488898

2013
CRE271210.1177/0269215513488898Clinical RehabilitationTanaka et al.

CLINICAL
Article REHABILITATION

Clinical Rehabilitation

Efficacy of strengthening or 27(12) 1059–1071

© The Author(s) 2013
Reprints and permissions:
aerobic exercise on pain relief in sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0269215513488898

people with knee osteoarthritis: cre.sagepub.com

a systematic review and

meta-analysis of randomized
controlled trials

Ryo Tanaka1, Junya Ozawa1, Nobuhiro Kito1 and Hideki Moriyama2

Abstract
Objective: We performed a systematic review and meta-analysis of randomized controlled trials to
investigate the differences in the efficacies between strengthening and aerobic exercises for pain relief in
people with knee osteoarthritis.
Data sources: This search was applied to Medline, Cochrane Central Register of Controlled Trials, the
Physiotherapy Evidence Database, and the Cumulative Index to Nursing and Allied Health Literature. All
literature published from each source’s earliest date to March 2013 was included.
Review methods: Trials comparing the effects of exercise intervention with those of either non-
intervention or psycho-educational intervention were collected. Meta-analysis was performed for trials in
which therapeutic exercise was carried out with more than three sessions per week up to eight weeks,
for pain in people with knee osteoarthritis. All trials were categorised into three subgroups (non-weight-
bearing strengthening exercise, weight-bearing strengthening exercise, and aerobic exercise). Subgroup
analyses were also performed.
Results: Data from eight studies were integrated. Overall effect of exercise was significant with a large
effect size (standardised mean difference (SMD): −0.94; 95% confidence interval −1.31 to −0.57). Subgroup
analyses showed a larger SMD for non-weight-bearing strengthening exercise (−1.42 [−2.09 to −0.75])
compared with weight-bearing strengthening exercise (−0.70 [−1.05 to −0.35]), and aerobic exercise
(−0.45 [−0.77 to −0.13]).
Conclusion: Muscle strengthening exercises with or without weight-bearing and aerobic exercises are
effective for pain relief in people with knee osteoarthritis. In particular, for pain relief by short-term exercise
intervention, the most effective exercise among the three types is non-weight–bearing strengthening exercise.

Keywords
Exercise, knee osteoarthritis, pain, systematic review, meta-analysis

1Department of Integrated Rehabilitation, Hiroshima Corresponding author:

International University, Japan Ryo Tanaka, Department of Integrated Rehabilitation,
2Department of Rehabilitation Science, Graduate School of Hiroshima International University, 555-36, Kurose-gakuendai,
Health Sciences, Kobe University, Japan Higashi-hiroshima, Hiroshima 739-2695, Japan.
Email: r-tanaka@hs.hirokoku-u.ac.jp

Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016

1060
Introduction
Osteoarthritis is a relatively common musculoskel-
etal disorder, with a high prevalence that increases
with age.1 Worldwide, hip and knee osteoarthritis is
one of the leading causes of disability, particularly
in the elderly.2,3 Osteoarthritis is a joint disorder
characterised by progressive degeneration of the
articular cartilage, resulting in the loss of joint space
and reduction in the formation of new marginal and
central bone.4 Structural abnormalities of all tissues
in the joints – including the cartilage, subchondral
bone, synovium, capsule, and ligaments – may also
be present.5 People with symptomatic osteoarthritis
chiefly complain of pain and functional limita-
tion.1,4 In particular, pain is believed to affect the
quality of life; therefore, pain relief is an important
issue in clinical rehabilitation.
Several systematic reviews have outlined and
showed the effectiveness of muscle strengthening
or aerobic exercises in the osteoarthritis popula-
tion,6–8 and a meta-analysis performed in another
review found no difference between the two types
of exercise.9 However, the meta-analysis in that
review included some trials incorporating addi-
tional interventions besides strengthening or aero-
bic exercise (e.g. range of movement exercise,
patellar taping, or non-steroidal anti-inflammatory
drug treatment). Consequently, the exact effects of
either strengthening or aerobic exercise on pain
relief remain unclear. Moreover, the effect of exer-
cise frequency and duration on pain relief was not
controlled, and the effect size of non-weight-bear-
ing strengthening exercise and that of weight-bear-
ing strengthening exercise were not investigated
separately in the meta-analysis. In order to deter-
mine the exact difference in efficacy between
strengthening and aerobic exercises, it is necessary
to perform a meta-analysis after confirming concep-
tual heterogeneity in exercise intervention, exercise
frequency, and duration.
The purpose of the present systematic review and
meta-analysis of randomized controlled trials was to
investigate the differences in the efficacies between
strengthening and aerobic exercises for pain relief in
people with knee osteoarthritis. We selected similar
trials in terms of the types of exercise and performed
Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016
Clinical Rehabilitation 27(12)
a meta-analysis and subgroup analysis for the trials,
while controlling for the influence of exercise fre-
quency and duration on pain relief.
Methods
In this study, we first performed a systematic review
to identify the relevant studies, and then followed it
up with a meta-analysis on the basis of the initial
systematic review. This systematic review and
meta-analysis was performed according to the pre-
ferred reporting items for systematic reviews and
meta-analyses statement protocol, as described
elsewhere.10
Inclusion and exclusion criteria for trials were
based on study design, participants, interventions,
comparisons, and outcomes as follows. Randomized
controlled trials on the efficacy of exercise were
included. Observational studies such as cohort stud-
ies were excluded.
Participants who had been diagnosed with knee
osteoarthritis were considered. No age, sex, or body
mass index restrictions were imposed. Preoperative
people were excluded from this review because
therapeutic exercise could not be applied to these
individuals. Furthermore, postoperative people (e.g.
total knee arthroplasty) were excluded from this
review.
The studies performing non-weight-bearing
strengthening exercise, weight-bearing strengthen-
ing exercise, or aerobic exercise were included. The
studies including intervention with combination
exercise (for example, trial including strengthening
exercise and aerobic exercise) were excluded from
our review, because of reducing the conceptual het-
erogeneity of exercise intervention. Moreover, in
order to minimise the influence of exercise fre-
quency and duration on pain relief, trials including
lower than the median of exercise frequency or lon-
ger than the median of exercise duration among
identified trials in our searching were excluded
from our analysis. In cardiac rehabilitation,
Berkhuysen et al.11 showed that high-frequency
patients (10 sessions per week) reported more posi-
tive change significantly in the self-reported out-
come than low-frequency patients (2 sessions per
Tanaka et al.
week). In addition, Fransen and McConnell12 con-
cluded in their the systematic review, which was
published in Cochrane library, that land-based ther-
apeutic exercise has at least short-term benefit in
terms of reduced knee pain, but there is a clinical
impression that the beneficial effects of exercise on
osteoarthritis of the knee disappear over time.13
Therefore, if there was heterogeneity in exercise
frequency or duration among trials, we could not
determine the exact difference in the efficacy
between strengthening and aerobic exercises.
Trials comparing therapeutic exercises with
either non-intervention or psycho-educational inter-
ventions in people with knee osteoarthritis were
included. Jamtvedt et al.14, in their systematic
review, wrote that psycho-educational interventions
improve psychological outcomes, but no clinically
important difference was found for pain.
The primary outcome measure was pain. Trials
using self-reported scales of pain were selected.
Studies were identified by searching electronic
databases. This search was applied to Medline,
Cochrane Central Register of Controlled Trials,
the Physiotherapy Evidence Database, and the
Cumulative Index to Nursing and Allied Health
Literature. Publication data was through 20 March
2013. Publication language was not restricted.
The first author (RT) developed and conducted
the search.
We used the following search terms to search all
trial registers and databases: osteoarthritis knee,
exercise, exercise therapy, and pain. The search
strategy consisted of a combination of free text
words and medical subject headings terms. Article
searches in the Medline database were limited to
clinical trials and randomized controlled trials.
Search strategies were peer reviewed as part of the
systematic review process. The search strategy is
available in the appendix, which is available online.
Two authors (RT and JO) scanned the articles
retrieved by the initial search to exclude obvi-
ously irrelevant studies. Study eligibility was
determined by reading the title and abstracts, and
obviously irrelevant trials were excluded at this
stage. Inclusion and exclusion criteria were estab-
lished a priori. Disagreements between reviewers
were resolved by consensus.
Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016
1061
Two authors (RT and JO) read the full-text arti-
cles and independently decided whether the
retrieved trials met the inclusion criteria or not. Two
authors (RT and JO) extracted the data using special
standardised forms developed for this review. We
contacted no author for further information. The
primary outcome data were entered into the Review
Manager 5.1 software (The Cochrane Collaboration,
Copenhagen, Denmark) for further statistical analy-
sis. All differences resulting from discrepant assess-
ments during data extraction and analysis were
resolved by consulting a third author (NK) or dis-
cussion among the authors. Information was
extracted from each included trial on: (a) the char-
acteristics of trial participants; (b) the contents of
exercise intervention and regimen; and (c) the out-
come scale.
The quality of the trials that were included in our
review was determined by using the PEDro scale.15
The PEDro scale is based on the Delphi list devel-
oped by Verhagen and colleagues at the Department
of Epidemiology, University of Maastricht.16 The
PEDro scale contains 10 check items: ‘random allo-
cation’, ‘concealed allocation’, ‘baseline compara-
bility’, ‘blind subjects’, ‘blind therapists’, ‘blind
assessors’, ‘adequate follow-up’, ‘intention-to-treat
analysis’, ‘between-group comparisons’, and ‘point
estimates and variability’.
Since the outcome consisted of continuous data,
which were standardised to the same scale, the stan-
dardised mean difference was used as the effect
measure. The standardised mean difference is the
difference in mean outcome between groups divided
by the standard deviation of the outcome.17 The
standardised mean difference and their correspond-
ing 95% confidence interval were calculated for the
continuous outcome data. The selection of using
either a fixed-effect model17 or a random-effect
model17 was decided after examination of statistical
heterogeneity between trials.
The inverse variance method17 was used to syn-
thesise the results by means of the Review Manager
5 software. If a standard deviation was not given, it
could be calculated from the 95% confidence inter-
val or the standard error of the mean values. Trials
that did not give means and standard deviation,
95% confidence interval, or the standard error of
1062
the mean values were excluded from the synthesis
in our study. Positive values indicate that the control
group improved on average more than the exercise
group. Effect sizes of 0.2–0.5 can be interpreted as
small, 0.5–0.8 as medium, and greater than 0.8 as
large effects.18 Heterogeneity between trials was
examined by the χ2 test and the l2 test.17 The l2 test
does not inherently depend on the number of studies
and is accompanied by an uncertainty interval. A
rough guide to interpretation is as follows:17 0%–
40% might not be important; 30%–60% may repre-
sent moderate heterogeneity; 50%–90% may
represent substantial heterogeneity; 75%–100%
considerable heterogeneity.
We stratified the trials according to the type of
exercise and subgroup analyses were performed. The
trials stratified according to type of exercise were
classified as non-weight-bearing strengthening exer-
cise category, weight-bearing strengthening exercise
category, and aerobic exercise category. In our study,
the stratification according to the delivery mode (e.g.
individual treatments, exercise classes, or home pro-
gramme) was not performed, because the mode of
treatment delivery was not significantly associated
with the magnitude of treatment benefit.19
For each trial, we plotted the effect by the inverse
of its standard error. The symmetry of such ‘funnel
plots’ was assessed both visually and formally. As a
subsequent check of publication bias, the test of
Egger et al.20 was incorporated. In the Egger test the
standard normal deviation is regressed on precision,
defined as the inverse of the standard error.21 This is
a formal statistical analysis that is intended to assess
the same assumption as the funnel plot, and may be
used as a ‘cross-check’ to the physical inspection of
the data.
Results
The search of Medline, Cochrane Central Register
of Controlled Trials, Physiotherapy Evidence
Database, and Cumulative Index to Nursing and
Allied Health Literature provided a total of 559 cita-
tions. We excluded 551 studies as shown in Figure 1
and performed the meta-analysis for eight studies
(11 exercise groups) consequently.
Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016
Clinical Rehabilitation 27(12)
A list of the included studies is shown in Table 1.
Of 11 exercise groups in eight studies, six exercise
groups in four studies22–25 examined the effect of
non-weight-bearing muscle strengthening only (i.e.
concentric-eccentric, isometric, or isokinetic exer-
cise under non-weight-bearing conditions). Two
exercise groups in one study26 examined the effect
of weight-bearing muscle strengthening (i.e. exer-
cise with a leg press machine) with or without non-
weight-bearing muscle strengthening, but did not
examine the effect of aerobic exercise. Three exer-
cise groups27–29 examined the effect of aerobic exer-
cise (i.e. walking, Baduanjin, or Tai Chi exercise).
The result of PEDro scoring is available in the
appendix, available online.
The means of original data, standard deviation,
sample sizes, and the standardised mean difference
(95% confidence interval) of each type of exercise
is seen in the forest plots (Figure 2). All eight stud-
ies showed that the effect of the exercise group was
better than that of the control group. Five of the
eight studies suggested significant differences
between the exercise and control groups. No studies
indicate that the effect of the control group was sig-
nificantly better than that of the exercise group. The
effect integrated with all trials was significant (P <
0.00001), and the pooled standardised mean differ-
ence was large (−0.94 [−1.31 to −0.57]).
All trials were stratified according to the type of
exercise and subgroup analyses were performed.
The results according to type of exercise are shown
in Table 2. Statistical substantial heterogeneity
between the trials for non-weight-bearing strength-
ening exercise and the trials for weight-bearing
strengthening exercise or aerobic exercise was
observed (l2 = 71.1% and 84.6%, respectively). The
pooled standardised mean difference of trials for
non-weight-bearing strengthening exercise was
larger than that of trials for weight-bearing strength-
ening exercise or aerobic exercise. There was not
important heterogeneity between the trials for
weight-bearing strengthening exercise and the trials
for aerobic exercise (l2 = 7.5%).
Analysis of funnel plots (Figure 3) as well as
results from the Egger’s test suggested negative evi-
dence of publication bias for all trials (P = 0.123).
The number of studies in each subgroup is four at
Tanaka et al. 1063

Figure 1. Flow diagram of included and excluded studies.

Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016

 Table 1. Summary of included studies.
1064
Study Participants Exercise contents/subgroup Frequency Duration Outcome PEDro
(Year) (per week) (week) scale score

Size Age Sex

(Average) (% female)
Gür et al. E1: 8 E1: 55 Unknown E1: Concentric-eccentric exercise/NWBSE 3 8 NRS 5
(2002) E2: 9 E2: 56 6 concentric extensions, eccentric extension
C: 9 C: 57 movements, followed by 6 concentric flexion, eccentric
flexion movements
Spectrum of angular velocities ranging from 30°/s to
180°/s at 30°/s-intervals (i.e. 30°, 60°, 90°, 120°, 150°,
and 180°)
Bilaterally
2-minute rest between knee extensors and flexors
5-minute rest between leg sides
E2: Concentric exercise/NWBSE
12 concentric extensions, concentric flexion movements
Spectrum of angular velocities ranging from 30°/s to
180°/s at 30°/s-intervals (i.e. 30°, 60°, 90°, 120°, 150°,
and 180°)
Bilaterally
5-minute rest between leg sides
Salli et al. E1: 25 E1: 55.7 E1: 83% E1: Combined concentric-eccentric exercise/NWBSE 3 8 VAS 7
(2010) E2: 25 E2: 57.1 E2: 83% 10 repetitions at each angular velocity: 60°, 90°, 120°,
C: 25 C: 58.3 C: 79% 150°, and 180°/s

Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016

Bilaterally
70% maximal voluntary contraction
10 repetitions at each angular velocity: 60°, 90°, 120°,
150°, and 180°/s
Bilaterally
70% maximal voluntary contraction
E2: Isometric exercise/NWBSE
5 concentric-concentric isokinetic exercise followed by
5 eccentric-eccentric isokinetic exercises
Clinical Rehabilitation 27(12)
 Table 1. (Continued)

Study Participants Exercise contents/subgroup Frequency Duration Outcome PEDro

(Year) (per week) (week) scale score
Tanaka et al.

Size Age Sex

(Average) (% female)

Bilaterally
Both at knee flexion and extension
20-second rest between velocity break
3-minute rest between knee breaks
Lin et al. E: 36 E: 61.6 E: 67% Concentric-eccentric exercise/NWBSE 3 8 WOMAC 8
(2009) C: 36 C: 62.2 C: 72% Concentric quadriceps action, followed by an eccentric
quadriceps action
Bilaterally
50% of 1-repetition maximum
Progressive increment of 5% of the original 1-repetition
maximum every 2 weeks
4 sets with 6 repetitions per set
1-minute rest between sets
5-minute rest between sides
Schilke E: 10 E: 64.5 85 Isokinetic strength training/NWBSE 3 8 OASI 4
et al. C: 10 C: 68.4
(2006) Knee extension and flexion

Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016

Angular velocity 90°/s
5 contractions
1 to 6 sets (progressively)
1-minute rest between sets and 15-minute rest between
sets 3 and 4 from 4 session onward
Jan et al. E1: 34 E1: 68.3 E1: 79% E1: Leg press machine, the high-resistance exercise/WBSE 3 8 WOMAC 7
(2008) E2: 34 E2: 61.8 E2: 79% 60% of 1 repetition maximum
C: 34 C: 62.8 C: 83% A progressive increment of 5% of the original 1
repetition maximum every 2 weeks
8 repetitions
3 sets
1065

(Continued)
 Table 1. (Continued)
Study Participants Exercise contents/subgroup Frequency Duration Outcome PEDro 1066
(Year) (per week) (week) scale score

Size Age Sex

(Average) (% female)
1-minute rest between sets
5-minute rest between left and right knee sessions
E2: Leg press machine, the low-resistance exercise /
WBSE
10% of 1 repetition maximum
Progressive increment of 5% of the original 1 repetition
maximum every 2 weeks
15 repetitions
10 sets
1-minute rest between sets
5-minute rest between left and right knee sessions.
An et al. E: 14 E: 65.4 E: 100% Baduanjin (AE)
(2008) C: 14 C: 64.6 C: 100% 8 sections
20 repetitions
30-minute session
Kovar et E: 52 E: 70.4 E: 77% Walking and patient education/AE 3 8 AIMS 5
al. C: 50 C: 68.5 C: 90% Walking: up to 30 minutes
(1992) Patient education: guest speakers on the medical aspects
of osteoarthritis and exercise; group discussion about
barriers and benefits of walking; instruction in proper

Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016

walking techniques and the maintenance of walking
program; supportive encouragement
Brismee E: 22 E: 70.8 E: 86.4% Tai chi training/AE 3 6 WOMAC 5
et al. C: 19 C: 68.8 C: 78.9% The 24-form simplified Yang-style tai chi
(2006) 6 minutes per routine
5 repetitions

NWBSE, non-weight-bearing strengthening exercise; WBSE, weight-bearing strengthening exercise; AB, aerobic exercise; E, exercise group; C, Control group; NRS, Numerical
Rating Scale;VAS,Visual Analogue Scale; WOMAC, Western Ontario and McMaster Universities Arthritis Index; OASI, Osteoarthritis Screening Index; AIMS, Arthritis Impact
Measurement Scale.
Some studies include multiple exercise, their groups are shown as E1, E2 in this table.
Clinical Rehabilitation 27(12)
Tanaka et al. 1067

Exercise Control Std. Mean Difference Std. Mean Difference

Study or Subgroup Mean SD Total Mean SD Total Weight IV, Random, 95% CI IV, Random, 95% CI
1.1.1 Non-weight-bearing strengthening exercise
Gur 2002 10.3 4.5 9 28 5.2 6 3.3% -3.48 [-5.27, -1.70]
Salli 2010 2.8 1.7 23 6.5 1.8 24 9.2% -2.08 [-2.80, -1.36]
Gur 2002 16.6 7.3 8 28 5.2 6 5.3% -1.64 [-2.92, -0.36]
Salli 2010 3.9 1.9 24 6.5 1.8 24 10.0% -1.38 [-2.02, -0.75]
Lin 2009 4.2 3 36 7.3 3.4 36 11.4% -0.96 [-1.45, -0.47]
Schilke 1996 9.7 4.72 10 10.1 6.44 10 7.9% -0.07 [-0.94, 0.81]
Subtotal (95% CI) 110 106 47.2% -1.42 [-2.09, -0.75]
Heterogeneity: Tau² = 0.48; Chi² = 19.98, df = 5 (P = 0.001); I² = 75%
Test for overall effect: Z = 4.13 (P < 0.0001)

1.1.2 Weight-bearing strengthening exercise

Jan 2008 4.8 2.7 34 7.1 3.4 34 11.4% -0.74 [-1.23, -0.25]
Jan 2008 4.8 3.5 34 7.1 3.4 34 11.4% -0.66 [-1.15, -0.17]
Subtotal (95% CI) 68 68 22.8% -0.70 [-1.05, -0.35]
Heterogeneity: Tau² = 0.00; Chi² = 0.05, df = 1 (P = 0.82); I² = 0%
Test for overall effect: Z = 3.95 (P < 0.0001)

1.1.3 Aerobic exercise

An 2008 71.1 110.1 11 138.2 112.6 10 7.9% -0.58 [-1.46, 0.30]
Kovar 1992 3.77 1.73 47 4.77 2.12 45 12.1% -0.51 [-0.93, -0.10]
Brismee 2006 15.39 5.7 22 16.64 4.57 18 10.1% -0.23 [-0.86, 0.39]
Subtotal (95% CI) 80 73 30.1% -0.45 [-0.77, -0.13]
Heterogeneity: Tau² = 0.00; Chi² = 0.63, df = 2 (P = 0.73); I² = 0%
Test for overall effect: Z = 2.73 (P = 0.006)

Total (95% CI) 258 247 100.0% -0.94 [-1.31, -0.57]

Heterogeneity: Tau² = 0.25; Chi² = 34.24, df = 10 (P = 0.0002); I² = 71%
-4 -2 0 2 4
Test for overall effect: Z = 4.96 (P < 0.00001) Favors exercise Favors control
Test for subgroup differences: Chi² = 6.59, df = 2 (P = 0.04), I² = 69.6%

Figure 2. Forest plots (eight studies, 11 exercise groups).

A random-effect model was used because of significant heterogeneity (l2 = 71%). The standardised mean difference of trials was
different among subgroups (l2 = 69.6%).

Table 2. The results of subgroup analysis according to the type of exercise.

The type of exercise The SMD of trials [95% CI] Heterogeneity (l2)

NWBSE WBSE AE
NWBSE —1.42 — 71.1% 84.6%
(6 exercise groups) [—2.09, —0.75]
WBSE —0.70 — 7.5%
(2 exercise groups) [—1.05, —0.35]
AE —0.45 —
(3 exercise groups ) [—0.77, —0.13]
NWBSE, non—weight—bearing strengthening exercise; WBSE, weight—bearing strengthening exercise; AE, aerobic exercise; SMD,
standardised mean difference; CI, confidence interval.

maximum. According to the Cochrane handbook there are at least 10 studies included in the meta-
for systematic reviews of interventions,17 tests for analysis, because when there are fewer studies the
funnel plot asymmetry should be used only when power of the tests is too low to distinguish chance

Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016

1068
SE(SMD)
0
0.2
0.4
0.6
0.8
1
-4 -2
Subgroups
Non-weight-bearing strengthening exercise
Weight-bearing strengthening exercise
Figure 3. Funnel plots (eight studies, 11 exercise groups).
Asymmetrical funnel plots suggest the negative publication bias. SE, standard error; SMD, standardised mean difference.
from real asymmetry. Consequently, we could not
assess the risk of publication bias for subgroups.
Discussion
Our study involved a meta-analysis and subgroup
analysis to compare the exact effect of strengthen-
ing and aerobic exercises on pain relief in people
with knee osteoarthritis. The synthesised trials data
suggested statistical heterogeneity among trials.
Subgroup analysis was performed after the stratifi-
cation of trials according to the type of exercise.
Both types of strengthening exercise were effec-
tive for pain relief, regardless of whether the weight-
bearing condition was used. Moreover, we observed
that the pooled effect size of non-weight-bearing
strengthening exercise was larger than that of weight-
bearing strengthening exercise. In only one trial
reported by Jan et al.30 that compared the effects of
non-weight-bearing versus weight-bearing exercise
in participants with knee osteoarthritis, but pain was
Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016
Clinical Rehabilitation 27(12)
SMD
0 2 4
Aerobic exercise
not included as an outcome measure. To our knowl-
edge, no studies have reported the comparative effec-
tiveness of weight-bearing positions in muscle
strengthening exercise for pain relief. Exercise in the
standing or weight-bearing positions for participants
with painful knee osteoarthritis might aggravate
symptoms such as pain, swelling, and inflammation
if the knee joint is overloaded.31, 32 Overloaded
weight-bearing positions should not be suggested for
strengthening exercise to relieve pain.
The pooled effect size of non-weight-bearing
strengthening exercise was also larger than that of
aerobic exercise. Roddy et al.9 indirectly compared
the effect of strengthening exercise and aerobic exer-
cise on pain relief and concluded that there were no
differences between them. Although they performed
a systematic review and meta-analysis, non-weight-
bearing strengthening exercise in people with knee
osteoarthritis was used in only three of 11 exercise
groups categorised as those involving strengthening
exercise. Another eight trials included subjects with
hip osteoarthritis33 and not knee osteoarthritis,34 an
Tanaka et al.
exercise group involving weight-bearing strengthen-
ing exercise,35–38 and a control group with exercise
(e.g. stretching or range of movement exercise).39
Accordingly, their findings did not always reflect a
lack of difference between non-weight-bearing
strengthening exercise and aerobic exercise in pain
relief in people with knee osteoarthritis.
Our findings suggest that the advantage of
strengthening exercise exists only for non-weight-
bearing regimens. Ettinger et al.35 showed that both
resistance exercise and aerobic exercise reduced
pain significantly compared with health education
over the course of the study. However, they did not
reveal an advantage of resistance exercise over aer-
obic exercise. Our findings seem to be inconsistent
with their results; this contradiction might be
explained by the difference of exercise content in
the resistance group, which included weight-bear-
ing strengthening exercise. As mentioned above,
exercise in weight-bearing positions for people with
painful knee osteoarthritis might aggravate pain.31,32
Our results might suggest that exercise therapy
for pain relief should be performed under reduced
loading conditions in people with knee osteoarthri-
tis. However, persistent reductions in joint loading
may have harmful effects on cartilage metabolism.
Hinterwimmer et al.40 showed that even during
short-term reduced loading conditions (partial
weight-bearing at the knee joint for seven weeks
after a surgical intervention at the ankle), a signifi-
cant degree of cartilage thinning was observed.
These data indicate that the cartilage undergoes
some process of atrophy in the absence of mechani-
cal stimulation. Furthermore, Jan et al.30 reported
that equally significant improvements were appar-
ent on both the Western Ontario and McMaster
Universities Arthritis Index function scale and mus-
cle torque after weight-bearing and non-weight-
bearing exercises, except for reposition error, for
which improvement was greater in the weight-bear-
ing exercise group. If people with knee osteoarthri-
tis performed non-weight-bearing exercises for pain
relief, exercise with a loading condition at the knee
joint should be performed immediately after suffi-
cient pain relief is achieved in order to inhibit carti-
lage thinning and improve position sense.
Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016
1069
The major strength of this study is its design,
which controls for the effects of confounding fac-
tors on the efficacy of exercise. Although previ-
ously published systematic reviews have reported
differences in efficacy between various types of
exercise,9,19 no systematic reviews including meta-
analyses have been performed that control for the
effects of exercise frequency and exercise duration
of the efficacy of exercise. In our study, almost all
of the included trials used frequencies of three exer-
cise sessions per week for a duration of eight weeks.
However, it should be pointed out that we excluded
studies that were very effective but fell outside the
scope of our study because of their lower exercise
frequency or longer term than that seen in other
studies. Accordingly, we would like to emphasise
our interpretation that the advantage of non-weight–
bearing strengthening exercise for pain relief com-
pared with weight-bearing strengthening or aerobic
exercise would be restricted to the first eight weeks
following the start of the exercise intervention with
over three sessions per week.
Our study has several limitations; one of these is
the quality of trials included in our study. The influ-
ence of the quality of trials on the results of meta-
analysis has been reported.19 This meta-analysis
showed that all three study categories (low, medium,
or high risk of bias) achieved significant mean treat-
ment benefits in terms of pain, and that studies at a
low risk of bias showed small mean treatment
effects for pain. The studies at a high risk of bias in
this review included those that were not blinded to
outcome assessment, had no appropriate handling
of participant attrition, and lacked adequate alloca-
tion concealment. Studies with a high risk of bias
showed mostly moderate mean treatment effects. In
our study, we performed the meta-analysis on trials
including those with low quality,22,25,27,28 which
means a high risk of bias in past meta-analyses.19
Therefore, the quality of trials may be considered to
have been incorporated in our study.
Evidence of publication bias for all trials was not
detected by Egger’s test in our results. This result
suggests that the effect size of overall exercise ther-
apy has not been influenced by publication bias.
However, an estimation of publication bias for each
1070
subgroup could not be determined, because the sub-
groups were too small to perform this test.
Consequently, we could not judge whether publica-
tion bias in the effect size for pain relief from
strengthening or aerobic exercise exists. As shown
in Figure 3, one exercise group in a trial reported by
Gur et al.,22 which was categorised as a subgroup
involved in non-weight-bearing exercise, showed
particularly higher standardised mean difference
than in other trials. If the existence of publication
bias is recognised and eliminated in future system-
atic reviews using an Egger’s test, it is probable that
the pooled effect size for pain relief from non-
weight-bearing exercise would become smaller than
our data at present.
We observed a difference in therapeutic efficacy
between strengthening and aerobic exercises. In
short-term, non-weight-bearing strengthening exer-
cises are recommended to reduce exercise pain in
people with knee osteoarthritis. Future studies
involving meta-analysis to examine the effects of
both strengthening exercise with or without weight-
bearing and aerobic exercise on pain relief in people
with knee osteoarthritis should consider the issues
of the quality of trials and publication bias.
Clinical messages
Muscle strengthening exercises with or
without weight-bearing and aerobic exer-
cises are effective for pain relief in peo-
ple with knee osteoarthritis.
In particular the first eight weeks follow-
ing the onset of the exercise interven-
tion, the most effective exercise among
the three types is non-weight-bearing
strengthening exercise.
Conflict of interest
The author declares that there is no conflict of interest.
Funding
This research received no specific grant from any funding
agency in the public, commercial, or not-for-profit sectors.
Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016
Clinical Rehabilitation 27(12)
References
1. Pisters MF, Veenhof C, van Meeteren NL, et al. Long-term
effectiveness of exercise therapy in patients with osteo-
arthritis of the hip or knee: a systematic review. Arthritis
Rheum 2007; 57: 1245–1253.
2. Guccione AA, Felson DT, Anderson JJ, et al. The effects of
specific medical conditions on the functional limitations of
elders in the Framingham Study. Am J Public Health 1994;
84: 351–358.
3. Woolf AD and Pfleger B. Burden of major musculoskeletal
conditions. Bull World Health Organ 2003; 81: 646–656.
4. Lin CW, Taylor D, Bierma-Zeinstra SM and Maher CG.
Exercise for osteoarthritis of the knee. Phys Ther 2010; 90:
839–842.
5. Felson DT, Lawrence RC, Hochberg MC, et al. Osteoarthri-
tis: new insights. Part 2: treatment approaches. Ann Intern
Med 2000; 133: 726–737.
6. Fransen M, McConnell S and Bell M. Therapeutic exercise
for people with osteoarthritis of the hip or knee. A system-
atic review. J Rheumatol 2002; 29: 1737–1745.
7. van Baar ME, Assendelft WJ, Dekker J, Oostendorp RA
and Bijlsma JW. Effectiveness of exercise therapy in
patients with osteoarthritis of the hip or knee: a systematic
review of randomized clinical trials. Arthritis Rheum 1999;
42: 1361–1369.
8. Smidt N, de Vet HC, Bouter LM, et al. Effectiveness of
exercise therapy: a best-evidence summary of systematic
reviews. Aust J Physiother 2005; 51: 71–85.
9. Roddy E, Zhang W and Doherty M. Aerobic walking or
strengthening exercise for osteoarthritis of the knee? A sys-
tematic review. Ann Rheum Dis 2005; 64: 544–548.
10. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA state-
ment for reporting systematic reviews and meta-analyses of
studies that evaluate health care interventions: explanation
and elaboration. PLoS Med 2009; 6: e1000100.
11. Berkhuysen MA, Nieuwland W, Buunk BP, Sanderman
R, Viersma JW and Rispens P. Effect of high- versus low-
frequency exercise training in multidisciplinary cardiac
rehabilitation on health-related quality of life. J Cardiopul
Rehabil 1999; 19: 22–28.
12. Fransen M and McConnell S. Exercise for osteoarthritis of
the knee. Cochrane Database Syst Rev 2008: CD004376.
13. Zhang W, Nuki G, Moskowitz RW, et al. OARSI recom-
mendations for the management of hip and knee osteoar-
thritis: part III: Changes in evidence following systematic
cumulative update of research published through January
2009. Osteoarthritis Cartilage 2009; 18: 476–499.
14. Jamtvedt G, Dahm KT, Christie A, et al. Physical therapy
interventions for patients with osteoarthritis of the knee: an
overview of systematic reviews. Phys Ther 2008; 88: 123–136.
15. Maher CG, Sherrington C, Herbert RD, Moseley AM and
Elkins M. Reliability of the PEDro scale for rating qual-
ity of randomized controlled trials. Phys Ther 2003; 83:
713–721.
16. Verhagen AP, de Vet HC, de Bie RA, et al. The Delphi list:
a criteria list for quality assessment of randomized clinical
Tanaka et al.
trials for conducting systematic reviews developed by Del-
phi consensus. J Clin Epidemiol 1998; 51: 1235–1241.
17. Deeks JJ, Higgins JPT and Altman DG. Analysing data and
undertaking meta-analyses. In: Higgins JPT and Green S
(eds) Cochrane Handbook for Systematic Reviews of Inter-
ventions. West Sussex: Wiley-Blackwell, 2008, pp.243–296.
18. Cohen J. Quantitative methods in psychology. Psychologi-
cal Bulletin 1992; 112: 155–159.
19. Fransen M and McConnell S. Land-based exercise for
osteoarthritis of the knee: a metaanalysis of randomized
controlled trials. J Rheumatol 2009; 36: 1109–1117.
20. Egger M, Davey Smith G, Schneider M and Minder C. Bias
in meta-analysis detected by a simple, graphical test. BMJ
1997; 315: 629–634.
21. Borenstein M. Software for Publication Bias. In: Rothstein
H, Sutton A and Borenstein M (eds) Publication Bias in
Meta-analysis: Prevention, Assessment and Adjustments.
Hoboken (NJ): Wiley, 2005, pp.391–403.
22. Gur H, Cakin N, Akova B, Okay E and Kucukoglu S. Con-
centric versus combined concentric-eccentric isokinetic
training: effects on functional capacity and symptoms in
patients with osteoarthrosis of the knee. Arch Phys Med
Rehabil 2002; 83: 308–316.
23. Lin DH, Lin CH, Lin YF and Jan MH. Efficacy of 2 non-
weight-bearing interventions, proprioception training ver-
sus strength training, for patients with knee osteoarthritis: a
randomized clinical trial. J Orthop Sports Phys Ther 2009;
39: 450–457.
24. Salli A, Sahin N, Baskent A and Ugurlu H. The effect of
two exercise programs on various functional outcome
measures in patients with osteoarthritis of the knee: A ran-
domized controlled clinical trial. Isokinetics and Exercise
Science 2010; 18: 201–209.
25. Schilke JM, Johnson GO, Housh TJ and O’Dell JR. Effects
of muscle-strength training on the functional status of
patients with osteoarthritis of the knee joint. Nurs Res 1996;
45: 68–72.
26. Jan MH, Lin JJ, Liau JJ, Lin YF and Lin DH. Investigation
of clinical effects of high- and low-resistance training for
patients with knee osteoarthritis: a randomized controlled
trial. Phys Ther 2008; 88: 427–436.
27. An B, Dai K, Zhu Z, et al. Baduanjin alleviates the symp-
toms of knee osteoarthritis. J Altern Complement Med
2008; 14: 167–174.
28. Kovar PA, Allegrante JP, MacKenzie CR, Peterson MG,
Gutin B and Charlson ME. Supervised fitness walking in
patients with osteoarthritis of the knee. A randomized, con-
trolled trial. Ann Intern Med 1992; 116: 529–534.
Downloaded from cre.sagepub.com at Hoegskolen i Bergen on January 9, 2016
1071
29. Brismee JM, Paige RL, Chyu MC, et al. Group and home-
based tai chi in elderly subjects with knee osteoarthritis: a
randomized controlled trial. Clin Rehabil 2007; 21: 99–111.
30. Jan MH, Lin CH, Lin YF, Lin JJ and Lin DH. Effects of
weight-bearing versus nonweight-bearing exercise on func-
tion, walking speed, and position sense in participants with
knee osteoarthritis: a randomized controlled trial. Arch Phys
Med Rehabil 2009; 90: 897–904.
31. Baliunas AJ, Hurwitz DE, Ryals AB, et al. Increased knee
joint loads during walking are present in subjects with
knee osteoarthritis. Osteoarthritis Cartilage 2002; 10:
573–579.
32. Lucchinetti E, Adams CS, Horton WE, Jr. and Torzilli PA.
Cartilage viability after repetitive loading: a preliminary
report. Osteoarthritis Cartilage 2002; 10: 71–81.
33. Hopman-Rock M and Westhoff MH. The effects of a health
educational and exercise program for older adults with
osteoarthritis for the hip or knee. J Rheumatol 2000; 27:
1947–1954.
34. Thomas KS, Muir KR, Doherty M, Jones AC, O’Reilly SC
and Bassey EJ. Home based exercise programme for knee
pain and knee osteoarthritis: randomised controlled trial.
BMJ 2002; 325: 752.
35. Ettinger WH, Jr, Burns R, Messier SP, et al. A random-
ized trial comparing aerobic exercise and resistance exer-
cise with a health education program in older adults with
knee osteoarthritis. The Fitness Arthritis and Seniors Trial
(FAST). JAMA 1997; 277: 25–31.
36. Baker KR, Nelson ME, Felson DT, Layne JE, Sarno R
and Roubenoff R. The efficacy of home based progres-
sive strength training in older adults with knee osteoar-
thritis: a randomized controlled trial. J Rheumatol 2001;
28: 1655–1665.
37. Fransen M, Crosbie J and Edmonds J. Physical therapy is
effective for patients with osteoarthritis of the knee: a ran-
domized controlled clinical trial. J Rheumatol 2001; 28:
156–164.
38. Quilty B, Tucker M, Campbell R and Dieppe P. Physio-
therapy, including quadriceps exercises and patellar taping,
for knee osteoarthritis with predominant patello-femoral
joint involvement: randomized controlled trial. J Rheumatol
2003; 30: 1311–1317.
39. Petrella RJ and Bartha C. Home based exercise therapy for
older patients with knee osteoarthritis: a randomized clini-
cal trial. J Rheumatol 2000; 27: 2215–2221.
40. Hinterwimmer S, Krammer M, Krotz M, et al. Cartilage
atrophy in the knees of patients after seven weeks of par-
tial load bearing. Arthritis Rheum 2004; 50: 2516–2520.