number 1
frequency and average annual decline in 6-minute walk
performance, usual-paced walking speed, and fast-paced
walking speed. Results for the association between walking
exercise frequency and the summary performance score
were nearly statistically significant (P ? 0.053); associa-
tions between walking exercise frequency and decline in
each component of the summary performance score were
statistically significant (P ? 0.048 for 5 repeated chair ris-
es; P ? 0.013 for standing balance).
The findings shown in Figure 2 were similar when
Figure 2. Associations between frequency of self-directed walking exercise and functional decline in patients with peripheral
arterial disease of the lower extremities.
Adjusted for age, sex, ethnicity, ankle–brachial index, body mass index, previous year’s performance, leg symptoms, cigarette use, depressive symptoms,
education level, comorbid conditions, and patterns of missing data. Exercise frequency was entered as a time-dependent variable. P for trend ? 0.037,
0.005, 0.013, and 0.053 for 6-minute walk distance, normal-pace 4-meter walking velocity, fast-pace 4-meter walking velocity, and summary perfor-
mance score, respectively. *P ? 0.05 for comparison with no-exercise group.
Exercise and Peripheral Arterial Disease
3 January 2006 Annals of Internal Medicine Volume 144 • Number 1 15
analyses were repeated with adjustments for use of aspirin,
statins, and angiotensin-converting enzyme inhibitors (data
not shown). Findings were also similar when analyses were
repeated in the subgroup of participants who were 65 years
of age and older. Results did not change substantially when
analyses were repeated after excluding participants re-
cruited from the general internal medicine practice.
We repeated the analyses summarized in Figure 2 after
stratifying participants according to tertiles of baseline
functional performance. Within these tertiles, the trends in
associations between walking exercise frequency and func-
tional decline were observed more frequently in partici-
pants in the lowest 2 tertiles of baseline performance.
These results suggest that self-directed walking exercise
may be of greatest benefit to patients with PAD who have
the greatest functional impairment. However, subgroups
were small and associations were not statistically significant
in most instances (Table 2).
After stratifying patients by leg symptom category and
adjusting for the previously mentioned confounding vari-
ables, we compared participants who walked 3 or more
times per week with those who walked fewer than 3 times
per week. Asymptomatic patients with PAD who walked
for exercise 3 or more times per week had less functional
decline than those who walked for exercise less frequently
or not at all; results showed a trend toward statistical sig-
nificance for 3 functional outcomes (Figure 3). Results for
symptomatic groups generally showed similar associations,
but these findings were not statistically significant in most
cases (probably because of the small sample size). In a
model using data from all participants, there was no signif-
icant interaction between leg symptom categories, and
walking exercise frequency in relation to each measure of
functional decline. Therefore, associations between exercise
frequency and functional decline were not significantly dif-
ferent among leg symptom groups.
All participants with PAD were stratified according to
the amount of time spent walking for exercise per week. At
baseline, 107 participants walked for exercise 90 or more
minutes per week, 118 walked for exercise less than 90
minutes per week, and 191 did not walk for exercise. We
observed significant stepwise associations between the
amount of time spent in self-directed walking exercise per
week and functional decline (Figure 4).
Finally, to determine whether substantial changes in
exercise status were associated with different rates of func-
tional decline, we defined 3 groups on the basis of changes
in exercise habits between 2 consecutive years. Group 1
contained participants who changed from no walking ex-
ercise to exercising 3 or more times per week (between
baseline and first follow-up visit, n ? 22); group 2 con-
tained those participants who did not fall into group 1 or
group 3 (between baseline and first follow-up visit, n ?
360); and group 3 contained participants who changed
from walking for exercise 3 or more times per week to
no walking exercise (between baseline and first fol-
low-up visit, n ? 24). In time-dependent analyses, we
observed a linear association among these 3 groups and
average annual decline in 6-minute walk performance
(?45.03 feet per year in group 1; ?60.65 feet per year
in group 2; and ?139.3 feet per year in group 3; P for
trend ? 0.045). We observed no significant associations
Table 2. Adjusted Associations between Frequency of Self-Directed Walking Exercise and Functional Decline over 3-Year
Follow-up according to Baseline Functional Performance (n ? 417)*
Baseline Functional Measure Functional Decline by Group
No WalkingWalking 1 or 2 Times/Week Walking >3 Times/Week
6-minute walk, ft
Lowest tertile (?1000 ft)
Middle tertile (?1000 and ?1333 ft)
Highest tertile (?1333 ft)
Usual-paced 4-meter walk, m/s
Lowest tertile (?0.80 m/s)
Middle tertile (?0.80 and ?0.97 m/s)
Highest tertile (?0.97 m/s)
Fastest-paced 4-meter walk, m/s
Lowest tertile (?1.10 m/s)
Middle tertile (?1.10 and ?1.33 m/s)
Highest tertile (?1.33 m/s)
Summary performance score, n
Lowest tertile (score of ?9)†
Middle tertile (score of 10 or 11)
Highest tertile (score of 12)
* Analyses adjusted for age, sex, ethnicity, body mass index, cigarette use, leg symptoms, ankle–brachial index, Geriatric Depression Scale score, education level, previous
year’s performance, and patterns of missing data.
† P for trend ?0.05.
Exercise and Peripheral Arterial Disease
16 3 January 2006 Annals of Internal Medicine Volume 144 • Number 1
between these groups and functional decline for other
Among 417 men and women with PAD, walking for
exercise 3 or more times per week was associated with
significantly less average annual functional decline com-
pared with less frequent exercise. To our knowledge, this
study is the first to report associations between self-directed
walking exercise and the natural history of lower-extremity
functioning in patients who have PAD with and without
intermittent claudication. Although supervised treadmill
walking exercise programs significantly improve walking
performance in patients with intermittent claudication (5,
31), exercise rehabilitation programs require travel to the
exercise facility and are often not reimbursed by medical
insurance. Because of these logistic barriers, few patients
with PAD participate (7). Our results suggest that self-
Figure 3. Associations between frequency of self-directed walking exercise and functional decline in asymptomatic patients with
peripheral arterial disease.
Adjusted for age, sex, ethnicity, ankle–brachial index, body mass index, previous year’s performance, cigarette use, depressive symptoms, education level,
comorbid conditions, and patterns of missing data. P ? 0.107, 0.065, 0.311, and 0.115 for 6-minute walk distance, normal-pace 4-meter walking
velocity, fast-pace 4-meter walking velocity, and summary performance score, respectively.
Exercise and Peripheral Arterial Disease
3 January 2006 Annals of Internal Medicine Volume 144 • Number 1 17
directed walking exercise outside of a supervised setting is a
rational alternative that may prevent functional decline in
Previous studies demonstrated that supervised exercise
training is underused in patients with PAD and that phy-
sicians do not routinely recommend walking exercise to
these individuals (32–34). In the Minnesota Regional Pe-
ripheral Arterial Disease Screening Program, only 6.5% of
92 persons with the disease were participating in a super-
vised exercise program (34). Of the 92 participants, 45.7%
had been previously told that they had “blockages in their
leg arteries.” In a national survey of 438 randomly selected
Figure 4. Associations between weekly duration of self-directed walking exercise and functional decline in patients with peripheral
P ? 0.008, 0.007, 0.012, and 0.040 for 6-minute walk distance, normal-pace 4-meter walking velocity, fast-pace 4-meter walking velocity, and summary
performance score, respectively.
Exercise and Peripheral Arterial Disease
18 3 January 2006 Annals of Internal Medicine Volume 144 • Number 1
physicians who regularly treat patients with PAD, only
34.6% reported that recommending aerobic exercise was
“extremely important” for these patients (33).
An earlier randomized, controlled clinical trial com-
pared a supervised treadmill exercise walking program with
an unsupervised exercise walking program in 20 persons
with intermittent claudication (35). Participants in the un-
supervised walking exercise program were telephoned
weekly and encouraged to adhere to their prescribed walk-
ing program. At 3-month follow-up, participants in the
supervised treadmill walking exercise program significantly
increased their pain-free and peak treadmill walking time;
those in the home walking program did not (35). On the
basis of our findings, a future randomized, controlled trial
should be performed in patients with PAD who do not
exercise to determine whether an intervention that pro-
motes self-directed exercise better prevents functional de-
cline than no-exercise interventions.
Our findings for associations between frequency of
walking exercise and decline in the summary performance
score were nearly statistically significant. However, we ob-
served significant associations between frequency of walk-
ing for exercise and functional decline on each component
of the summary performance score. These results suggest
that walking exercise in an individual participant with
PAD is not associated with simultaneous increases in all
summary performance score components.
Recent data show that patients with PAD who are
asymptomatic are at particularly high risk for functional
decline (4). Although our sample size of asymptomatic par-
ticipants with PAD was limited, our data suggest that self-
directed walking exercise may prevent functional decline in
this group. To our knowledge, medical therapies to im-
prove lower-extremity functioning have not been tested in
a large group of patients who are asymptomatic or who
have symptoms other than intermittent claudication.
These data also underscore the importance of screening for
and identifying asymptomatic patients because they may
also benefit from walking exercise.
Only a small proportion of African-American patients
in our cohort walked for exercise 3 or more times per week.
Previous cross-sectional analyses show that African-Ameri-
can patients with PAD have greater functional impairment
than white patients (36). In addition, the prevalence of
PAD is higher in African-American individuals than in
white patients (37). Our data suggest that clinicians should
take steps to encourage African-American patients to in-
crease their frequency of walking exercise.
Our observation that individuals in the highest tertiles
of performance at baseline (Table 2) had greater average
annual decline in performance than those in lower tertiles
is probably attributable to the regression-to-the-mean phe-
nomenon. In addition, participants in the former group
had more opportunity for decline because of a “floor” ef-
fect for individuals in the lowest tertiles.
This observational study has 6 noteworthy limitations.
First, these data do not clearly indicate whether walking
exercise or the other healthy behaviors that are associated
with exercise are responsible for protection against func-
tional decline. However, we found no significant associa-
tions between BMI or pack-years of smoking and more
regular walking exercise. Second, we cannot rule out the
possibility that residual confounding contributed to the
observed differences in functional decline across exercise
groups. However, our analyses adjusted for multiple con-
founding variables, including comorbid conditions, BMI,
ankle–brachial index, cigarette use, education, depressive
symptoms, previous year’s performance, and patterns of
missing data. Additional unidentified confounders (such as
self-efficacy), however, may have contributed to the ob-
served differences across the exercise categories. Third, our
exercise frequency measure was based on self-report, but
the validity of our measure is supported by the observation
that more frequent exercisers had higher objectively as-
sessed activity levels than less frequent exercisers. In addi-
tion, misclassification of exercise status is likely to have
attenuated associations between exercise frequency and
functional decline. Fourth, we cannot rule out the possi-
bility that some participants who reported walking for ex-
ercise were participating in supervised exercise programs.
Fifth, we had insufficient numbers of accelerometers to
measure physical activity levels in all study participants.
Sixth, sample sizes were small, which limited statistical
power for subgroup comparisons.
In conclusion, self-directed walking exercise is associ-
ated with less functional decline among persons with PAD.
Similar findings were observed in symptomatic and asymp-
tomatic participants. Our data suggest that patients with
PAD who are unable or unwilling to participate in super-
vised walking exercise programs may benefit from self-
directed walking at home. Our data also suggest that Afri-
can-American patients with PAD should be a particular
focus of walking exercise recommendations.
From Northwestern University Feinberg School of Medicine, Chicago,
Illinois; Evanston Northwestern Hospital, Evanston, Illinois; National
Institute on Aging, Baltimore, Maryland; and University of California at
San Diego, La Jolla, California.
Potential Financial Conflicts of Interest: None disclosed.
Grant Support: By the National Heart, Lung, and Blood Institute (grant
numbers R01-HL58099 and R01-HL64739) and by the National Cen-
ter for Research Resources, National Institutes of Health (grant number
Requests for Single Reprints: Mary M. McDermott, MD, 675 North St.
Clair, Suite 18-200, Chicago, IL 60611; e-mail, mdm608@northwestern
Current author addresses and author contributions are available at www
Exercise and Peripheral Arterial Disease
3 January 2006 Annals of Internal Medicine Volume 144 • Number 1 19
1. Hirsch AT, Criqui MH, Treat-Jacobson D, Regensteiner JG, Creager MA,
Olin JW, et al. Peripheral arterial disease detection, awareness, and treatment in
primary care. JAMA. 2001;286:1317-24. [PMID: 11560536]
2. McDermott MM, Kerwin DR, Liu K, Martin GJ, O’Brien E, Kaplan H, et
al. Prevalence and significance of unrecognized lower extremity peripheral arterial
disease in general medicine practice. J Gen Intern Med. 2001;16:384-90.
3. McDermott MM, Greenland P, Liu K, Guralnik JM, Criqui MH, Dolan
NC, et al. Leg symptoms in peripheral arterial disease: associated clinical charac-
teristics and functional impairment. JAMA. 2001;286:1599-606. [PMID:
4. McDermott MM, Liu K, Greenland P, Guralnik JM, Criqui MH, Chan C,
et al. Functional decline in peripheral arterial disease: associations with the ankle
brachial index and leg symptoms. JAMA. 2004;292:453-61. [PMID: 15280343]
5. Gardner AW, Poehlman ET. Exercise rehabilitation programs for the treat-
ment of claudication pain. A meta-analysis. JAMA. 1995;274:975-80. [PMID:
6. Falcone RA, Hirsch AT, Regensteiner JG, Treat-Jacobson D, Williams MA,
Hiatt WR, et al. Peripheral arterial disease rehabilitation: a review. J Cardiopulm
Rehabil. 2003;23:170-5. [PMID: 12782899]
7. Regensteiner JG. Exercise rehabilitation for the patient with intermittent clau-
dication: a highly effective yet underutilized treatment. Curr Drug Targets Car-
diovasc Haematol Disord. 2004;4:233-9. [PMID: 15379615]
8. Stewart KJ, Hiatt WR, Regensteiner JG, Hirsch AT. Exercise training for
claudication. N Engl J Med. 2002;347:1941-51. [PMID: 12477945]
9. TransAtlantic Inter-Society Consensus (TASC). Management of peripheral
arterial disease (PAD). J Vasc Surg 2000;31:S83-S86, S118.
10. Olin JW. The clinical evaluation and office-based detection of peripheral
arterial disease. In: Hirsch AT, Olin JW, eds. An Office-Based Approach to the
Diagnosis and Treatment of Peripheral Arterial Disease, I: The Epidemiology and
Practical Detection of Peripheral Arterial Disease. Am J Med. Continuing Edu-
cation Series. Belle Meade, NJ: Excerpta Medica; 1998:10-7.
11. Ogren M, Hedblad B, Isacsson SO, Janzon L, Jungquist G, Lindell SE. Ten
year cerebrovascular morbidity and mortality in 68 year old men with asymptom-
atic carotid stenosis. BMJ. 1995;310:1294-8. [PMID: 7773042]
12. Bernstein EF, Fronek A. Current status of noninvasive tests in the diagnosis
of peripheral arterial disease. Surg Clin North Am. 1982;62:473-87. [PMID:
13. McDermott MM, Criqui MH, Liu K, Guralnik JM, Greenland P, Martin
GJ, et al. Lower ankle/brachial index, as calculated by averaging the dorsalis pedis
and posterior tibial arterial pressures, and association with leg functioning in
peripheral arterial disease. J Vasc Surg. 2000;32:1164-71. [PMID: 11107089]
14. Hiatt WR, Hoag S, Hamman RF. Effect of diagnostic criteria on the prev-
alence of peripheral arterial disease. The San Luis Valley Diabetes Study. Circu-
lation. 1995;91:1472-9. [PMID: 7867189]
15. Criqui MH, Denenberg JO, Bird CE, Fronek A, Klauber MR, Langer RD.
The correlation between symptoms and non-invasive test results in patients re-
ferred for peripheral arterial disease testing. Vasc Med. 1996;1:65-71. [PMID:
16. Guralnik JM, Fried LP, Simonsick EM. The Women’s Health and Aging
Study: Health and Social Characteristics of Older Women with Disability. NIH
publication 95-400. Bethesda, MD: National Institute on Aging; 1995.
17. Altman R, Alarcon G, Appelrouth D, Bloch D, Borenstein D, Brandt K, et
al. The American College of Rheumatology criteria for the classification and
reporting of osteoarthritis of the hip. Arthritis Rheum. 1991;34:505-14. [PMID:
18. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al.
Development of criteria for the classification and reporting of osteoarthritis. Clas-
sification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Com-
mittee of the American Rheumatism Association. Arthritis Rheum. 1986;29:
1039-49. [PMID: 3741515]
19. McDermott MM, Greenland P, Liu K, Guralnik JM, Celic L, Criqui MH,
et al. The ankle brachial index is associated with leg function and physical activ-
ity: the Walking and Leg Circulation Study. Ann Intern Med. 2002;136:873-83.
20. Montgomery PS, Gardner AW. The clinical utility of a six-minute walk test
in peripheral arterial occlusive disease patients. J Am Geriatr Soc. 1998;46:706-
11. [PMID: 9625185]
21. Guyatt GH, Sullivan MJ, Thompson PJ, Fallen EL, Pugsley SO, Taylor
DW, et al. The 6-minute walk: a new measure of exercise capacity in patients
with chronic heart failure. Can Med Assoc J. 1985;132:919-23. [PMID:
22. Guralnik JM, Simonsick EM, Ferrucci L, Glynn RJ, Berkman LF, Blazer
DG, et al. A short physical performance battery assessing lower extremity func-
tion: association with self-reported disability and prediction of mortality and
nursing home admission. J Gerontol. 1994;49:M85-94. [PMID: 8126356]
23. Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower-
extremity function in persons over the age of 70 years as a predictor of subsequent
disability. N Engl J Med. 1995;332:556-61. [PMID: 7838189]
24. Lyness JM, Noel TK, Cox C, King DA, Conwell Y, Caine ED. Screening
for depression in elderly primary care patients. A comparison of the Center for
Epidemiologic Studies Depression Scale and the Geriatric Depression Scale. Arch
Intern Med. 1997;157:449-54. [PMID: 9046897]
25. Richardson MT, Leon AS, Jacobs DR Jr, Ainsworth BE, Serfass R. Ability
of the Caltrac accelerometer to assess daily physical activity levels. J Cardiopulm
Rehabil. 1995;15:107-13. [PMID: 8542513]
26. Miller DJ, Freedson PS, Kline GM. Comparison of activity levels using the
Caltrac accelerometer and five questionnaires. Med Sci Sports Exerc. 1994;26:
376-82. [PMID: 8183104]
27. Sallis JF, Buono MJ, Roby JJ, Carlson D, Nelson JA. The Caltrac acceler-
ometer as a physical activity monitor for school-age children. Med Sci Sports
Exerc. 1990;22:698-703. [PMID: 2233210]
28. Swan PD, Byrnes WC, Haymes EM. Energy expenditure estimates of the
Caltrac accelerometer for running, race walking, and stepping. Br J Sports Med.
1997;31:235-9. [PMID: 9298560]
29. McDermott MM, Liu K, O’Brien E, Guralnik JM, Criqui MH, Martin GJ,
et al. Measuring physical activity in peripheral arterial disease: a comparison of
two physical activity questionnaires with an accelerometer. Angiology. 2000;51:
91-100. [PMID: 10701716]
30. Laird NM, Ware JH. Random-effects models for longitudinal data. Biomet-
rics. 1982;38:963-74. [PMID: 7168798]
31. Leng GC, Fowler B, Ernst E. Exercise for intermittent claudication. Co-
chrane Database Syst. Rev. 2000;2:CD000990.
32. Hirsch AT, Halverson SL, Treat-Jacobson D, Hotvedt PS, Lunzer MM,
Krook S, et al. The Minnesota Regional Peripheral Arterial Disease Screening
Program: toward a definition of community standards of care. Vasc Med. 2001;
6:87-96. [PMID: 11530970]
33. McDermott MM, Hahn EA, Greenland P, Cella D, Ockene JK, Brogan D,
et al. Atherosclerotic risk factor reduction in peripheral arterial disease: results of
a national physician survey. J Gen Intern Med. 2002;17:895-904. [PMID:
34. McDermott MM, Mehta S, Ahn H, Greenland P. Atherosclerotic risk
factors are less intensively treated in patients with peripheral arterial disease than
in patients with coronary artery disease. J Gen Intern Med. 1997;12:209-15.
35. Regensteiner JG, Meyer TJ, Krupski WC, Cranford LS, Hiatt WR. Hos-
pital vs home-based exercise rehabilitation for patients with peripheral arterial
occlusive disease. Angiology. 1997;48:291-300. [PMID: 9112877]
36. Rucker-Whitaker C, Greenland P, Liu K, Chan C, Guralnik JM, Criqui
MH, et al. Peripheral arterial disease in African Americans: clinical characteristics,
leg symptoms, and lower extremity functioning. J Am Geriatr Soc. 2004;52:922-
30. [PMID: 15161456]
37. Selvin E, Erlinger TP. Prevalence of and risk factors for peripheral arterial
disease in the United States: results from the National Health and Nutrition
Examination Survey, 1999-2000. Circulation. 2004;110:738-43. [PMID:
38. Little RJ. Modeling the drop-out mechanism in repeated-measures studies.
Journal of the American Statistical Association. 1995;90:1112-21.
39. Fitzmaurice GM, Laird NM, Shneyer L. An alternative parameterization of
the general linear mixture model for longitudinal data with non-ignorable drop-
outs. Stat Med. 2001;20:1009-21. [PMID: 11276032]
Exercise and Peripheral Arterial Disease
20 3 January 2006 Annals of Internal Medicine Volume 144 • Number 1
Current Author Addresses: Drs. McDermott and Martin: 675 North Download full-text
St. Clair, Suite 18-200, Chicago, IL 60611.
Drs. Liu, Greenland, and Tian and Mr. Tan: 680 N. Lake Shore Drive,
Suite 1102, Chicago, IL 60611.
Dr. Ferrucci: National Institute on Aging, Clinical Research Branch,
Harbor Hospital, 3001 South Hanover Street, Fifth Floor, Baltimore,
Dr. Criqui: University of California at San Diego, Family & Preventive
Medicine, 9500 Gilman Drive, La Jolla, CA 92093.
Dr. Guralnik: National Institute on Aging, Gateway Building, Room
3-C309, 7201 Wisconsin Avenue, Bethesda, MD 20892.
Dr. Schneider: Evanston Northwestern Hospital, Department of Sur-
gery, 2650 Ridge Avenue, Burch 100, Evanston, IL 60201.
Dr. Pearce: 201 East Huron, Suite 10-105, Chicago, IL 60611.
Author Contributions: Conception and design: M.M. McDermott, K.
Liu, L. Ferrucci, P. Greenland, J.M. Guralnik, J.R. Schneider, G.J. Mar-
Analysis and interpretation of the data: M.M. McDermott, K. Liu, L.
Ferrucci, M.H. Criqui, P. Greenland, J.M. Guralnik, L. Tian, J. Tan,
Drafting of the article: M.M. McDermott, M.H. Criqui.
Critical revision of the article for important intellectual content: M.M.
McDermott, L. Ferrucci, M.H. Criqui, P. Greenland, J.M. Guralnik,
J.R. Schneider, G.J. Martin.
Final approval of the article: M.M. McDermott, K. Liu, M.H. Criqui, P.
Greenland, J.M. Guralnik, L. Tian, J.R. Schneider, W.H. Pearce, J. Tan,
Provision of study materials or patients: J.R. Schneider, W.H. Pearce,
Statistical expertise: K. Liu, L. Tian.
Obtaining of funding: M.M. McDermott, K. Liu, M.H. Criqui, G.J.
Administrative, technical, or logistic support: W.H. Pearce, J. Tan.
Collection and assembly of data: M.M. McDermott, W.H. Pearce.
Statistical Analyses and Strategies for Handling Missing
The associations between longitudinal changes in functional
measures and exercise frequency were evaluated by using mixed-
effects models in which a subject-specific random effect is used to
account for the potential correlations among successive annual
differences in each functional measure of the same participant.
Dependent variables in each mixed-effects regression analysis
were the successive annual differences in each functional measure.
For example, for the 6-minute walk, the dependent variable was
defined as the successive differences in 6-minute walking dis-
tances (that is, the difference in distances measured at baseline
and at the first follow-up visit, the difference in distances mea-
sured at the first and second annual follow-up visits, and the
difference in distances measured at the second and third annual
follow-up visits). Statistically valid inference is guaranteed in this
initial mixed-effects regression analysis, provided that missing
data caused by patient dropout are unrelated to unobserved data
(that is, any missing data are missing at random). As a safeguard
against violations to this assumption, we used a repeated-mea-
sures pattern-mixture analysis of covariance to repeat the fully
adjusted comparisons (38, 39). In this model, patients may be
classified into possible patterns of missing data. Because data were
analyzed by using successive differences, there were 2 observed
patterns of missing differences in our analyses: missing the suc-
cessive differences from the second follow-up visit to the third
and missing the successive differences from second follow-up visit
to the third and from the first follow-up visit to the second. The
different patterns of missing data were included as 2 binary in-
dicator covariates (centered about their means). By including pat-
terns of missing data in analyses as centered covariates and by
using the adjusted least-square means to find the averages of these
patterns, one can obtain an unbiased estimate of the marginal
means, adjusted for covariates (39).
Annals of Internal Medicine
W-2 3 January 2006 Annals of Internal Medicine Volume 144 • Number 1