Академический Документы
Профессиональный Документы
Культура Документы
research-article2016
CRE0010.1177/0269215516683000Clinical RehabilitationWang et al.
CLINICAL
Original Article REHABILITATION
Clinical Rehabilitation
meta-analysis
Abstract
Objective: To evaluate the efficacy and safety of short-wave therapy with sham or no intervention for
the management of patients with knee osteoarthritis.
Methods: We searched the following databases from their inception up to 26 October 2016: MEDLINE,
CENTRAL, EMBASE, Physiotherapy Evidence Database, CINAHL and OpenGrey. Studies included
randomized controlled trials compared with a sham or no intervention in patients with knee osteoarthritis.
The results were calculated via standardized mean difference (SMD) and risk ratio for continuous variables
outcomes as well as dichotomous variables, respectively. Heterogeneity was explored by the I2 test and
inverse-variance random effects analysis was applied to all studies.
Results: Eight trials (542 patients) met the inclusion criteria. The effect of short-wave therapy on pain was
found positive (SMD, −0.53; 95% CI, −0.84 to −0.21). The pain subgroup showed that patients received
pulse modality achieved clinical improvement (SMD, –0.83; 95% CI, –1.14 to −0.52) and the pain scale in
female patients decreased (SMD, −0.53; 95% CI, −0.98 to −0.08). In terms of extensor strength, short-
wave therapy was superior to the control group (p < 0.05, I2 = 0%). There was no significant difference in
the physical function (SMD, −0.16; 95% CI, −0.36 to 0.05). For adverse effects, there was no significant
difference between the treatment and control group.
Conclusion: Short-wave therapy is beneficial for relieving pain caused by knee osteoarthritis (the pulse
modality seems superior to the continuous modality), and knee extensor muscle combining with isokinetic
strength. Function is not improved.
Keywords
Knee osteoarthritis, short-wave therapy, diathermy, systematic review, meta-analysis
scores,19 Lequesne indexes,20 arthritis impact (SMD) was employed to estimate the treatment
measurement scale21 and other functional effect. All outcomes were compared applying the
scales. Secondary outcomes were stiffness of random-effect model with 95% confidence inter-
knee, muscle strength and adverse events vals (CI). The high and low dose short-wave ther-
caused by short-wave therapy. apy subgroups in one trial were combined into a
single group, and the subgroups containing both
placebo short-wave therapy as well as control sub-
Data extraction and
group without therapy in the same trial were also
management combined into one single group.10 This was the for-
Pairs of review authors extracted data indepen- mula for calculation following the Cochrane
dently. We resolved any disagreements by consen- Handbook for systematic reviews (version 5.1.0).
sus or by consulting a third review author. A test The posttreatment means and deviations of trials
data form was applied to extract the following data: were pooled on the basis that the trials were suffi-
Study participants (first author, year published, cient. Also, subgroup analysis was conducted
clinical setting, samples size, gender and age), based on short-wave therapy mode and partici-
short-wave therapy interventions used (frequency, pants’ gender. We planned to test the heterogeneity
wave-mode, intensity, dose and device), combined between comparable trials applying a standard I2
intervention, outcome measures (characteristics of test considered statistically significant with a P
pain, physical function, safety outcomes, other value <0.05. The I2 statistic value thresholds23 were
measures mentioned and follow-up). For any effect interpreted as follows: 0% to 40%, might not be
sizes that were not reported or could not be calcu- important; 30% to 60%, may represent moderate
lated in the study, the authors would be contacted. heterogeneity; 50% to 90%, may represent sub-
The extracted data were entered into a database stantial heterogeneity; and 75% to 100%, consider-
(Table 1, available online). able heterogeneity. If the heterogeneity was
considered to have serious effects on the validity of
results, we would give up combining the data.
Assessment of risk of bias Also, any evidence of heterogeneity would be
included studies identified so that it could determine whether there
The risk of bias was assessed in accordance with were significant differences between trials that
methods of the Cochrane Collaboration’s ‘Risk of might trigger heterogeneity. If the I2 test was not
bias’. Figures generated by the RevMan 5.3 (http:// significant (p > 0.05), sensitivity analysis would be
ims.cochrane.org/revman) were presented to pro- conducted to investigate the potential sources of
vide summary assessments of the risk of bias. heterogeneity. Such analysis was performed
Moreover, the methodological quality of the through omitting one study and identifying its
included studies was also assessed by the impact on the overall pooled estimate. P < 0.05 was
Physiotherapy Evidence Database (PEDro)22 tool, regarded as statistically significant. All statistical
providing a score between 0 and 10 points (http:// analyses were performed by RevMan 5.3.
www.pedro.fhs.usyd.edu.au/scale_item.html).
Scores of trials were either provided by the PEDro Results
website, or scored by our research group. Two
authors independently assessed the risk of bias in
Study selection
included trials. We resolved any disagreement by Eight studies (542 patients) were included in this
consensus or by consulting a third review author. research. Figure 1 shows the flow details of the
included trials selected in different phases. In the
phase of full text articles assessed for eligibility,
Statistical analyses two trials10,24 that were conducted in the same hos-
Where available and appropriate, with respect pital by the same group at different time (2008 and
to continuous data, standardized mean difference 2011) met the inclusion criteria, with the subject
4 Clinical Rehabilitation
Figure 1. Flow of information through the different phases of our systematic review.
collections of the former covered by the latter. ranged from 42 to 85 years old. Four trials9–11,26
Unfortunately, we were not able to get in touch included only females, while the other four tri-
with the authors. In order to draw a more conserva- als6,27–29 contained both females and males, with a
tive result, we included the latest trial, in accord- greater proportion of females compared with
ance with the previous systematic review.12 males. The duration of intervention varied between
Preferred Reporting Items for Systematic reviews two to eight weeks, and the sessions of which
and Meta-Analyses (PRISMA) guidance25 were ranged from six to 24 treatments. The frequency of
confirmed during this research flow figure. short-wave therapy was 27.12 HZ, with exception
of one trial27 (27 HZ). Furthermore, the frequency
applied in another trial28 was not mentioned. The
Description of included studies output intensity of all trials ranged from 1.8 W to
Eight parallel groups randomized placebo con- 20 W, while three studies9,26,28 did not present the
trolled studies published between 2005 and 2012 specific intensity. For short-wave mode, four tri-
were included. Three trials were conducted in als6,9,11,26 applied the continuous short-wave ther-
Turkey,6,9,26 one in each of United Kingdom,27 apy (we could not make contact with the authors of
Italy,28 Brazil,10 Israel29 and Thailand.11 In total, the one trial,26 but the previous systematic review’s
studies included 542 participants who were author12 did, and hence we employed their opin-
enrolled from a clinical setting. The ages of patients ion), while four others10,27–29 applied the pulse
Wang et al. 5
(one28 of which was inferred through the author’s trial10 conducted both a control group without
objective). Two trials10,29 included two treatment treatment and a placebo short-wave therapy with-
groups that contained both high- and low-dose out output. The other three trials9,26,28 involved the
short-wave therapy, while the other six trials additional modality or modalities alone without
involved only one treatment group. When it came blinding the patients.
to the control groups, no short-wave therapy was Four trials9,11,28,29 conducted a less than three-
given to three trials.9,26,28 Four trials6,11,27,29 included month follow-up, while one trial6 conducted both a
a placebo or sham short-wave intervention and the three-month and six-month follow-up. Only one
device of all of them was operationalized without trial10 was reported with a 12-month follow-up.
current output or power, appearing similar in Three trials6,10,11 performed intention-to-treat anal-
appearance to the treatment group device. One ysis in outcomes analysis. As for financing, only
trial10 involved two control groups, receiving no one trial27 received the funding support supplied by
short-wave therapy treatment and placebo short- their hospital trust fund. The other seven tri-
wave therapy without output. The description and als6,9–11,26,28,29 did not mention information whether
characteristics of the included trials are summa- they were supported by any commercial organiza-
rized in Table 1, available online. tions or other funds.
Figure 3. Forest plot of trials comparing the effects of short-wave therapy and control changes on knee
osteoarthritis pain. (a) The aggregated results of pain changes (six trials). (b) The aggregated results of pain changes
in groups contained female only (three trials). (c) The aggregated results of pain changes in groups contained
both female and male (three trials). (d) The aggregated results of pain changes in groups applied pulse short-wave
therapy (three trials). (e) The aggregated results of pain changes in groups applied continuous short-wave therapy
(three trials).
Physical function. All of the eight included trials assess physical function, high scores showing high
assessed physical function. The data supplied in the level function opposite to the other tools (WOMAC,
articles of Rattanachaiyanont (2008)11 and Cantar- Lequesne indexes or arthritis impact measurement
ini (2007)28 were insufficient. Although efforts scale). Therefore, this study was excluded. As a
were made to make attempts to contact both authors, result, six trials involving 377 patients were aggre-
we failed to obtain physical function results from gated for meta-analysis, indicating no beneficial
Cantarini. However, Rattanachaiyanont did provide effects. We also conducted a subgroup analysis
us with specific data. One study10 employed the based on gender and different modes (pulse mode
Knee Injury and Osteoarthritis Outcome Score to and continuous mode). Both gender and different
Wang et al. 7
Figure 4. Forest plot of trials comparing the effects of short-wave therapy and control on physical function
changes. (a) The aggregated results of physical function changes (six trials). (b) The aggregated results of physical
function changes in groups only contained female people (three trials). (c) The aggregated results of physical
function changes in groups contained both female and male people (three trials). (d) The aggregated results of
pain changes in groups applied pulse short-wave therapy (two trials). (e) The aggregated results of pain changes in
groups applied continuous short-wave therapy (four trials).
modes subgroups analysis suggested that short- The pooled results of these trials were not in favour
wave therapy was not able to make any improve- of short-wave therapy (Figure 5).
ment in physical function without heterogeneity
(Figure 4). Muscle strength. Three trials assessed muscle
strength with a focus on isokinetic knee motion.
Stiffness. Three trials9,11,29 conducted the WOMAC Two of these trials conducted strength at three
stiffness subscale. We contacted one author11 to speeds (60, 120 and 180°/s), while the other one
obtain the specific data as the original article did not only assessed the speed at 90°/s. Thus, we aggre-
manifest sufficient information about the results. gated the two trials in three different speeds
8 Clinical Rehabilitation
Figure 5. Forest plot of trials comparing the effects of short-wave therapy and control on WOMAC stiffness
subscale changes (three trials).
involving both extensor and flexor. The extensor therapy group, while the other11 was observed with
strength in all three speeds showed significant light events in four patients similar with the pla-
improvement in the treatment group combining with cebo group, showing no statistical differences. No
isokinetic exercise, without heterogeneity (p < 0.05, adverse event was reported in the other six trials.
I2 = 0%). Additionally, no significant difference was Safety results manifested that short-wave therapy
found in the flexor strength of both groups in all was a safe treatment, and the risk ratio could not be
three speeds with very low heterogeneity (p > 0.05, calculated.
I2 < 28%) (Appendix 2, available online).
Figure 6. Forest plot of trials comparing the follow-up effects of short-wave therapy and control. (a) The
aggregated no more than three months pain follow-up results of pain changes (three trials). (b) The aggregated
more than three months pain follow-up results (two trials). (c) The aggregated no more than three months physical
function follow-up results (four trials). (d) The aggregated no more than three months WOMAC stiffness subscale
follow-up results (three trials).
assessment of knee osteoarthritis pain employing therapy subgroups were combined together to
WOMAC is easily confounded by physical disabil- make up the treatment group. In terms of compari-
ity.17 To decrease the heterogeneity and potential son groups, both placebo short-wave therapy and
bias, we adopted the unidimensional pain question- the control subgroup without therapy in the same
naires to evaluate pain, and function measurement study were also combined into one single group,
to assess physical function, respectively. We also while the previous systematic review chose the
contacted with the authors to obtain the outcomes strongest positive effect group to pool the results.
of two trials.11,28 One11 of the two trials was We excluded one trial30 of individuals with both
included in the previous review without pooled hip and knee osteoarthritis in our study, also differ-
effect size. In addition, different dose short-wave ent from the previous review.14
10 Clinical Rehabilitation
The positive effect on pain relief was in favour being performed in the previous systematic review.
of short-wave therapy, and the pooled pain esti- Three trials included in our study assessed muscle
mate was unchanged by the removal of each study, strength. Two of the three trials conducting strength
showing a robust outcome without changing the at the three speeds (60, 120 and 180°/s) could be
heterogeneity. In clinical practice, which type of pooled for meta-analysis, while the other trial27
short-wave mode is much more effective in pain only reported the 90°/s speed. As suggested by the
relief for knee osteoarthritis patients is an impor- pooled results, short-wave therapy brought bene-
tant problem the physicians want to know. Hence, fits to the knee extensor muscle combining with
different modes of short-wave therapy were aggre- isokinetic strength, showing no improvement of
gated in this study. The results suggested that pulse the knee flexor muscle. Though the main effect of
short-wave therapy was of superiority compared short-wave therapy is not to improve muscle
with the continuous mode when it comes to reliev- strength, the effect of combining it with isokinetic
ing pain. Such a result was robust without hetero- strength is valuable in clinical experiment, for the
geneity, exhibited by sensitivity analysis. Clinically, important role of extensor in knee osteoarthritis.
whether the short-wave effects are different in dif- Recent reviews32,33 regarded the impaired muscle
ferent genders is another problem of physicians’ function as an important risk factor for develop-
concern, as more women are epidemiologically ment of knee osteoarthritis, and the meta-analysis34
affected by knee osteoarthritis than men.1 Hence, also demonstrated that knee extensor muscle weak-
we pooled the trials containing only female people ness was associated with an increased risk of devel-
and trials containing both female and male people. oping knee osteoarthritis in both men and women.
The outcomes were indicative for the phenomenon, The finding of our study was indicative of the
which was that short-wave therapy seems effective potential of the combination of short-wave therapy
in women with knee osteoarthritis. and isokinetic exercise and they are becoming an
In the physical function assessment, six trials effective treatment for the improvement of exten-
were pooled for analysis results. The results sor strength of knee osteoarthritis patients.
revealed that four trials used the WOMAC physical When it comes to interpreting the results of our
function subscale, one trial applied Lequesne study, some limitations should be cautiously con-
indexes, and the other employed the arthritis impact sidered. First, though we searched six main data-
measurement scale. We performed sensitivity anal- bases and the references of the included trials, we
yses and subgroup analysis, and the aggregated may miss some other grey literature and hand
results did not favour short-wave therapy in the search of relevant literature, which may cause
improvement of physical function. Our study basi- selection bias. Second, there are only a limited
cally agreed with the previous results. The meas- number of trials added compared with the previ-
urement of the WOMAC stiffness subscale was ous review. So, the evidence was limited owing to
only reported in three trials. The aggregated result the small number of included studies and hetero-
did not support short-wave therapy with moderate geneity. Further high-quality studies are required
heterogeneity. As joint stiffness is one of the most to provide stronger evidence. However, the stabil-
common symptoms in osteoarthritis,31 we sug- ity of the results of the sensitivity analysis con-
gested clinical trials are needed to assess whether firms that the main outcomes of this study are
short-wave therapy can improve the stiffness of robust. Third, there is variability in treatment
knee osteoarthritis people. We also pooled the fol- duration, number of sessions, intensity of output
low-up data for meta-analysis, which was missing and other parameters in the included trials, which
in the previous review. The aggregated results indi- may cause heterogeneity and also limit the evi-
cated that short-wave therapy could act as an dence. Fourth, the included trials were only writ-
immediate pain relief strategy after treatment. ten in English, though we searched for studies
In terms of muscle strength, both extensor and without restriction of languages, which may also
flexor were conducted in all reported speeds, produce a selection bias. Last, the follow-up data
whereas different from only the extensor of 60°/s were not sufficient, especially the long-term item,
Wang et al. 11
and the pooled results indicated moderate to sub- States: Impact of extending indications for total knee
stantial heterogeneity. arthroplasty. Arthritis Care Res. 2015; 67: 203–215.
4. Smith SR, Deshpande BR, Collins JE, Katz JN and Losina
E. Comparative pain reduction of oral non-steroidal anti-
inflammatory drugs and opioids for knee osteoarthritis:
Systematic analytic review. Osteoarth Cart 2016; 24:
Clinical messages 962–972.
•• Short-wave therapy is beneficial for 5. Shields N, Gormley J and O’Hare N. Short-wave dia-
relieving pain of knee osteoarthritis. thermy: Current clinical and safety practices. Physiother
Res Int 2002; 7: 191–202.
•• Pulse mode short-wave therapy seems 6. Atamaz FC, Durmaz B, Baydar M, et al. Comparison of
superior to continuous modality. the efficacy of transcutaneous electrical nerve stimulation,
•• Short-wave therapy increases knee exten- interferential currents, and shortwave diathermy in knee
sor strength. osteoarthritis: A double-blind, randomized, controlled, mul-
ticenter study. Arch Phys Med Rehabil 2012; 93: 748–756.
7. Rawe IM. The case for over-the-counter shortwave ther-
apy: Safe and effective devices for pain management.
Acknowledgements Pain Management 2014; 4: 37–43.
The authors thank Vilai Kuptniratsaikul MD and Professor 8. McAlindon TE, Bannuru RR, Sullivan MC, et al. OARSI
Antonella Fioravanti for providing the initial data. guidelines for the non-surgical management of knee oste-
oarthritis. Osteoarth Cart 2014; 22: 363–388.
9. Akyol Y, Durmus D, Alayli G, et al. Does short-wave
Contributors diathermy increase the effectiveness of isokinetic exercise
HaiMing Wang and Chi Zhang were responsible for the on pain, function, knee muscle strength, quality of life,
literature search and data extraction. ChengFei Gao and and depression in the patients with knee osteoarthritis?: A
randomized controlled clinical study. Eur J Phys Rehabil
SiYi Zhu performed the blinding literature search and
Med 2010; 46: 325–336.
the data extraction. Quan Wei and Lijie Yang were in
10. Fukuda TY, Alves da Cunha R, Fukuda VO, et al. Pulsed
charge of data interpretation and technical guidance. shortwave treatment in women with knee osteoarthritis: A
ChengQi He made important intellectual contribution in multicenter, randomized, placebo-controlled clinical trial.
research design and manuscript revision. Phys Ther 2011; 91: 1009–1017.
11. Rattanachaiyanont M and Kuptniratsaikul V. No additional
Declaration of Conflicting Interests benefit of shortwave diathermy over exercise program for
knee osteoarthritis in peri-/post-menopausal women: An
The author(s) declared no potential conflicts of interest equivalence trial. Osteoarth Cart 2008; 16: 823–828.
with respect to the research, authorship, and/or publica- 12. Laufer Y and Dar G. Effectiveness of thermal and ather-
tion of this article. mal short-wave diathermy for the management of knee
osteoarthritis: A systematic review and meta-analysis.
Osteoarth Cart 2012; 20: 957–966.
Funding
13. Laufer Y and Dar G. Response to Letter to the Editor:
The author(s) disclosed receipt of the following financial Comment on Laufer et al. entitled “Effectiveness of ther-
support for the research, authorship, and/or publication mal and athermal short-wave diathermy for the manage-
of this article: The financial support for the research, ment of knee osteoarthritis: A systematic review and
authorship and/or publication of this article was spon- meta-analysis”. Osteoarth Cart 2014; 22: 607–608.
sored by National Natural Science Foundation of China 14. Zeng C, Gao SG and Lei GH. Comment on Laufer et al.
[Grant Number 81572236]. entitled “Effectiveness of thermal and athermal short-
wave diathermy for the management of knee osteoarthri-
tis: A systematic review and meta-analysis”. Osteoarth
Reference Cart 2014; 22: 605–606.
1. Woolf AD and Pfleger B. Burden of major musculo- 15. Altman R, Asch E, Bloch D, et al. Development of cri-
skeletal conditions. Bull World Health Org 2003; 81: teria for the classification and reporting of osteoarthritis.
646–656. Classification of osteoarthritis of the knee. Diagnostic
2. Bennell KL, Hall M and Hinman RS. Osteoarthritis year and Therapeutic Criteria Committee of the American
in review 2015: Rehabilitation and outcomes. Osteoarth Rheumatism Association. Arthritis Rheum 1986; 29:
Cart 2016; 24: 58–70. 1039–1049.
3. Losina E, Paltiel AD, Weinstein AM, et al. Lifetime medi- 16. Gentelle-Bonnassies S, Le Claire P, Mezieres M, Ayral
cal costs of knee osteoarthritis management in the United X and Dougados M. Comparison of the responsiveness
12 Clinical Rehabilitation
of symptomatic outcome measures in knee osteoarthritis. 25. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA state-
Arthritis Care Res 2000; 13: 280–285. ment for reporting systematic reviews and meta-analyses of
17. Hawker GA, Mian S, Kendzerska T and French M. studies that evaluate health care interventions: Explanation
Measures of adult pain: Visual Analog Scale for Pain and elaboration. Ann Intern Med 2009; 151: W65–94.
(VAS Pain), Numeric Rating Scale for Pain (NRS Pain), 26. Cetin N, Aytar A, Atalay A and Akman MN. Comparing
McGill Pain Questionnaire (MPQ), Short-Form McGill hot pack, short-wave diathermy, ultrasound, and TENS on
Pain Questionnaire (SF-MPQ), Chronic Pain Grade Scale isokinetic strength, pain, and functional status of women
(CPGS), Short Form-36 Bodily Pain Scale (SF-36 BPS), with osteoarthritic knees: A single-blind, randomized,
and Measure of Intermittent and Constant Osteoarthritis controlled trial. Am J Phys Med Rehabil 87: 443–451.
Pain (ICOAP). Arthritis Care Res 2011; 63 Suppl 11: 27. Callaghan MJ, Whittaker PE, Grimes S and Smith L. An
S240–252. evaluation of pulsed shortwave on knee osteoarthritis
18. Ornetti P, Dougados M, Paternotte S, Logeart I and using radioleucoscintigraphy: A randomised, double blind,
Gossec L. Validation of a numerical rating scale to assess controlled trial. Joint Bone Spine 2005; 72: 150–155.
functional impairment in hip and knee osteoarthritis: 28. Cantarini L, Leo G, Giannitti C, Cevenini G, Barberini P
Comparison with the WOMAC function scale. Annals and Fioravanti A. Therapeutic effect of spa therapy and
Rheumatic Dis 2011; 70: 740–746. short wave therapy in knee osteoarthritis: A randomized,
19. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J single blind, controlled trial. Rheumatol Int 27: 523–529.
and Stitt LW. Validation study of WOMAC: a health status 29. Laufer Y, Zilberman R, Porat R and Nahir AM. Effect
instrument for measuring clinically important patient rel- of pulsed short-wave diathermy on pain and function of
evant outcomes to antirheumatic drug therapy in patients subjects with osteoarthritis of the knee: A placebo-con-
with osteoarthritis of the hip or knee. J Rheumatology trolled double-blind clinical trial. Clin Rehabil 2005; 19:
1988; 15: 1833–1840. 255–263.
20. Lequesne MG, Mery C, Samson M and Gerard P. 30. Moffett JA, Richardson PH, Frost H and Osborn A. A
Indexes of severity for osteoarthritis of the hip and knee. placebo controlled double blind trial to evaluate the
Validation—value in comparison with other assessment effectiveness of pulsed short wave therapy for osteoar-
tests. Scan J Rheumatology 1987; 65: 85–89. thritic hip and knee pain. Pain 67: 121–127.
21. Meenan RF, Gertman PM and Mason JH. Measuring 31. Arya RK and Jain V. Osteoarthritis of the knee joint: An
health status in arthritis. The arthritis impact measurement overview. J Indian Academy Clin Med 2013; 14: 154–162.
scales. Arthritis Rheum 1980; 23: 146–152. 32. Roos EM, Herzog W, Block JA and Bennell KL. Muscle
22. Maher CG, Sherrington C, Herbert RD, Moseley AM and weakness, afferent sensory dysfunction and exercise in
Elkins M. Reliability of the PEDro scale for rating qual- knee osteoarthritis. Nature Rev Rheumatol 2011; 7: 57–63.
ity of randomized controlled trials. Phys Ther 2003; 83: 33. Bennell KL, Wrigley TV, Hunt MA, Lim B-W and
713–721. Hinman RS. Update on the role of muscle in the genesis
23. Higgins JPT GS. Cochrane Handbook for Systematic and management of knee osteoarthritis. Rheumatic Dis
Reviews of Interventions Version 5.1.0 [updated March Clinics N Am 2013; 39: 145–176.
2011]. Available from www.cochrane-handbook.org2011. 34. Øiestad BE, Juhl CB, Eitzen I and Thorlund JB. Knee
24. Fukuda TY, Ovanessian V, Cunha RAD, et al. Pulsed extensor muscle weakness is a risk factor for development
short wave effect in pain and function in patients with of knee osteoarthritis. A systematic review and meta-
knee osteoarthritis. J App Res 2008; 8: 189–198. analysis. Osteoarth Cart 2015; 23: 171–177.