Upper- vs lower-limb aerobic exercise

rehabilitation in patients with symptomatic

peripheral arterial disease: A randomized
controlled trial
Irena Zwierska, MSc,a Richard D. Walker, PhD,b Sohail A. Choksy, FRCS,b Jonathan S. Male, BSc,a
A. Graham Pockley, PhD,b and John M. Saxton, PhD,a Sheffield, United Kingdom

Objectives: To investigate the effects of a 24-week program of upper- and lower-limb aerobic exercise training on walking
performance in patients with symptomatic peripheral arterial disease (PAD) and to study the mechanisms that could
influence symptomatic improvement.
Methods: After approval from the North Sheffield Local Research Ethics Committee, 104 patients (median age, 69 years;
range, 50 to 85 years) with stable PAD were randomized into an upper- or lower-limb aerobic exercise training group
(UL-Ex or LL-Ex), or to a nonexercise training control group. Training was performed twice weekly for 24 weeks at
equivalent relative exercise intensities. An incremental arm- and leg-crank test (ACT and LCT) to maximum exercise
tolerance was performed before and at 6, 12, 18, and 24 weeks of the intervention to determine peak oxygen consumption
(V̇O2). Walking performance, defined as the claudicating distance (CD) and maximum walking distance (MWD) achieved
before intolerable claudication pain, was assessed at the same time points by using a shuttle-walk protocol. Peak blood
lactate concentration, Borg ratings of perceived exertion (RPE) and pain category ratio (CR-10) were recorded during all
assessments.
Results: CD and MWD increased over time (P < .001) in both training groups. At 24 weeks, CD had improved by 51%
and 57%, and MWD had improved by 29% and 31% (all P < .001) in the UL-Ex and LL-Ex groups, respectively. An
increase in peak heart rate at MWD in the UL-Ex group (109 ⴞ 4 vs 115 ⴞ 4 beats/min; P < .01) and LL-Ex group
(107 ⴞ 3 vs 118 ⴞ 3 beats/min; P ⴝ .01) was accompanied by an increase in the amount of pain experienced in both
groups (P < .05), suggesting that exercising patients could tolerate a higher level of cardiovascular stress and an increased
intensity of claudication pain before test termination after training. Patients assigned to exercise training also showed an
increase in LCT peak V̇O2 at the 24-week time point in relation to baseline (P < .01) and control patients (P < .01),
whereas ACT peak V̇O2 was only improved in the UL-Ex group (P < .05).
Conclusions: Our results suggest that a combination of physiologic adaptations and improved exercise pain tolerance
account for the improvement in walking performance achieved through upper-limb aerobic exercise training in patients
with PAD. Furthermore, that both arm- and leg-crank training could be useful exercise training modalities for improving
cardiovascular function, walking performance, and exercise pain tolerance in patients with symptomatic PAD. ( J Vasc
Surg 2005;42:1122–30.)

Despite a considerable body of evidence supporting the
positive effects of lower-limb exercise rehabilitation, overall
physical activity levels are lower in patients with symptom-
atic peripheral arterial disease (PAD) than in the non-PAD
population.1 Furthermore, a significant proportion of pa-
tients refrain from engaging in walking exercise rehabilita-
tion,2 and evidence suggests that self-reported walking
ability declines at the rate of 9.2 yards (8.4 meters) per
year.3
From the Centre for Sport and Exercise Science, Sheffield Hallam Univer-
sity;a and Division of Clinical Sciences (North), University of Sheffield,
Northern General Hospital.b
This research was supported by a grant from the British Heart Foundation
(Grant PG/2000042).
Competition of interest: none.
Correspondence: John M. Saxton, PhD, Centre for Sport and Exercise
Science, Sheffield Hallam University, Collegiate Crescent Campus, Shef-
field, S10 2BP, UK (e-mail: j.m.saxton@shu.ac.uk).
0741-5214/$30.00
Copyright © 2005 by The Society for Vascular Surgery.
doi:10.1016/j.jvs.2005.08.021
1122
The degree of ischemic pain evoked by walking might
deter some patients from engaging in lower-limb exercise
rehabilitation, and the reduced levels of physical activity
could potentially contribute to subsequent disability4 and
the enhanced risk of cardiovascular morbidity and mortality
that is observed in this patient group.5 Alternative exercise
modalities that avoid the ischemic pain that is associated
with lower-limb exercise might help some patients to over-
come the barriers to conventional forms of exercise reha-
bilitation (ie, walking exercise) and increase their level of
enthusiasm for exercise rehabilitation.
As symptomatic arterial insufficiency of the upper limbs
is over 20 times less frequent than lower-limb arterial
occlusion in patients with PAD,6,7 patients are less likely to
experience ischemic pain during arm exercise. Preliminary
evidence from our laboratory has shown that upper-limb
aerobic exercise training is well-tolerated in patients with
symptomatic PAD and that it can induce improvements in
cardiovascular function and walking ability.8 However, the
effects of upper-limb aerobic exercise training on walking
JOURNAL OF VASCULAR SURGERY
Volume 42, Number 6 Zwierska et al 1123

Table I. Demographics of the three study groups

Leg training Arm training Control

Variable (n ⫽ 37) (n ⫽ 34) (n ⫽ 33) P

Median age (range) 69 (50-85) 66 (54-84) 72 (56-84) NS*

Body weight (kg) 76.3 ⫾ 2.2 80.8 ⫾ 1.9 78.3 ⫾ 3.2 NS*
Height (m) 1.69 ⫾ 0.01 1.68 ⫾ 0.01 1.67 ⫾ 0.01 NS*
Body mass index 26.6 ⫾ 0.6 28.6 ⫾ 0.6 27.8 ⫾ 1.0 NS*
Sex
Male (%) 81 79 73 NS†
Female (%) 19 21 27
Resting ABPI 0.64 ⫾ 0.03 0.65 ⫾ 0.03 0.69 ⫾ 0.03 NS†
Duration of claudication (months) 59 50 55 NS*
Angina (%) 14 18 24 NS†
Previous MI (%) 3 9 12 NS†
Previous stroke (%) 19 18 15 NS†
Diabetes mellitus (%) 8 18 27 .05†
Smoking status (%)
Current 38 24 33 NS†
Previous 54 73 61
Never smoked 8 3 6
ABPI, Ankle-brachial pressure index; MI, myocardial infarction; NS, no significant difference between the groups.
ABPI data are presented for the most symptomatic limb. Patients with diabetes mellitus were taking oral hypoglycemics or insulin. Unless otherwise stated,
values are means ⫾ SE.
*Analysis of variance.
␹ test.
† 2

performance and the mechanisms underpinning symptom-
atic improvement have not been systematically investigated
in a large-scale, randomized controlled trial.
This study investigated the effects of a 24-week pro-
gram of upper- and comparable lower-limb aerobic exercise
training on walking performance in a large cohort of pa-
tients with PAD. We also investigated the possible mecha-
nisms that could influence symptomatic improvement after
these exercise rehabilitation regimens.
METHODS
Patients. A total of 104 patients (median age, 69
years; range, 50 to 85 years) with stable intermittent clau-
dication were recruited from the Sheffield Vascular Insti-
tute at the Northern General Hospital, Sheffield, United
Kingdom. Approval was obtained from the North Sheffield
Local Research Ethics Committee, and all patients pro-
vided informed consent before entering the study.
The clinical diagnosis of PAD was established from the
patient’s history and a physical examination and was con-
firmed by the Doppler assessment of ankle-brachial pressure
index (ABPI), a noninvasive, reliable measure of lower-
extremity hemodynamics,9 in accordance with current UK
medical practice.
In two-thirds of the patients, claudication was due to
superficial femoral artery disease as determined by angio-
gram, duplex scanning, or examination of pulses. The rest
of the patients had tibial vessel disease. The absence of
significant upper-extremity arterial disease was confirmed
by pulse examination and the observation that brachial
artery systolic pressures were within 10 mm Hg of each
other.9
Patients experiencing symptoms of intermittent claudi-
cation for ⬍12 months or reporting a significant change in
walking ability within this time period were considered to
have unstable disease and were, as a consequence, ex-
cluded. Patients were also excluded if they exhibited fea-
tures of critical ischemia, had undergone a revascularization
procedure ⱕ12 months, or if the initial assessment estab-
lished that they had severe arthritis (ie, were unable to walk
unaided or perform either upper- or lower-limb cranking
exercise due to joint pain), severe lumbar spine disease, or
unstable cardiorespiratory conditions (ie, unstable blood
pressure, recent electrocardiographic changes or acute
myocardial infarction, unstable angina, third-degree heart
block, acute congestive heart failure and severe respiratory
conditions).
No patients were taking medication for claudication
during the course of this study, but patients on long-term
medication continued with their treatment. Most patients
were taking statins and aspirin in accordance with guide-
lines for patients with intermittent claudication. Some pa-
tients were also taking ␤-blockers, angiotensin-converting
enzyme inhibitors, calcium-channel blockers, nitrates, di-
uretics, or warfarin. Demographic data for the patient
cohort are summarized in Table I.
Outcome measures. Outcome measures were assessed
at baseline and at weeks 6, 12, 18, and 24 in all patients. In
the exercise groups, assessment of outcome variables re-
placed the training sessions in these weeks. During an initial
consultation session lasting approximately 1.5 hours, a full
explanation of the testing procedures was followed by a dry
run of each assessment under test conditions, ensuring that
patients were fully accustomed to all assessment protocols
before the study commenced. Patients were instructed not
to perform any vigorous exercise in the 24 hours before
assessment sessions and to abstain from caffeinated bever-
ages and nicotine intake for at least 2 hours before each test.

1124 Zwierska et al
Assessment of outcome measures was conducted by the
same two members of staff. Procedures were checked for
consistency at random by the lead investigator at Sheffield
Hallam University (J. S.), who was blinded to group assign-
ment.
Walking performance. Walking performance was the
primary outcome variable, and this was assessed with an
incremental shuttle-walk test,10 which has good test-retest
reproducibility and can be used to assess patients with
wide-ranging functional ability. We have previously re-
ported an intraclass correlation coefficient of 0.87 for max-
imum walking distance (MWD) between repeated incre-
mental shuttle-walk tests in patients with symptomatic
PAD, which was similar to that observed for repeated
standardized treadmill testing in the same patient group.10
Briefly, the test involves the patient walking back and
forth between two cones placed 10 meters apart on a flat
floor, with walking speed being controlled by bleeps re-
corded onto an audiotape. The initial walking speed for the
test is 3 km/h, and there is a step-increase in walking speed
of 0.5 km/h every minute. Patients start walking on hear-
ing the first bleep and aim to reach the opposite cone by the
next, generally achieving the correct walking pace within
two or three repetitions (20 to 30 meters). Distance to the
onset of claudication pain (claudicating distance [CD]) and
the MWD before test termination due to intolerable clau-
dication pain were recorded. Measurements of the CD and
MWD began at the first bleep, when the patient started
walking. The MWD was the primary outcome measure of
the study.
Assessment of upper- and lower-limb peak aerobic
power. Peak upper- and lower-limb aerobic power was
assessed with a calibrated electronically braked cycle er-
gometer (Gould, Bilthoven, The Netherlands), which was
raised on a platform for upper-limb testing. Patients per-
formed an incremental arm-cranking test (ACT) and an
incremental leg-cranking test (LCT) to maximum exercise
tolerance on separate occasions, with the order of the tests
being randomized. For the ACT, the mid-point of the
sprocket was set at shoulder height; and for the LCT, the
seat height was adjusted to allow slight knee flexion at
bottom dead center.
The ACT and LCT were performed at a similar time of
the day and were discontinuous in nature, comprising
3-minute bouts of constant work at a cranking rate of
50 revs/min interpolated with 2-minute rest intervals, to
maximum exercise tolerance. The initial intensity was 9 W
for both tests, with power output being increased by 7 and
14 W per increment in the ACT and LCT, respectively.
Before each assessment, resting blood pressure and
heart rate were recorded, and a small (25 ␮L) blood sample
was taken from a fingertip for blood lactate analysis (1500
Sport, YSI, Inc, Yellow Springs, Ohio). During each work
increment, pulmonary gas exchange variables, including oxy-
gen consumption V̇O2 were measured breath-by-breath (Pul-
moLab EX670, Morgan Medical, Kent, UK) and averaged
from the last 30 seconds of each work increment. Heart rate
was continuously monitored using a 6-lead electrocardio-
JOURNAL OF VASCULAR SURGERY
December 2005
graph (CASE 15, Marquette/G.E., Milwaukee, Wis). Per-
ceived exertion using the Borg Ratings of Perceived Exer-
tion (RPE) 15-graded scale11 and the degree of pain
experienced using the Borg Category Ratio-10 (CR-10)
scale11 were also recorded during the last 30 seconds of
each work increment, and additional 25 ␮L of fingertip
blood samples for lactate analysis were obtained at the end
of each work increment.
Physical activity levels. Self-reported physical activity
levels were assessed with the PAD-Physical Activity Recall
(PAR) questionnaire in all groups weekly12 (data from
control patients were collected by telephone). Patients
were asked to evaluate the time they had spent over the
previous week performing work activities, household tasks
and yard work, and leisure activities, classified as heavy,
moderate, light, and very light determined by metabolic
equivalents (METs). One MET is equivalent to a resting
V̇O2 of 3.5 mL · kg⫺1 · min⫺1. The number of hours of
physical activity per week in each of the four categories was
summed to give global physical activity status and the data
were recorded in MET hours (MET-h) per week (MET-
h/wk: hours per week multiplied by the MET value of the
activity).
Supervised exercise training. Patients were randomly
assigned to an upper-limb aerobic exercise-training group
(UL-Ex), a lower-limb aerobic exercise-training group
(LL-Ex), or a control group (no exercise training). Super-
vised exercise training was undertaken twice a week for 24
weeks at equivalent limb-specific relative exercise intensi-
ties. Briefly, exercise training intensities were based on the
penultimate power output achieved in the respective arm-
or leg-cranking assessment. Use of the penultimate power
output enabled training intensity to be balanced at 85% to
90% of the limb-specific peak V̇O2 between the UL-Ex and
LL-Ex groups, thus ensuring a similar cardiovascular stim-
ulus. In addition to power output, monitoring of heart rate
and perceived exertion (Borg RPE 15-graded scale11) en-
sured that the correct training intensities were used in the
supervised exercise sessions, with most patients exercising
between the RPE ratings of 13 (“somewhat hard”) to 16
(“hard to very hard”).
For each of the supervised training sessions, patients
exercised in cycles of 2 minutes exercise at a crank rate of
50 rev/min, followed by 2 minutes rest, for a total exercise
time of 20 minutes in a 40-minute session. For the upper-
limb aerobic exercise training, the arm-crank ergometer
was placed on a table in front of the seated patient with the
mid-point of the sprocket set at shoulder height. For lower-
limb aerobic exercise training, the seat height was adjusted
to allow slight knee flexion at bottom dead centre. Up to
eight patients exercised together in the same session.
The incremental ACTs and LCTs in weeks 6, 12, and
18 were used to adjust the respective upper- and lower-limb
training intensities on an individual basis (where appropriate)
to ensure progression of the exercise stimulus as patients
adapted to the training load. Patients in the control group
were given lifestyle advice, including encouragement to
JOURNAL OF VASCULAR SURGERY
Volume 42, Number 6
undertake regular exercise, but did not undertake any su-
pervised exercise.
Statistical analysis. Intention-to-treat analysis was
used to compare patients in the groups to which they were
originally randomly assigned, with data being carried over
from previous visits in cases of patient withdrawal. Depen-
dent variables were first tested for normal distribution using
the Kolmogorov-Smirnov goodness of fit test. With the
exceptions of peak power output, RPE, CR-10, and CD, all
other variables were normally distributed. A mixed factorial
analysis of variance (ANOVA) was used to test for changes
in normally distributed variables over time within groups
and for differences between groups at the same time points
(peak power output and CD were normalized using loga-
rithmic transformation before analysis). Where significance
was indicated, the mixed factorial ANOVA analysis was
followed up with one-way ANOVAs within and between
groups and post hoc comparisons.
As the RPE and CR-10 responses could not be normal-
ized using data transformation techniques, they were ana-
lyzed using the Kruskal-Wallis test, with group differences
at each time point being identified using the Mann-Whitney
U test and changes over time within groups analyzed using
the Friedman and Wilcoxon signed ranks test. The magni-
tude of the changes in CD and MWD between baseline and
the 24-week time point between groups was compared
using one-way ANOVA. Statistical significance was estab-
lished at P ⬍ .05, and data were analyzed using SPSS
statistical software (SPSS UK Ltd, Woking, UK). The
results are expressed as means ⫾ SE unless otherwise stated.
RESULTS
Of the 104 patients recruited, 4 withdrew from the
UL-Ex group, 5 withdrew from the LL-Ex group, and 1
withdrew from the control group, primarily for medical
reasons. Three patients withdrew in between their initial
assessment and their first 6-week reassessment. One was
diagnosed with bowel cancer, another became a full-time
carer for his wife, and a third found the assessments too
stressful. Five patients withdrew in between weeks 6 and 12
of the intervention due to gout, heart attack, pneumonia
(which resulted in death), and other medical conditions.
Two further patients withdrew between weeks 12 and 18 of
the intervention, one because of lower-limb ulcers and the
other because he broke a bone in his foot. In the 94 patients
who completed the 24-week intervention (30 in UL-Ex, 32
in LL-Ex; 32 in control), compliance with the twice-weekly
training sessions was 99%.
MWD increased over time in both training groups (P ⬍
.001) and was significantly higher than in the controls at 12
to 24 weeks of the intervention (Fig 1). In the control
subjects, MWD was unchanged from baseline over the 24
weeks of the study. CD was also increased from baseline at
the 6-week time point in the exercising groups (P ⬍ .05)
and showed further increases up to 24 weeks of the inter-
vention (Table II). The magnitude of improvement in CD
and MWD over the 24 weeks of the study was comparable
between the exercising groups (Fig 1; Table II). At 24
Zwierska et al 1125
Leg training group Arm training group Control group
††
†† **
450 ††
Maximum walking distance (m)
**
†† **
*
400
††
350 †† ††
†† *
* *
300
250
0
0 6 12 18 24
Time (weeks)
Fig 1. Maximum walking distance at 6-weekly intervals in all
groups. Data are presented as means with error bars representing
SE. Statistical significance: *P ⬍ .05; **P ⬍ .01: indicates signifi-
cantly higher than control patients; ††P ⬍ .01: indicates signifi-
cantly higher than baseline.
weeks, CD had improved by 51% and 57% and MWD had
improved by 29% and 31% in the UL-Ex and LL-Ex
groups, respectively. These changes in CD and MWD in
the exercising groups were significantly different from the
control group (all P ⬍ .001).
Heart rate, RPE, CR-10, and blood lactate responses
were measured during the incremental walking test to
assess the consistency of effort and level of pain experienced
in the walking assessments at the different time-points.
Heart rate at CD was unchanged in all groups at the end of
the intervention period. However, heart rate at MWD was
increased from baseline in the UL-Ex group (P ⬍ .01) and
LL-Ex group (P ⫽ .01) as well as in relation to the controls
at the 24-week time point in both exercise groups (P ⬍
.01), indicating a higher level of cardiovascular stress at
MWD (Table III).
A higher RPE at MWD was also observed at the 24-
week time point in the UL-Ex group (P ⬍ .01), although
this was not significantly different from the controls. An
increase in the amount of pain experienced at MWD was
observed in both exercising groups (P ⬍ .05) (Table III).
The blood lactate concentration at MWD was higher
than baseline and greater than in the control group at the
24-week time point only in the UL-Ex group (P ⬍ .05).
Resting blood lactate levels before the walking assessment
were 1.21 ⫾ 0.13, 1.09 ⫾ .08, and 1.29 ⫾ 0.10 mM in the
LL-Ex, UL-Ex, and control groups, respectively, at base-
line, corresponding to resting levels of 1.11 ⫾ .08, 1.28 ⫾
0.12, and 1.17 ⫾ .09 mM, respectively, at the 24-week
time point.
Peak power outputs for the ACT and the LCT were
increased from baseline (P ⬍ .01) and were higher than in
the control patients (P ⬍ .01) in both exercise groups at the
24-week time point, indicating an improvement in peak
 JOURNAL OF VASCULAR SURGERY
1126 Zwierska et al December 2005

Table II. Claudication distance at 6-week intervals in all groups

Weeks of exercise Baseline 6 12 18 24

† †
Leg-training group (m) 123 (20-568) 140 (23-394)* 156 (20-692) 161 (26-465) 168 (34-579)†
Arm-training group (m) 136 (25-430) 161 (38-480)* 175 (39-430)† 177 (29-438)† 187 (42-508)†
Control group (m) 135 (28-408) 138 (24-480) 123 (30-463) 129 (22-397) 134 (25-333)
Data were not normally distributed and are presented as medians with range in parentheses.
Statistical significance: *P ⬍ .05; †P ⬍ .01: indicates significantly higher than baseline.

Table III. Heart rate, blood lactate, Borg Category Ratio-10 pain levels and Ratings of Perceived Exertion responses
during the shuttle-walk test, leg-cranking test, and arm-cranking test

Leg-training group Arm-training group Control group

Shuttle-walk test Baseline 24 weeks Baseline 24 weeks Baseline 24 weeks

HR at CD (beats/min) 92 ⫾ 3 92 ⫾ 2 92 ⫾ 2 94 ⫾ 2 88 ⫾ 3 88 ⫾ 3
HR at MWD
(beats/min) 107 ⫾ 3 118 ⫾ 3†ⴱⴱ 109 ⫾ 4 115 ⫾ 4†§ 100 ⫾ 3 99 ⫾ 3
Blood lactate at MWD
(mM) 1.87 ⫾ 0.09 2.18 ⫾ 0.12* 1.92 ⫾ 0.14 2.33 ⫾ 0.16†‡ 1.67 ⫾ 0.08 1.70 ⫾ 0.10
CR-10 at CD 1.0 (0.3-3.0) 1.0 (0.3-3.0) 1.5 (0.3-3.0) 1.5 (0.3-3.0) 1.5 (0.5-3.0) 1.5 (0.3-5.0)
CR-10 at MWD 6.0 (3.0-11.0) 7.0 (2.0-11.0)† 6.0 (2.5-11.0) 7.0 (2.5-11.0)§ 5.5 (2.5-11.0) 6.0 (1.0-10.0)
§
RPE at MWD 14.4 (9.0-20.0) 15.2 (7.0-19.0) 13.6 (7.0-20.0) 15.3 (6.0-20.0) 14.1 (7.0-19.0) 13.8 (6.0-20.0)
Peak values for LCT and ACT
Heart rate
LCT 114 ⫾ 4 124 ⫾ 3†§ 115 ⫾ 4 119 ⫾ 4ⴱ 109 ⫾ 4 104 ⫾ 4
ACT 120 ⫾ 3 122 ⫾ 3† 117 ⫾ 4 124 ⫾ 4† 105 ⫾ 4 100 ⫾ 4
Lactate (mM)
LCT 3.26 ⫾ 0.15 3.95 ⫾ 0.15†§ 3.21 ⫾ 0.18 3.87 ⫾ 0.18†§ 2.71 ⫾ 0.17 2.59 ⫾ 0.16
ACT 3.54 ⫾ 0.20 3.73 ⫾ 0.17† 3.27 ⫾ 0.20 4.28 ⫾ 0.22†§ 3.07 ⫾ 0.18 2.61 ⫾ 0.17
Power (W)
LCT 72 (38-126) 95 (53-169)†§ 69 (9-140) 84 (24-155)†§ 60 (24-105) 60 (9-96)
ACT 48 (24-82) 54 (24-89)†§ 46 (17-82) 61 (24-118)†§ 40 (17-60) 37 (9-67)
RPE
LCT 17.2 (13.0-20.0) 16.3 (12.0-20.0) 16.7 (12.0-20.0) 15.8 (7.0-20.0) 16.7 (13.0-20.0) 16.2 (12.0-20.0)
ACT 17.2 (13.0-20.0) 16.3 (10.0-20.0) 16.1 (10.0-20.0) 16.0 (8.0-20.0) 17.0 (13.0-20.0) 16.8 (11.0-20.0)
CR-10
LCT 7.5 (0.0-11.0) 7.0 (0.0-10.0) 6.5 (3.0-11.0) 6.5 (3.0-11.0) 6.0 (0.0-11.0) 7.0 (3.0-10.0)
ACT 5.0 (0.0-11.0) 6.5 (0.0-11.0)‡ 3.5 (0.0-10.0) 5.5 (0.0-11.0)§ 5.0 (0.0-11.0) 6.0 (0.0-10.0)‡
HR, Heart rate; CD, claudicating distance; MWD, maximum walking distance; CR-10, Borg Category Ratio pain level; RPE, Borg Ratings of Perceived
Exertion; LCT, leg-cranking test; ACT, arm-cranking test.
Heart rate and blood lactate are reported as means ⫾ SE; Power, CR-10 pain levels and RPE responses are shown as median (range).
*P ⬍ .05 and †P ⬍ .01 indicate significantly higher than control patients at 24 weeks.
‡
P ⬍ .05, §P ⬍ .01, **P ⫽ .01, indicate significantly higher than baseline at 24 weeks.

power output for both trained and untrained skeletal mus-
cle groups in exercised individuals (Table III). Patients in
the UL-Ex and LL-Ex groups showed an increase in LCT
peak V̇O2 at the 24-week time point in relation to baseline
measures (P ⬍ .01) and control patients (P ⬍ .01) (Fig 2),
whereas ACT peak V̇O2 was only improved with respect to
baseline in patients in the UL-Ex group (P ⬍ .05) (Fig 2).
Nevertheless, upper-limb peak aerobic power in both exer-
cising groups was significantly greater than in the control
patients at weeks 6 to 24 of the intervention period (Fig 2).
The increase in LCT peak V̇O2 was accompanied by an
increase in peak heart rate in the LL-Ex group (P ⬍ .01)
and by an increase in peak blood lactate concentration from
baseline in both exercise training groups (P ⬍ .01) (Table III)
at the end of the intervention period.
These peak responses were all greater than those re-
corded for the controls. Neither the peak RPE responses for
the ACT and LCT, nor peak CR-10 pain level for the LCT
were increased in either training group at the end of the
intervention period, nor was there any change from base-
line in control patients. However, peak CR-10 pain level for
the ACT was increased from baseline in all groups at the
24-week time point (P ⱖ .045).
Resting ABPI was unchanged from baseline in all
groups over the 24-week intervention period. At baseline,
resting ABPI was 0.64 ⫾ .03, 0.65 ⫾ .03, and 0.69 ⫾ .03
in the LL-Ex, UL-Ex, and control groups, respectively.
After the 24-week intervention period, the corresponding
values were 0.66 ⫾ .03, 0.68 ⫾ .04, and 0.68 ⫾ .03 for the
LL-Ex, UL-Ex, and control groups, respectively.
JOURNAL OF VASCULAR SURGERY
Volume 42, Number 6 Zwierska et al 1127

LCT tion regimens. Both exercise regimens evoked a significant

†† †† improvement in walking performance (MWD) that was
** **
1.40
Peak oxygen uptake (l.min-1)

†† apparent after 12 weeks of twice-weekly supervised exercise

** training. The change in walking performance was accom-
††
1.30 panied by improvements in lower-limb aerobic power and
††
peak oxygen consumption in both the UL-Ex and LL-Ex
1.20
†† † †† ** groups. Compliance with the exercise training sessions was
* * excellent (99%) in the 94 patients who completed the
1.10
intervention, and only 10 patients had to withdraw from
1.00 the study, primarily for medical reasons. The excellent
compliance and low patient drop-out in this study is attrib-
0.00 uted to the flexible approach taken when arranging exercise
0 6 12 18 24 training sessions and the regular contact maintained with
all patients, including weekly telephone contact with the
controls.
ACT The time-course of improvement in CD and MWD was
†† †† † very similar in the exercising groups, and a comparable
1.25 ** ** ** magnitude of change occurred in CD and MWD over the
Peak oxygen uptake (l.min-1)

**
1.15
1.05
**
0.95
0.85
0.00
0 6
Time (weeks)
Leg training group
Fig 2. Changes in peak oxygen uptake (V̇O2) for the leg-crank-
ing test (LCT) and arm-cranking test (ACT) at 6-week intervals
for all groups. Data are presented as means with error bars repre-
senting SE. Statistical significance: *P ⬍ .05; **P ⬍ .01: indicates
significantly higher than control patients; †P ⬍ .05; ††P ⬍ .01:
indicates significantly higher than baseline.
Self-reported physical activity level, as assessed by the
PAD-PAR questionnaire, increased by approximately 10
MET-h/wk from the baseline level in both exercise train-
ing groups between weeks 6 and 24 of the intervention
(P ⬍ .01) (Table IV). This was mainly due to an increase in
leisure time activity resulting from participation in the
supervised exercise sessions, as physical activity status for
work activities and household tasks remained unchanged.
An increase in self-reported leisure time physical activity
level was also observed for the control subjects at the
12-week time point (P ⬍ .05) (Table IV).
DISCUSSION
The primary aim of this study was to compare the
effects of a 24-week program of upper- and lower-limb
aerobic exercise training on walking performance in a large
cohort of patients with PAD. A secondary aim was to
investigate the possible mechanisms that could influence
symptomatic improvement after these exercise rehabilita-
24 weeks of the study. At 24 weeks, CD had improved by
51% and 57%, and MWD had improved by 29% and 31% in
the UL-Ex and LL-Ex groups, respectively. These changes
**
* * in CD and MWD are lower than some reported in the
literature after randomized, controlled walking exercise
interventions, in the range of 28% to 125% for CD and 51%
to 210% for MWD.13
Although the exercise stimulus might be greater in
weight-bearing walking exercise programs than in weight-
12 18 24 supported arm- and leg-cranking exercise, the use of
constant-pace treadmill testing to assess walking perfor-
mance in previous studies14 might also have had some
Arm training group Control group bearing on the magnitudes of the previously reported im-
provements. Constant-pace walking tests are considered to
be less rigorous at assessing walking distances than incre-
mental tests due to ceiling effects, higher coefficients of
variation, and because improvements in walking perfor-
mance are commonly observed when control patients per-
form such repeated tests (placebo response).12 Placebo
responses of 25% to 36% have been reported for control
patients when they have been assessed using constant-pace
tests, an effect that is not usually observed when incremen-
tal tests are used,12 and this is in accordance with our
findings. It has therefore been acknowledged that much
smaller improvements measured using incremental tests
can be interpreted as being of clinical relevance.12
We used an incremental shuttle-walk protocol to assess
walking performance in our cohort, as this exhibits good
test-retest reproducibility and might provide a better reflec-
tion of changes in everyday walking ability than incline
treadmill assessments.10 An improvement in MWD of 30%
on a graded (incremental) test is considered clinically sig-
nificant15 and on the shuttle-walk test equates to a distance
of 100 meters.
Changes in walking performance after programs of exer-
cise rehabilitation might be explained by a number of physio-
logic mechanisms. These include localized metabolic16-18 and
blood flow adaptations18-21 within the ischemic tissues as well
as central cardiovascular adaptations.8,18,21 Changes in nitric
oxide metabolism22 and endothelial vasoreactivity,21 in con-

1128 Zwierska et al
Table IV. Physical activity status, as assessed by the peripheral arterial disease-physical activity recall questionnaire
Group Baseline 6 weeks
Work activities
Leg training 82 ⫾ 16 82 ⫾ 16
Arm training 86 ⫾ 23 83 ⫾ 24
Control 78 ⫾ 15 80 ⫾ 16
Household work
Leg training 76 ⫾ 5 76 ⫾ 6
Arm training 78 ⫾ 5 78 ⫾ 5
Control 75 ⫾ 8 76 ⫾ 8
Leisure activities
Leg training 66 ⫾ 5 75 ⫾ 4*
Arm training 65 ⫾ 4 73 ⫾ 3*
Control 64 ⫾ 5 68 ⫾ 5
Data (metabolic equivalents · hour · week⫺1) are presented as means ⫾ SE.
*P ⬍ .01; †P ⬍ .05; indicates significantly higher than baseline.
junction with altered blood rheology,23 could also enhance
microcirculatory blood flow to exercising skeletal muscles.
In addition, improvements in maximum exercise tolerance
could be explained by an adaptation in exercise pain thresh-
old after a program of exercise rehabilitation.
Patients in both exercise-training groups exhibited im-
provements in their lower-limb peak power and peak V̇O2
on the LCT, with similar peak responses for perceived
exertion (RPE) and pain (CR-10). However, the higher
peak heart rate for the LL-Ex group and higher blood
lactate response for both exercise groups suggests that
patients in the exercising groups were motivated to perform
better on this test and pushed themselves harder after the
period of supervised exercise training.
In the walking assessment, higher peak heart rates and
CR-10 responses were observed at MWD at the 24-week
time point in both exercise groups, suggesting that irre-
spective of training group assignment, patients were able to
tolerate a higher level of cardiovascular stress and were able
to push themselves further beyond the “pain barrier” after
training. The increase in the amount of pain experienced
before test termination was particularly interesting, as it
suggests that the improvement in MWD was at least par-
tially attributable to an alteration in exercise pain tolerance.
This response was not observed in the control patients.
Although a limitation of this study is that assessors were not
blinded to training group assignment, consideration of the
peak heart rate, RPE and CR-10 responses suggests that
patients in both exercise groups were equally motivated to
perform well on tests of physical function.
Given that CD reflects the first perception or onset of
claudication pain, the increase in CD that was observed in
both the UL-Ex and LL-Ex groups cannot be explained by
an alteration in exercise pain tolerance. The similar level of
pain experienced at CD before and after the exercise
training intervention also suggests consistency of pain
perception. This observation suggests that the onset of
the exercise-induced ischemic pain response was delayed at
lower intensities of walking, even though there was no
change in resting ABPI after exercise training in either
group.
JOURNAL OF VASCULAR SURGERY
December 2005
12 weeks 18 weeks 24 weeks
82 ⫾ 16 82 ⫾ 16 82 ⫾ 16
86 ⫾ 23 86 ⫾ 23 86 ⫾ 23
79 ⫾ 15 79 ⫾ 15 79 ⫾ 15
77 ⫾ 6 77 ⫾ 6 77 ⫾ 6
79 ⫾ 5 78 ⫾ 5 79 ⫾ 5
75 ⫾ 8 75 ⫾ 8 76 ⫾ 8
75 ⫾ 4* 75 ⫾ 4* 76 ⫾ 4*
73 ⫾ 3* 73 ⫾ 3* 75 ⫾ 4*
71 ⫾ 5† 68 ⫾ 5 70 ⫾ 5
Although localized metabolic adaptations within the
skeletal muscles of the lower limbs could have contributed
to the improvement in CD observed in the LL-Ex group, it
is not feasible that such a physiologic adaptation accounted
for the improvement in CD observed in the UL-Ex group.
The observed improvement in CD in the UL-Ex group was
more likely to be related to central and/or systemic cardio-
vascular adaptations that improved lower-limb perfusion
during exercise.
An increase in cardiac stroke volume is known to ac-
count for most of the rapid improvement in cardiorespira-
tory capacity observed after short-term programs of aerobic
training in previously sedentary individuals.24 Furthermore,
the increase in stroke volume observed after a program of
exercise training in patients with chronic heart failure has
been shown to be accompanied by a decrease in total
peripheral resistance at rest and during exercise.25,26 An
improvement in cardiac stroke volume, combined with a
decrease in total peripheral resistance, would facilitate im-
proved blood flow to lower-limb skeletal muscle groups
during walking and might explain the improvement in CD
observed in the UL-Ex group. However, other systemic
adaptations such as improved endothelial vasoreactivity,21
nitric oxide metabolism,22 and blood rheologic adapta-
tions23 cannot be dismissed on the basis of the evidence
presented here.
Knowledge of physical activity levels and tight con-
trol of the exercise training stimulus allows more robust
conclusions to be drawn about the relative efficacy of
different exercise regimens and the extent to which
symptomatic improvement can be ascribed to the super-
vised exercise, as opposed to changes in home-based
physical activity. As expected, self-reported physical ac-
tivity level was increased from baseline in the UL-Ex and
LL-Ex groups between weeks 6 and 24 of the interven-
tion. These increases in self-reported physical level,
amounting to approximately 10 MET-h/wk in both
exercise training groups, were largely attributable to the
additional supervised exercise training sessions, as evi-
denced by the change in leisure activities classification.
The increase in self-reported physical activity levels re-
JOURNAL OF VASCULAR SURGERY
Volume 42, Number 6
ported by the control patients at the 12-week time point
was due to an increase in community-based activities
such as walking, swimming, and bowling and the use of
personal exercise ergometers.
Our values for self-reported physical activity levels are in
agreement with the values reported by Regensteiner et al27 for
patients with PAD. Although PAD-PAR responses cor-
relate poorly with objectively measured physical activity
levels and are prone to over-reporting bias,5 the primary
reason for using this measure in the present study was to
ensure that self-reported physical activity status was bal-
anced between the two exercising groups rather than to
precisely quantify the level of physical activity performed
by any one group. As it is likely that any reporting bias
would be controlled for by randomization, our results
show that the increase in self-reported physical activity
level was well balanced across the supervised exercise
training groups. In addition, the exercise intensity used
in the supervised training sessions was equivalent be-
tween the groups, equating to 85% to 90% of the respec-
tive limb-specific peak V̇O2. The reassessment of patients
at 6-week intervals ensured that the exercise stimulus was
maintained at this level throughout the intervention
period, and the interval-style training regimen enabled
exercise of this intensity to be well-tolerated by the
patients.
CONCLUSION
The results of this randomized controlled trial confirm
our earlier preliminary findings that both upper- and lower-
limb weight-supported aerobic exercise training provide an
adequate stimulus for evoking improvements in walking per-
formance in patients with PAD. Evidence from the present
study suggests that the improvement in walking performance
after upper-limb training is due to a combination of central
cardiovascular and/or systemic mechanisms in addition to an
adaptation in exercise pain tolerance that enables patients
to endure a greater intensity of claudication pain before test
termination. These findings demonstrate the effectiveness
of alternative aerobic exercise interventions for patients
with symptomatic PAD.
Arm-cranking was very well tolerated by our patient
cohort and at high exercise intensities using the interval
training regimen. This, and other alternative exercise train-
ing modalities such as leg-cranking, could be a very useful
strategy for improving cardiovascular function and exercise
pain tolerance in patients who have become physically
inactive due to the discomfort that they encounter during
walking, particularly during the early stages of a rehabilita-
tion program.
This program of research is dedicated to the memory of
Professor Richard Wood, MD FRCS, who was involved in
its conception and whose inspiration, enthusiasm, and
counsel are greatly missed.
Zwierska et al 1129
REFERENCES
1. Regensteiner JG, Hiatt WR. Exercise rehabilitation for patients with
peripheral arterial disease. Exerc Sport Sci Rev 1995;23:1-24.
2. Bartelink ML, Stoffers HE, Biesheuvel CJ, Hoes AW. Walking exercise
in patients with intermittent claudication. Experience in routine clinical
practice. Br J Gen Pract 2004;54:196-200.
3. Aquino R, Johnnides C, Makaroun M, Whittle JC, Muluk VS, Kelley
ME, Muluk SC. Natural history of claudication: long-term serial follow-
up study of 1244 claudicants. J Vasc Surg 2001;34:962-70.
4. McDermott MM, Greenland P, Ferrucci L, Criqui MH, Liu K, Sharma L,
et al. Lower extremity performance is associated with daily life physical
activity in individuals with and without peripheral arterial disease. J Am
Geriatr Soc 2002;50:247-55.
5. Otis RB, Brown AS, Womack CJ, Fonong T, Gardner AW. Relationship
between physical activity recall and free-living daily physical activity in
older claudicants. Angiology 2000;51:181-8.
6. Welling RE, Cranley JJ, Krause RJ, Hafner CD. Obliterative arterial
disease of the upper extremity. Arch Surg 1981;116:1593-6.
7. Laroche GP, Bernatz PE, Joyce JW, MacCarty CS. Chronic arterial
insufficiency of the upper extremity. Mayo Clin Proc 1976;51:180-6.
8. Walker RD, Nawaz S, Wilkinson CH, Saxton JM, Pockley AG, Wood
RF. Influence of upper- and lower-limb exercise training on cardiovas-
cular function and walking distances in patients with intermittent clau-
dication. J Vasc Surg 2000;31:662-9.
9. 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 activity: the Walking and Leg Circulation Study. Ann Intern
Med 2002;136:873-83.
10. Zwierska I, Nawaz S, Walker RD, Wood RF, Pockley AG, Saxton JM.
Treadmill versus shuttle walk tests of walking ability in intermittent
claudication. Med Sci Sports Exerc 2004;36:1835-40.
11. Borg G. Borg’s Perceived Exertion and Pain Scales. Leeds, UK: Human
Kinetics; 1998.
12. Hiatt WR, Hirsch AT, Regensteiner JG, Brass EP. Clinical trials for
claudication: assessment of exercise performance, functional status and
clinical end points. Circulation 1995;92:614-21.
13. Robeer GG, Brandsma JW, van den Heuvel SP, Smit B, Oostendorp
RA, Wittens CH. Exercise therapy for intermittent claudication: a
review of the quality of randomised clinical trials and evaluation of
predictive factors. Eur J Vasc Endovasc Surg 1998;15:36-43.
14. Gardner AW, Poehlman ET. Exercise rehabilitation programs for the
treatment of claudication pain. A meta-analysis. JAMA 1995;274:
975-80.
15. Hiatt WR. Functional assessment of the claudicant. Importance of
treatment and follow-up strategies. Minerva Cardioangiol 1999;47:
385-6.
16. Dahllof AG, Bjorntorp P, Holm J, Schersten T. Metabolic activity of
skeletal muscle in patients with peripheral arterial insufficiency. Eur
J Clin Invest 1974;4:9-15.
17. Holm J, Dahllof AG, Bjorntorp P, Schersten T. Enzyme studies in
muscles of patients with intermittent claudication. Effect of training.
Scand J Clin Lab Invest Suppl 1973;128:201-5.
18. Hiatt WR, Regensteiner JG, Hargarten ME, Wolfel EE. Benefit
of exercise conditioning for patients with peripheral arterial disease.
Circulation 1990;81:602-9.
19. Alpert JS, Larsen OA, Lassen NA. Exercise and intermittent claudica-
tion. Blood flow in the calf muscle during walking studied by the
xenon-133 clearance method. Circulation 1969;39:353-9.
20. Gardner AW, Katzel LI, Sorkin JD, Killewich LA, Ryan A, Flinn WR, et al.
Improved functional outcomes following exercise rehabilitation in pa-
tients with intermittent claudication. J Gerontol A Biol Sci Med Sci
2000;55:M570-M577.
21. Brendle DC, Joseph LJ, Corretti MC, Gardner AW, Katzel LI. Effects
of exercise rehabilitation on endothelial reactivity in older patients with
peripheral arterial disease. Am J Cardiol 2001;87:324-9.
22. Arosio E, Cuzzolin L, De Marchi S, Minuz P, Degan M, Crivellente F,
et al. Increased endogenous nitric oxide production induced by physical
exercise in peripheral arterial occlusive disease patients. Life Sci 1999;
65:2815-22.
 JOURNAL OF VASCULAR SURGERY
1130 Zwierska et al December 2005

23. Ernst EE, Matrai A. Intermittent claudication, exercise, and blood
rheology. Circulation 1987;76:1110-4.
24. Saltin B, Blomqvist G, Mitchell JH, Johnson RL Jr, Wildenthal K,
Chapman CB. Response to exercise after bed rest and after training.
Circulation 1968;38(5 Suppl):VII1-78.
25. Hambrecht R, Gielen S, Linke A, Fiehn E, Yu J, Walther C, et al. Effects
of exercise training on left ventricular function and peripheral resistance
in patients with chronic heart failure: A randomized trial. JAMA 2000;
283:3095-101.
26. Gielen S, Erbs S, Schuler G, Hambrecht R. Exercise training and
endothelial dysfunction in coronary artery disease and chronic heart
failure. From molecular biology to clinical benefits. Minerva Cardioan-
giol 2002;50:95-106.
27. Regensteiner JG, Steiner JF, Hiatt WR. Exercise training improves
functional status in patients with peripheral arterial disease. J Vasc Surg
1996;23:104-15.
Submitted Jun 24, 2005; accepted Aug 22, 2005.

COLLECTIONS OF PAPERS
On the Web version of the Journal, selected articles have been grouped together for the convenience of the
readers. The current collections include the following:

American Board of Vascular Surgery Basic Science Reviews

Editorial Comments Guidelines
History Lifeline Research Meeting Abstracts
Reporting Standards Reviews
Technical Notes