CLINICAL EPIDEMIOLOGY
www.jasn.org
Exercise in Patients on Dialysis: A Multicenter,
Randomized Clinical Trial
Fabio Manf redini,*
†
Francesca Mallamaci,
‡§
Graziella D’Arrigo,
‡
Rossella Baggetta,
‡
Davide Bolignano,
‡
Claudia Torino,
‡
Nicola Lamberti,* Silvio Bertoli,
|
Daniele Ciurlino,
|
Lisa Rocca-Rey,
|
Antonio Barillà,
¶
Yuri Battaglia,
¶
Renato Mario Rapanà,**
Alessandro Zuccalà,** Graziella Bonanno,
††
Pasquale Fatuzzo,
‡‡
Francesco Rapisarda,
‡‡
Stefania Rastelli,
‡‡
Fabrizio Fabrizi,
§§
Piergiorgio Messa,
§§
Luciano De Paola,
||
Luigi Lombardi,
||
Adamasco Cupisti,
¶¶
Giorgio Fuiano,*** Gaetano Lucisano,***
Chiara Summaria,*** Michele Felisatti,
†
Enrico Pozzato,
†
Anna Maria Malagoni,
†
Pietro Castellino,
‡‡
Filippo Aucella,
†††
Samar Abd ElHafeez,
‡‡‡
Pasquale Fabio Provenzano,
‡
Giovanni Tripepi,
‡
Luigi Catizone,
¶
and Carm ine Zoccali
‡
Due to the number of contributing authors, the affiliations are listed at the end of this article.
ABSTRACT
Previous studies have suggested the benefits of physical exercise for patients on dialysis. We conducted the Exercise
Introduction to Enhance Performance in Dialysis trial, a 6-month randomized, multicenter trial to test whether a
simple, personalized walking exercise program at home, managed by dialysis staff, improves functional status in adult
patients on dialysis. The main study outcomes included change in physical performance at 6 months, assessed by the
6-minute walking test and the five times sit-to-stand test, and in quality of life, assessed by the Kidney Disease Quality
of Life Short Form (KDQOL-SF) questionnaire. We randomized 296 patients to normal physical activity (control;
n=145) or walking exercise (n=151); 227 patients (exercise n=104; control n=123) repeated the 6-month evaluations.
The distance covered during the 6-minute walking test improved in the exercise group (mean distance6SD: baseline,
328696 m; 6 months, 3676113 m) but not in the control group (baseline, 3216107 m; 6 months, 3246116 m;
P,0.001 between groups). Similarly, the five times sit-to-stand test time improved in the exercise group (mean
time6SD: baseline, 20.566.0 seconds; 6 months, 18.265.7 seconds) but not in the control group (baseline, 20.965.8
seconds; 6 months, 20.266.4 seconds; P=0.001 between groups). The cognitive function score (P=0.04) and quality
of social interaction score (P=0.01) in the kidney disease component of the KDQOL-SF improved significantly in the
exercise arm compared with the control arm. Hence, a simple, personalized, home-based, low-intensity exercise
program managed by dialysis staff may improve physical performance and quality of life in patients on dialysis.
J Am Soc Nephrol 28: 1259–1268, 2017. doi: 10.1681/ASN.2016030378
Poor physical func tioning is perhaps the most per-
vasive and disabling distur bance in patients with
stage G5 CKD maintained on chronic dialysis
(CKD-5D).
1–3
National Kidney Foundation Kidney
Disease Outcomes Quality Initiative Guidelines
formally recommend that patients with CKD-5D
be “counseled and regularly encouraged by ne-
phrology and dialysis staf f to increase the ir level
of physical activity.”
4
However, the evidentiary basis
for recommending exercise training in CKD-5D is
still limited. Although the effect of regular physical
exercise training on physical performance in
selected patients w ith CKD-5D studied in stan-
dardi zed experimental settings in the lab oratory is
Received March 31, 2016. Accepted September 19, 2016.
F. Manfredini and F. Mallamaci contributed equally to this work.
Published online ahead of print. Publication date available at www.j asn.org.
Correspondence: Prof. Carmine Zoccali, National Council of Re-
search, Institute of Clinical Physiology, Clinical Epidemiology and
Physiopathology of Renal Diseases and Hypertension of Reggio
Calabria, c/o Nefrologia e CNR Ospedali Riuniti, 89124 Reggio
Calabria, Italy. Email: carmine.zoccali@tin.it
Copyright © 2017 by the American Society of Nephrology
J Am S oc Nephrol 28: 1259–1268, 2017 ISSN : 1046-6673/2804-1259
1259
well documented,
5
how exercise training should be articulated
(intradialysis or off-dialysis, in-center only, daily versus other
schedules) and implemented (duration and intensity) still re-
mains an open problem. In studies perf ormed so far, physical
exercise was mainly proposed under supervision during the
dialysis session or between two dialysis sessions.
5–7
However,
organization and cost problems mainly related with instru-
ments, personnel, and intensific ation of visits to the dialysis
cente r may hinder patients’ acceptability of exercise progra ms
and, ultimat ely, the diffusion of such programs.
On the basis of a mod el developed for peripheral arterial
disease rehabilitation,
8
we designed an easy-to-implement
home program of physical exercise for patients with CKD-
5D and documented the feasibilit y of such a program in a pilot
study.
9
After this pilot experience, we further simplified this
prog ram into a format whereby a home-based, individuali zed,
low-intensity exercise program could be m anaged by the di-
alysis staff, without extra v isits to the di alysis center (see video
at: https://www.youtube.com/watch?v=ki8YX_t-0jA). We
have now tested in a multicenter, randomized clinical trial,
the EXerCise I ntroduction To Enhan ce performance in dialy-
sis patients trial (EXCITE), whether this home exercise pro-
gram improves the degree of fitness and
quality of life in patients with CKD-5D.
RESULTS
All eligible patients were recruited between
November o f 2009 and February of 20 11.
The Consolidated Standards of Repo rti ng
Trials (CONSORT) diagram describing the
flow of patients through this open, paral lel,
randomized, two-group trial is shown in
Figure 1.
The source population in the 13 Ne-
phrology Units participating in the tr ial
was composed of 714 pati ents receiving di-
alysis. Of these, 296 patients in nine centers
(59.6% eligible patients, 41.5% of total
population) were randomized to walking
exercise (n=151; hemodialysis, n=127;
continuous ambulator y per itoneal dialysis
[CAPD], n=24) or usual care and normal
physical activit y (n=145; hemodialysis,
n=120; CAPD, n=25). As depicted in Fig-
ure 1, 227 of 296 patients (77%) could be
retested after 6 months (104 in the active
arm [hemodialysis, n=90; CAPD, n=14]
and 123 in the control arm [hemodialysis,
n=102; CAPD, n=21]). The reasons for
nonparticipation and exit from the study
are detailed in Figure 1. The two study
groups did not differ for demographic,
clinical, and biochemical data (Table 1)
but did for systolic B P (mean6SD; 138618 versus 127618
mmHg), which tended to be hig her (P=0.06) in patients in the
active arm than in those in the control arm.
Adherence to the Exercise Program in the Active Arm
Out of 104 patients in the exercise arm who were re-evaluated
after 6 months, 76 correctly filled the study diary and 81 re-
turned the metronome for battery verifica tion. Overall, 91
patients (87.5%) documented their degree of compliance to
the exercise program with at least one of these two instruments.
As reported in the personal diarie s, the average number of
sessions p erformed was 1196103 (range, 7–336), correspond-
ing to 83% of the 144 prescribed sessions. Forty-six patients
exceeded the number of prescribed sessions because th ey did
extra session s on the dialysi s days, whereas 29 performed just a
minimal amount (,10%) of the pres cribed sessions. The level
of adhe rence to the exercise program was high for 55 patients
and low for 49 patients. The main determinants of low adher-
ence w ere scarce interest (n=22), orthopedic problems (n=7),
intercurrent nonorthopedic problems (n=10), and problems
related wi th work (n =10). The residual batter y charge was
significantly higher in patients with poor adherence compared
Figu re 1. CONSORT diagram of the flow of patients across the various phases of the trial.
All baseline measurements (including 6MWT and 5STS) were done after randomization.
Baseline demographic data, main cardiovascular comorbidities, and results of the 6MWT
and 5STS in patients who completed the 6-month training program and those who
dropped out in the two study arms are detailed in Supplemental Table 4.
1260 Journal of the American Society of Nephrology J Am So c Nephrol 28: 1259–1268, 2017
CLINICAL EPIDEMIOLOGY
www.jasn.org
with those w ith high adherence (2.982 versus 2.961 mV;
P,0.05), but this indicator had very modest discriminatory
ability to ide ntify patients with a high degree of adherence
from those with a low de gre e, and was no t applied for the
stratification of patients.
Effect of the Home-Based Training Program on
Functional Capacity and Other Parameters
BP and heart rate remained unchang ed after the 6-month ex-
ercise program in the exercise group (6-month BP, 133615/
7369 mmHg; heart rate, 7469 beats/m in). Sim ilarly, serum
creatinine (9.962.7 mg/dl), urea ( 150641 mg/dl), Kt/V (he-
modialysis, 1.4660.30; CAPD, 1.9360.66), albumin (3.760. 4
g/dl), phosphate (5.061. 5 mg/dl), calcium (8.861.0 mg/dl),
parathyroid hormone (269 pg/ml [range, 167– 429]),
choles terol (166638 mg/dl), triglycerides
(1826112 mg/dl), and glucose (115 670
mg/dl) remained the same in the active
group. No change in the same parameters
was obs erved i n the control group. Drug
therapy at baseline was similar in the two
groups and did not change across the trial
(see Supplemental Table 1).
The 6-minute walking distance (6MWD)
improved in the exercise group both
in analysis made in all randomized pa-
tients who started the trial (6 months
versusbaseline:exercisegroup,+39m
[95% confidence interval (95% CI), 33 to
46 m], control group, +2 m [95% CI, 25
to 10]; P,0.001) and in analysis re-
stricted to patients who completed the trial
(exerci se grou p: +41 m, 9 5% CI, 31 to
51 m; control g roup +3 m, 95% CI, 27
to 12 m; P,0 .001), (Figure 2, Supplemen-
tal Table 2). Among patients wh o com-
pleted the tr ial there was a dose–response
relationship between the adherence to ex-
ercise program and 6MWD changes across
the tr ial (Figure 3) (P,0.001). Among the
same patients at the 6-month testing ses-
sion, the average rate of the patient’sper-
ceived exertion by the Borg CR10 Sc ale was
unmodified both in the active group (2.6;
95% CI, 2.2 to 2.9 at baseline, and 2.8; 95%
CI, 2.5 to 3.1 after 6 months) and in the
control group (2.7; 95% CI, 2.4 to 3.0 at
baseline, and 3.2; 95% CI, 2.8 to 3.5 after 6
months). No effect modification by dialy-
sis treatment modality (hemodialysis ver-
sus C APD) was found on the relationship
between allocation arm and changes in
6MWD (P=0.86) and five times sit- to-
stand test (5STS; P=0.26).
At baseline, 13 patie nts in the exercise
arm an d eight in the control a rm were unable to perform any
repetition in the 5STS. In addition, 25 p atients in the active arm
and 21 in the control arm were unable to complete the test for
asthenia and per ipheral fatigue. All these patients were main-
tained in the trial and after 6 months, the corresponding figures
of patients unable to perform any repetition were four (active
arm ) and s even (control arm), and of those unable to complete
the test were nine and 15, respectively. Overall, across the trial
the proportion of patients completely or partially un able to
perform the 5STS was significantly reduced in the first group
compared with the second group (P,0.001, chi- squared test).
Among patients who completed the test, 5STS time im-
proved in the exercise group but not in the control group
(Figure 2) (between-g roup difference, P=0.001) and there
was a dose–respon se rel atio nshi p b etwe en t he 5 STS time
Table 1. Demographic, clinical, and biochemical data of patients that
completed the study
Active Arm (n=104) Control Arm (n=123) P Value
Age, yr 63613 64614 0.60
Men, % 64 68 0.54
Hemodialysis/CAPD, n 90/14 102/21 0.45
BMI, kg/m
2
266427660.32
Smoking , % (0=no; 1=yes) 18 19 0.93
Diabetes, % (0=no; 1=yes) 18 18 0.88
Systolic BP, mmHg 132618 127618 0.06
Diastol ic BP, mmHg 72610 71612 0.43
HR, beats/min 756974680.51
Total cholesterol, mg/dl 164639 166639 0.67
Triglyc erides, mg/dl 1666116 160686 0.68
Hemoglo bin, g/dl 116111620.22
Albumin, g/dl 3.960.4 3.860.5 0.44
Calcium, mg/dl 8.860.7 8.960.7 0.42
Phosphate, mg/dl 4.961.5 4.861.4 0.35
PTH, pg/ml 280 (179–456) 283 (156–396) 0.55
Creatinine, md/dl 10.562.7 9.862.6 0.41
Glycemia, mg/dl 111664 102636 0.23
Urea, mg/dl 153642 148640 0.33
CRP, mg/L 5.0 (3.1–9. 0) 4.6 (3.0–8. 0) 0.60
Kt/V hemodialysis 1.4260.25 1.4360 .30 0.68
Kt/V CAPD 1.9660.29 1.8060 .60 0.36
Myocardial infarction, % 15 17 0.73
Stroke/transient ischemic attack, % 8 14 0.14
Anginal episodes, % 11 13 0.74
Arrhythmia, % 12 7 0.19
Heart failure, % 17 24 0.24
Peripheral vascular disease, % 7 12 0.16
Histor y of neoplasia, % 22 18 0.52
Antihyp ertensive therapy, % 77 70 0.27
NYHA class, %
I38340.46
II 14 16
III–IV 4 10
Mobility, %
Assisted 4 3 0.56
Independent 96 97
BMI, body mass inde x; HR, heart rate; PTH, paratohormone; CRP, C-reactive protein; NYHA, New York
Heart Association.
J Am S oc Nephrol 28: 1259–1268, 2017 Home-Based Exercise Trial in Patients on Dialysis 1261
www.jasn.org
CLINICAL EPIDEMIOLOGY
chan ges (6 months versus baseline) and adherence to the ex-
ercise program (P=0.01) (Figure 3).
Quality of Life
Overall, the global score (Table 2, last line) of the 19 items
composing the Kidney Disease Quality of Life, Short Form,
version 1.3 (KDQOL-SF) on av erage changed
more favora bly in the exercise than in the
control arm, but the diff erence largely
failed to achieve statistical significance
(P=0.17). When compared with changes
in the control arm only two items, both in
the kidney disease component (cognitive
function [P=0.04] and quality of social
interaction [P=0.01]), achieved formal
statistical significance.
Safety of the Exercise Program
Sym ptoms of moderate intensit y, not lim-
iting the program execution, were reported
by 44 patients and included moderate fa-
tigue ( n=31), “ heavy legs” or leg pain
(n=35), moderate dyspnea (n=29), or
other symptoms, including joint pain
(n=17). Five patients reported four symp-
toms, 22 reported three sy mptoms, nine
reported at least two symptoms, and eight
only reported one symptom during th e ex-
ercise sessions. Overall, the training pro-
gram was w ell tolerated a nd o nly five
telephone calls were received by the reha-
bilitation team across the trial. No angina
episode or other major symptoms/compli-
cations during exercise were rep orted in the
active arm of the trial. No systematic symp-
toms collection was undertaken in the con-
trol arm.
Secondary Outcomes: Death,
Cardiovascular Events, and
Hospitalizations
In a Kaplan–Meier analysis including ran-
dom ized patients who started the trial (151
in the exercise and 145 in the control arm),
there was a largely nonsignifica nt reduc-
tion in the risk of hospitalization. However,
in an analysis restricted to patients who
completed the trial, the hospitalization-
free survival was lower (P=0.04) in patients
in the a ctive g roup than in the control
group(Figure4).AsdetailedinSupple-
mental Table 3, among these patients there
were 18 ho spitalizations in ten pat ients in
the exercise group (35 hospitaliz ations per
100 person-years; 95% CI, 21 to 55) and 35
hospitalizations in 24 pati ents in the control group (57 hos-
pitalizations per 100 person-years; 95% CI, 40 to 7 9).
An analysis of AV-fistula events in randomiz ed patients who
started the trial showed that the incidence rate of these out-
comes did not significantly di ffer (P=0.22) between th e exer-
cise (22 events in six patients, 35 ev ents per 100 person-years;
Figure 2. Effect of the study intervent ions on walking capacity (6MWT) and lower limb
strength (5STS) in the two arms of the study. The figures overly ing the columns are SD
of the corresponding mean values. The P value compares c hanges (6 months vers us
baseline) between the two groups (exercise versus control).
1262 Journal of the American Society of Nephrology J Am So c Nephrol 28: 1259–1268, 2017
CLINICAL EPIDEMIOLOGY
www.jasn.org
95% CI, 22 to 53 ) and control (13 events in six patients, 23
events per 100 person-years; 95% CI, 12 to 39) group. A par-
allel analysis in patients who completed the tr ial provided
similar results (ten events in six patients [19 events per 100
person-years; 95% CI, 9 to 35] in the exercise group and nine
events in seven patients [15 events per 100 person-years; 95%
CI, 7 to 28] in the control group).
DISCUSSION
In this multicenter, randomized trial in patients with stage
5D-CKD a simple , personalized, low-intensity, home-based
walking program managed by the dialysis staff improved the
functional status compared with usual care in these patients.
Two items in the kidney disease component of the KDQOL-SF,
nam ely cognitive function and quality of soc ial interaction,
showed a stat istically significan t improvement in the exercise
group compared with the control group.
A comp rehensive meta-analysis published in 2014 identi -
fied 29 randomized trial s, all focused on a single cen ter, testing
physica l exercise programs in patients on di alysis.
5
In the ma-
jority of these trials (14 trials) exercise was performed in-
cente r, during the hemodialysis session. O verall , these trials
documented an improvement in muscular strength and aero-
bic capacit y. T he number of patients in these trials ranged
from 13 to 96, and the vast majorit y had ,40 patients. In
most of these trials the outcome measure was aerobic capacity
and/or cardiovascular/hemodynamic biomarkers (BP, heart
rate, left ventricular mass) or muscle strength. Walking
capacity was a main outcome measure in just one of these
trials. The most recent meta-analysis, published in late
2015,
7
added 11 trials to the 29 analyzed in the previous
meta-analysis. Again, these tri als w ere relatively small (from
eight to 26 patients) , almost all in- center (during hemodialy-
sis, n=9), and looked at disparate outcome measures. A global
analysis of the scientific quality of trials performed so far ac-
cording to the Cochrane col laborat ion recommendations
10
showed that the random sequence generation and the alloca-
tion concealment in these trials was unclear. Furthermore, the
flow diagram of the prog ress of pati ents throug h the phases of
the trial and the outcome data in these trials were largely in-
compl ete, and reporting was selective in most trials.
7
These
limitations notwithstanding, it is indisputable that research on
physic al exercise performed so far in ESRD forms a solid basis
for considering physical exercise as a potentially valuable in-
tervention to improve health outcomes in this po pula tion.
The majority of physical exercise tria ls performed so far
test ed in terventions performed during the hemodialys is ses-
sion and consisted either of aerobic exercise (cycling) or re-
sistance training of some muscular groups against elastic
bands, or a combination thereof.
5,7
The se in-center programs
allow effective, superv ised exercise training. On the other
hand, promoting therapeutic programs embedded in the ac-
tual familial and social conte xt of individual patients,
11
like the
home-based exercise program tested in this trial, is an impor-
tant opportunity for expanding the application of these pro-
grams and for patients empowerment.
The walking capacity encompasses cardiorespiratory and
muscle endurance, muscle strength, and balance and coordi-
nation, which is fundamental in daily li ving in patients with
chronic disease and in the eld erly.
12
In the systematic review
and meta-analysis by Heiwe and Jacobson,
5
only one physical
exercise program was specifically focused on w alking exer-
cise,
13
and just four
13–16
contemplated walking capacity as
an outcome measure. Of no te, t he intervention being tested
in did n ot improve walking capacity in any of these studies.
In a study of 26 patients, subseq uent to Hewe i meta-analysis , a
combined resistance and aerobic intervention
17
was more ef-
fective than isol ated resistance training to improve walking
capacity.
This multicenter trial tested a low-intensit y (20 minut es of
walking at low-to-m oderate speed every second day) exercise
program of gradu ally increasing intensity. The CONSORT di-
ag ram (Figure 1) shows that 32% (2 27 of 714) of the whole
population in participating centers actually completed the
training program, which is a significant proportion of the di-
alysis p opulation. Th is progra m was well accepted by patients
and among th e 151 patients randomized to the exercise arm,
104 (69%) completed the 6-month training. Such an inter-
vention allowed a meaningful increase in walking distance
(41 m). At least in theory, such an increase may favorably affect
clinical outcomes because, beyond the actual duration of the
trial (6 months), in a separate, long-term observational anal-
ysis of the EXCITE cohort exte nded to 3.3 years, we found
Figure 3. Do se–response relationship between achieved physi-
cal performance by the 6MWT and 5STS across the control arm
and the low adherence and high adherence to th e exercise
program (active arm). The bars are SD of the mean. Low and high
adherence we re defined as performance of ,60% and $60% of
the prescribed sessions, respectively. See also adherence to ex-
ercise program in Supplemental Material.
J Am S oc Nephrol 28: 1259–1268, 2017 Home-Based Exercise Trial in Patients on Dialysis 1263
www.jasn.org
CLINICAL EPIDEMIOLOGY