Osteoarthritis and Cartilage 22 (2014) 1090e1099

Effectiveness of continuous and pulsed ultrasound for the

management of knee osteoarthritis: a systematic review and network
meta-analysis
C. Zeng, H. Li, T. Yang, Z.-h. Deng, Y. Yang, Y. Zhang, X. Ding, G.-h. Lei*
Department of Orthopaedics, Xiangya Hospital, Central South University, Changsha, Hunan Province, China

a r t i c l e i n f o s u m m a r y

Article history: Background: To investigate the efficacy of continuous and pulsed ultrasound (US) in the management of
Received 12 March 2014 knee osteoarthritis (OA).
Accepted 28 June 2014 Design: This systematic review and network meta-analysis covered 12 trials in total. Electronic databases
including MEDLINE, Embase and Cochrane Library were searched through to identify randomized
Keywords: controlled trials comparing the two modes of US with control interventions (sham or blank) or with each
Ultrasound
other. Bayesian network meta-analysis was used to integrate both the direct and indirect evidences on
Pain
treatment effectiveness.
Function
Osteoarthritis
Results: Pulsed US (PUS) is more effective in both pain relief and function improvement when compared
Meta-analysis with the control group; but for continuous US (CUS), there is only a significant difference in pain relief in
comparison with the control group. In addition, no matter in terms of pain intensity or function at the
last follow-up time point, PUS always exhibited a greater probability of being the preferred mode.
However, the evidence of heterogeneity and the limitation in sample size of some studies could be a
potential threat to the validity of results.
Conclusions: Our findings indicated that PUS, with a greater probability of being the preferred mode, is
more effective in both pain relief and function improvement when compared with the control group.
However, CUS could only be considered as a pain relief treatment in the management of knee OA. The
findings also confirmed that none of these modes is dangerous.
Level of evidence: Level II, systematic review and network meta-analysis of randomized controlled trials.
© 2014 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Introduction
Osteoarthritis (OA) is a progressive rheumatic disease, the
incidence of which keeps rising with the growth of the ageing
population in many societies. About 9 million people suffer from
symptomatic, radiographically confirmed knee OA in the USA1. The
main complaints of knee OA include: pain during body moving or
weight bearing, stiffness, swelling, deformity and decreased
walking time and distance. Therefore, the relevant therapies
generally aim to relieve pain, improve functionality and mitigate
* Address correspondence and reprint requests to: G.-h. Lei, Department of Or-
thopaedics, Xiangya Hospital, Central South University, #87 Xiangya Road, Chang-
sha, Hunan 410008, China. Tel: 86-0731-84327326.
E-mail addresses: zengchao19880505@sina.com (C. Zeng), lgh9640@sina.cn,
125562108@qq.com (G.-h. Lei).
disability for enhancing the quality of life. As the prevalence of knee
OA increases2, the increasing rates of knee arthroplasty, which is
the only effective way of late phase, has made the identification of
effective conservative treatments a high priority.
Therapeutic ultrasound (US), based on the application of sound
waves in tissues, is one of the most widely used physical agents in
knee OA. There are two modes of US, continuous and pulsed.
Continuous US (CUS) generates thermal effects by stimulating the
process of tissue regeneration, changing cell membrane perme-
ability and increasing intracellular calcium, while pulsed US (PUS)
mainly produces non-thermal effects to increase tissue meta-
bolism, enhance fibrous tissue extensibility and elevate pain
threshold3e5. In the past few years, a rapidly growing interest in
testing the effects of US has been observed, but no agreement was
reached yet; especially, there is no exact conclusion on which
modality of US (continuous or pulsed) is more effective in pain
relief and functionality improvement of knee OA.

http://dx.doi.org/10.1016/j.joca.2014.06.028
1063-4584/© 2014 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099
Recently, the Osteoarthritis Research Society International
(OARSI) developed the guidelines (2014) for non-surgical man-
agement of knee OA6, which is the most up-to-date guideline till
present. The recommendation level of US for knee-only OA was
uncertain based on the results of two systematic reviews7,8 pub-
lished in 2010 and one randomized controlled trial (RCT)9 pub-
lished in 2012. Although these two previous systematic reviews
covered several RCTs (one of them reported a statistically signifi-
cant pain relief for knee OA7), evidence was limited because only six
RCTs were available at that time. It is worth mentioning that
whether the physical function was actually improved remains un-
clear. Furthermore, the authors admitted that the small sample size
and low quality of trials included in their mode subgroup analyses
(continuous or pulsed) inhibited further inferences7.
Network meta-analysis has been recently developed to evaluate
the relative effectiveness of several interventions and synthesize
evidence across a network of randomized trials. The mainly
strength of network meta-analysis is to combine the direct (studies
compared PUS with CUS directly) and indirect estimates (studies
compared PUS with CUS via control group) and calculated a mixed
effect size as the weighted average of the direct evidence and the
indirect evidence10. Comparing with traditional meta-analysis,
network meta-analysis is able to offer information on compari-
sons for which no trails exist. In our case, the application of indirect
evidence can improve the precision of the direct estimate by
shortening the width of the confidence intervals in contrast to
direct estimates alone11.
Considering the accumulation of new trials, this paper intended
to conduct an up-to-date systematic review and Bayesian network
meta-analysis covering all the available direct and indirect evi-
dences on US effects in order to generate a unified and coherent
analysis for all RCTs12e15. The objective of this study was to assess
the effects of US (continuous or pulsed) on pain relief and function
improvement for knee OA.
Materials and methods
Literature search
The electronic databases of Medline, Embase and Cochrane li-
brary were searched through using a series of logic combination of
keywords and text words related to OA, interested interventions
and randomized controlled trials (Appendix 1). The latest electronic
search was conducted in February 2014 to identify the references of
the retrieved papers and reviews. In addition, the following web-
sites were searched through manually to identify unpublished and
ongoing studies: Current Controlled Trials (http://www.controlled-
trials.com/), ClinicalTrials.gov (http://www.clinicaltrials.gov/)and
the World Health Organization International Clinical Trials Registry
(http://apps.who.int/trialsearch/Default.aspx).
Study selection
Two researchers reviewed all the retrieved abstracts and full
texts independently. Disagreements were resolved through dis-
cussions and consultations to another researcher. Those papers
meeting the following criteria were included in the analysis: (1)
randomized controlled trials; (2) patients with knee OA (3)
studies containing at least two of the following eligible treat-
ments: PUS, CUS and control group (blank or sham); (4) studies
reporting the pain or function outcomes of patients; (5) English
literature. On the other hand, the trials with unbalanced addi-
tional modality (e.g., education or exercise) between groups were
excluded.
1091
Quality assessment
Two researchers evaluated the methodological quality sepa-
rately. The modified oxford score16,17, a scale ranged from 0 to 7
according to the descriptions of randomization, concealment allo-
cation, blinding method and reporting of participant withdrawals,
was used to measure the methodological quality of included
studies.
Outcome measures
The primary goal of this study was to identify the effectiveness
of pain management and function improvement by applying the
PUS or CUS therapy respectively. The degree of pain intensity after
treatment was used as the measure of pain management effect. If a
study reported multiple pain scales, the highest one on the hier-
archy of pain scale related outcomes was selected, as described by
Jüni and colleagues18. The WOMAC function was the preferred
measure for function. If a study did not measure or report the
WOMAC function, WOMAC total, Lequesne Index or other func-
tional measurement scales were used in the analysis instead. If a
study reported outcomes at multiple time points after treatment,
only the data from the final follow-up time point was extracted for
analysis. The effect of pain management and function improvement
was expressed as the standard mean differences (SMD) between
different treatment arms. A negative value of SMD indicates lower
pain intensity and better status of function after treatment.
Statistical analysis
The random effect Bayesian network meta-analysis was per-
formed to compare the pain management and function improve-
ment effect between different modes of US therapy for knee OA.
The advantage of network meta-analysis is that it allows indirect
comparisons of interventions among primary trials. In this study,
the random effect Bayesian network meta-analysis was performed
using WinBUGS software (version 1.4.3, MRC Biostatistics Unit,
Cambridge, UK), R version 3.0.2 (The R Foundation for Statistical
Computing) and STATA software (version 11.0, StataCorp, College
Station, TX). The codes of random effect models for multi-arm trials
were available at http://www.mtm.uoi.gr/ (Appendix 2). Three
Markov chains ran simultaneously with different initial values
chosen arbitrarily. A total of 50,000 simulations were generated for
each of the three sets of initial values, and the first 10,000 simu-
lations were discarded due to the burn-in period. The pooled effect
sizes were reported from the median of the posterior distribution.
The corresponding 95% credible intervals with the 2.5th and 97.5th
percentiles of the posterior distribution were used in this study,
which could be interpreted in a way similar to the conventional 95%
confidence intervals. In order to estimate the network inconsis-
tency between indirect and direct estimates in each closed loop, the
absolute difference between the direct and indirect treatment ef-
fect estimates was calculated. Loops with the lower CI limit did not
reach the zero line were considered to represent statistically sig-
nificant inconsistency19. The fit of the model to data can be
measured by calculating the posterior mean residual deviance. If
the mean of the residual deviance is similar to the number of data
points of the model, it indicates that the model fits the data
adequately20. All treatments were then ranked based on the level of
effectiveness according to their posterior probabilities (first best,
second best, third best, etc.). The probability values were summa-
rized and reported as the surface under the cumulative ranking
(SUCRA)21. SUCRA is equal to 1 for the best treatment, and 0 for the
worst treatment.
1092 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099
The classic pairwise meta-analysis was also performed by using
STATA software (version 11.0, StataCorp, College Station, TX). The
heterogeneity was tested by Q statistics (P
0.05 was considered
heterogeneous) and I2 statistics, which measures the percentage of
the total variation across various studies (I2  50% was considered
heterogeneous). To evaluate the publication bias, Begg's tests22
were performed, where a P value less than 0.5 was considered to
reflect publication bias.
Results
Study selection and characteristics
Figure 1 showed the selection process of the included trials in
this study. A total of 1769 records were identified from the pre-
liminary database and website search and 993 records were
selected after removing duplicates. Then 927 records excluded with
reasons and 66 full-text trials were evaluated for eligibility. Even-
tually, 12 studies were included in this research9,23e33. The char-
acteristics of the included studies were listed in Table I. The
methodological quality assessment (Table II) showed that three
studies belonged to low quality (score
3), four belonged to me-
dium quality (scored 4 or 5), and the rest belonged to high quality
(scored 6 or 7). Data from eight studies involving a total of 525
patients with knee OA was available for network meta-ana-
lysis9,23e29. Figure 2 showed the network structure of the com-
parisons for the primary outcomes.
Fig. 1. Flowchart for the selection of included trials.
Pain intensity
The results of network meta-analysis and pairwise meta-
analysis were reported in Table III. PUS achieved a significantly
lower pain intensity compared to the control group (SMD: 0.59,
95%CI: 0.89 to 0.26), and CUS achieved a significantly better
effect of pain management compared to the control group
(SMD: 0.41, 95%CI: 0.67 to 0.07). However, there was no sig-
nificant difference between the PUS and CUS group in terms of pain
intensity after treatment (SMD: 0.18, 95%CI: 0.94 to 0.37). No
evidence of inconsistency between direct and indirect estimates
was found in this network meta-analysis. Evaluation of the good-
ness of fit indicated adequate fit with a posterior mean residual
deviance of 20.07 (19 data points). The distribution of probability of
each treatment for this outcome was shown in Fig. 3. PUS is most
likely to be the best treatment (93%) compared to CUS (56%) and the
control group (0%). Pairwise meta-analysis showed nearly the same
results (Table III). Significant evidence of heterogeneity was only
observed in the comparison between PUS and the control group
(P ¼ 0.07, I2 ¼ 54%). There was no publication bias among various
studies. The original pain outcomes of the studies which were not
included in meta-analysis were reported in Appendix 3.
Function
Table III also showed the outcomes of network meta-analysis for
function. It indicated that only the PUS therapy was significantly
more effective in terms of function improvement compared to the
 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099 1093

Table I
Characteristics of the included studies

Study Groups Balance* N Age Gender Mean Parameters of intervention Interested outcome Test timez
(years) (M/F) BMI (pain/function)

Loyola-S

anchez G1: PUS None 14 62.6 3/11 34 9.5 min  3 times  8 weeks, VAS/WOMAC function 8 weeks
2012 G2: sham 13 61.2 3/10 30.4 20% duty cycle, 1 MHz,
average temporal intensity:
0.2 W/cm2, therapeutic
dose: 112.5 J/cm2
Ulus 2012 G1: CUS Hot pack, 20 60.7 3/17 31.6 10 min  5 times  3 weeks, VAS/WOMAC function 0 week
G2: sham interferential 20 60.3 3/17 31.1 1 MHz, 1 W/cm2 and LI
current,
exercise
Tascioglu 2010 G1: CUS None 27 59.7 10/17 30 5 min  5 times  2 weeks, VAS/WOMAC total 2 days
G2: PUS 28 61.4 7/21 30.8 1 MHz, 2 W/cm2
G3: sham 27 60 9/18 28.7
€ o
Ozg € nenel 2009 G1: CUS None 34x 53.6 6/28 None 5 min  5 times  2 weeks, VAS/WOMAC function 2 days
G2: sham 33 56.2 7/26 1 MHz, 1 W/cm2
Huang 2005a G1: PUS Exercise 32 65y 27/113y None 5 min  3 times  8 weeks, VAS/LI 0, 1 year
G2: blank 31 1 MHz, 2.5 W/cm2
Huang 2005b G1: CUS Exercise, 48 62y 23/97y None 5 min  3 times  8 weeks, VAS/LI 0, 1 year
G2: PUS hot pack 56 1 MHz, continuous:
G3: blank 42 1.5 W/cm2; pulsed: 2.5 W/cm2
Cetin 2008 G1: CUS Exercise, 20x 57.6 Only 29.8 10 min  3 times  8 weeks, VAS/LI 0 week
G2: blank hot pack 20 61.1 Females 27.7 1 MHz, 1.5 W/cm2
Cakir 2013 G1: CUS Exercise 20 56.9 6/14 27.9 12 min  5 times  2 weeks, VAS/WOMAC function 0, 6 months
G2: PUS 20 58.2 4/16 30.9 CUS: 1 MHz, 1 W/cm2;
G3: sham 20 57.1 3/17 29.5 PUS: 1 MHz, 1 W/cm2
Yang PF 2011 G1: PUS None 50 58.3y 72/15y None 30 min  5 treatments VAS/LI 0,4 weeks
G2: sham 50
Falconer J 1992 G1: CUS None 34x 69.4x 8/26 BMI 30 min  12 treatments VAS/gait velocity 0, 12 weeks
G2: sham 35x 65.7x 11/35 (2e3 times per week over
4e6 weeks); increments of
0.1 W/cm2 to maximum 2.5 W/cm2
Külcü 2009 G1: CUS None 15 63.1 2/13 e 10 min  5 sessions  3 weeks; VAS 0 week
G2: blank 15 62 3/12 1 MHz, 1.5 W/cm2 WOMAC
Kapidzic 2011 G1: CUS Thermotherapy 40 60y e e 5 min, 3 weeks, 0.8 W/cm2 Likert's scale/WOMAC 0 week
G2: sham and exercises 50 function and LI

W, watt; ms, microsecond; N, number of subjects; M, male; F, female; BMI, Body Mass Index; VAS, visual analogue scale (0e10); WOMAC function, function subscale of
Western Ontario and McMaster University Osteoarthritis Index (0e68); WOMAC total, Western Ontario and McMaster University Osteoarthritis Index consists three subscales
(pain, stiffness and functional status, 0e96); LI, Lequesne index (0e26).
*
Usual cares which were balance between groups.
y
Only data for the whole trial is available.
z
0 means at the end of the treatment.
x
Only data from the baseline is available.

control group (SMD: 0.57, 95%CI: 1.12 to 0.01). There was no
significant difference between CUS and the control group
(SMD: 0.26, 95%CI: 0.79 to 0.28), and between PUS and CUS
(SMD: 0.31, 95%CI: 0.94 to 0.37) in terms of function improve-
ment. There was no evidence of inconsistency between the direct
and indirect evidences. The model had an adequate fit to data, with
a posterior mean residual deviance of 19.64 (19 data points).
Figure 4 displayed the distribution of probability of each treatment
ranked at each of the three possible positions. According to the
result of the posterior probability values of rank, it was found that
PUS is most likely (98%) to be the best treatment for function
improvement, followed by CUS (51%) and the control group (9%).
The pairwise meta-analysis also ended up with similar results as
the network meta-analysis in terms of function. However, there
was significant heterogeneity between the two direct comparisons
(PUS vs control: P ¼ 0.00, I2 ¼ 83%; PUS vs CUS: P ¼ 0.005, I2 ¼ 81%).
Based on the P values of Beggs' test, publication bias did not exist
among various studies. The functional results of the studies
excluded from this network meta-analysis were reported in
Appendix 3.
Adverse effects
There were only three trials that reported adverse effects. Two
of them claimed that no adverse effect related to the CUS
treatment was observed, and one study reported the adverse ef-
fect of electric shock/stinging sensation related to the PUS treat-
ment. However, there was no significant difference between the
intervention and the control group in terms of the incidence of
adverse effect.
Discussion
This network meta-analysis reviewed two kinds of therapeutic
US (continuous or pulsed) in terms of pain relief and function
improvement for patients with knee OA. It was found that PUS is
more effective in terms of both pain relief and function improve-
ment when compared with the control group (blank and sham), but
for CUS, the significant difference only exists in pain relief in
comparison with the control group. In addition, no matter in terms
of pain intensity or function at the last follow-up time point, PUS is
most likely to be the preferred mode.
Recently, a provincial survey conducted in Canada indicated
that a majority of physical therapists (85%) expressed interest in
the effectiveness of US in pain relief and physical function34.
However, the recommendations mentioned in the current clinical
practice guidelines have not arrived at any agreement which
against helping policy makers, service commissioners, and pro-
viders to judge the efficacy of US therapy. One guideline rec-
ommended US for OA management35; three guidelines were
1094 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099

Table II relief and function improvement) for patients with knee OA7,8.
Methodological evaluation of included studies With the accumulation of evidences, especially for the mode
Study Randomized Concealment Blinding Follow-up Total (continuous or pulsed) subgroup, this network meta-analysis
method allocation method score further supported that PUS could be the preferred mode in
Loyola-Sa
nchez 2012 2 2 2 1 7 terms of both pain relief and function improvement without
Ulus 2012 2 2 2 1 7 significant adverse effects.
Tascioglu 2010 1 2 2 1 6 In vitro studies have supported the effects of PUS in inducing
€ o
Ozg € nenel 2009 1 0 2 1 4
chondrocyte proliferation and matrix production in human artic-
Huang 2005a 1 2 2 1 6
Huang 2005b 1 2 2 1 6 ular cartilage45e49. Meanwhile, based on a number of researches
Cetin 2008 1 0 2 1 4 establishing animal models of cartilage injury, in vivo experiments
Cakir 2013 2 0 2 1 5 also showed that low intensity PUS was beneficial to cartilage
Yang 2011 1 0 0 1 2
health50e55. There was also a study suggesting that PUS with topical
Falconer 1992 1 0 2 1 4
Külcü 2009 2 0 0 1 3
lidocaine gel could induce greater anaesthetic effects compared to
Kapidzic 2011 1 0 0 1 2 CUS with topical lidocaine gel and lidocaine application alone56.
The findings of these studies were consistent with the results ob-
tained here. On the other hand, unlike the athermal mechanical
effects generated by the application of PUS, CUS aims to generate
thermal effects which could increase capillary permeability and
against the use of US36e38; three guidelines, including the latest tissue metabolism, enhance fibrous tissue extensibility, and espe-
one, regarded US as an uncertain appropriate modality6,39,40; cially elevate the pain threshold5,57. Likewise, CUS was proved to be
other four guidelines did not even mention about US as a treat- an effective way only in pain relief for patients with knee OA in this
ment option41e44. Two reviews published in 2010 involving five study.
and six randomized controlled trials respectively reached a As far as we know, this is the first network meta-analysis of
similar conclusion that therapeutic US might be beneficial (pain US for knee OA, which compiled evidences from both direct and
indirect comparisons for evaluating the relative effectiveness in
pain relief and function improvement. In addition, this study
overcame two limitations which were admitted in the previous
PUS
classical meta-analysis7. One is the effects of co-interventions
(including education or exercise) which could be effective in
pain relief and function improvement. The control group of this
3(147) study was limited to sham or blank control, and the additional
modality was required to be balanced between groups in order to
3(199) eliminate the potential impact of standard care. Under such
circumstance, it is meaningful to conclude that US was effective
as long as the results were significant. Otherwise, it is uncom-
5(224) 5(328) CUS fortable to draw a definite conclusion whether US is beneficial or
not when the results were non-significant. The other limitation
resolved is the powerful mode subgroup analysis which was
6(331)
conducted in this network meta-analysis. Furthermore, Beggs'
test suggested that publication bias was not observed among the
6(286) included studies.
Nevertheless, the limitations of this study should not be ignored.
Firstly, variations of US sessions, doses and differences in final
control follow-up time point might contribute to the significant evidence of
heterogeneity, particularly for the possible doseeresponse patterns
Fig. 2. Structure of network formed by interventions and their direct comparisons. The
lines between treatment nodes indicate the direct comparisons made within ran- which could greatly affect the performance of US. However,
domized trials. The width of the lines is proportional to the number of trials comparing because of the limited number of included trials and insufficient
each pair of treatments. Numbers represent numbers of trials (number of analyzed descriptions of treatment parameters, this study could not explore
patients) per comparison. Inside numbers are for pain outcome and outside numbers
it in depth. Fortunately, no obvious evidence of inconsistency was
are for function outcome.
observed in this network meta-analysis. Secondly, several included

Table III
Results of network meta-analysis and pairwise meta-analysis

Comparison Network meta-analysis, Pairwise meta-analysis, Heterogeneity (P/I2) Publication bias

SMD (95%CI) SMD (95%CI) (P value of Beggs' test)

PUS vs control
Pain 0.59 (0.89, 0.26) 0.55 (0.90, 0.21) 0.07/54% 0.086
Function 0.57 (1.12, 0.01) 0.63 (1.31, 0.06) 0.00/83% 0.462
CUS vs control
Pain 0.41 (0.67, 0.07) 0.40 (0.67, 0.14) 0.21/30% 1.000
Function 0.26 (0.79, 0.28) 0.24 (0.47, 0.00) 0.29/18% 1.000
PUS vs CUS
Pain 0.18 (0.49, 0.17) 0.13 (0.41, 0.15) 0.71/0% 1.000
Function 0.31 (0.94, 0.37) 0.33 (1.10, 0.45) 0.005/81% 0.296
 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099 1095

CUS PUS

1
.8

.8
.6

.6
.4

.4
Cumulative Probabilities
.2

.2
0

0
1 2 3 1 2 3

control
1
.8
.6
.4
.2
0

1 2 3

Graphs by Treatment Rank

Fig. 3. Rankings for pain intensity. Graph displays distribution of probabilities for each treatment. X-axis represents the possible rank of each treatment (from the best rank to worse
according to the outcomes), Y-axis represents the cumulative probability for each treatment to be the best option, among the best two options, among the best three options, and so on.

studies measured the pain or function parameters too shortly after
all treatment courses. It is uncertain whether these effects may
diminish after a long time. Thirdly, the low level of methodological
quality and the limitation in sample size of some studies could be a
potential threat to the validity of results. Fourthly, this study only
focused on the effectiveness of pain relief and function improve-
ment, it is difficult to measure other indices in this Bayesian
network meta-analysis.
Conclusion
Our findings indicated that PUS, with a greater probability of
being the preferred mode, is more effective in both pain relief and
function improvement when compared with the control group.
However, CUS could only be considered as a pain relief treatment in
the management of knee OA. The findings also confirmed that none
of these modes is dangerous.
Contribution of authors
All authors had full access to all the data in the study and take
responsibility for the integrity of the data and the accuracy of the
data analysis. CZ and GL contributed to the study concept and
design and drafted the manuscript. HL, TY and ZD contributed to
data collection. YY, YZ and XD contributed to preparation and
data analysis. GL contributed to revision of the manuscript. All

CUS PUS
1

1
.8

.8
.6

.6
.4

.4
Cumulative Probabilities
.2

.2
0

1 2 3 1 2 3

control
1
.8
.6
.4
.2
0

1 2 3
Rank
Graphs by Treatment

Fig. 4. Rankings for function. Graph displays distribution of probabilities for each treatment. X-axis represents the possible rank of each treatment (from the best rank to worse
according to the outcome), Y-axis represents the cumulative probability for each treatment to be the best option, among the best two options, among the best three options, and so on.
1096 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099

the authors contributed to the interpretation of the data and 7. coxarthro$.ti,ab,sh.

critically reviewed the manuscript for publication. 8. arthros$.ti,ab.
9. arthrot$.ti,ab.
10. ((knee$ or joint$) adj3 (pain$ or ach$ or discomfort$)).ti,ab.
Conflict of interest 11. ((knee$ or joint$) adj3 stiff$).ti,ab
The authors declare that they have no conflict of interest. 12. Or/1-11
13. randomized controlled trial.pt.
14. controlled clinical trial.pt.
Acknowledgements 15. randomi?ed.ab.
16. placebo.ab,tw.
This work was supported by Hunan Provincial Innovation 17. controlled.ti,ab
Foundation for Postgraduate, the Fundamental Research Funds 18. randomly.ti,ab.
for the Central Universities of Central South University, the Na- 19. trial.ti,ab.
tional Natural Science Foundation of China (Nos. 81201420, 20. groups.ti,ab
81272034), the Provincial Science Foundation of Hunan (No. 21. ((randomized controlled trials) or (random$ allocation) or
14JJ3032), the Scientific Research Project of the Development and (double blind) or (single blind)).tw
Reform Commission of Hunan Province (No. [2013]1199), the 22. ((singl$ or doubl$ or tripl$) and (mask$ or blind$)).tw
Scientific Research Project of Science and Technology Office of 23. Or/13-22
Hunan Province (No. 2013SK2018), the Doctoral Scientific Fund 24. exp ultrasonography/
Project of the Ministry of Education of China (No. 25. exp Ultrasonic Therapy/
20120162110036). 26. (ultrasound$ or ultrasonic$).tw.
27. short wave therapy.tw.
28. ultrasonograph$.tw.
29. Ultraso$.ti,ab,sh
Appendix 1. Search strategies for Pubmed, Ovid/MEDLINE, 30. Or/24-29
The Cochrane Library and Ovid/EMBASE database 31. 12 and 23 and 30

Pubmed

The Cochrane Library

1. osteoarthro*[tiab] or gonarthriti*[tiab] or gonarthro*[tiab] or
coxarthriti*[tiab] or coxarthro*[tiab] or osteo?arthritis[tiab]
#1 (osteoarthritis* OR osteoarthro* OR gonarthriti* OR gonar-
2. (knee*[tiab] or joint*[tiab]) and (pain*[tiab] or discomfort*
thro* OR coxarthriti* OR coxarthro* OR arthros* OR arthrot*
[tiab])
OR ((knee* OR joint*) near/3 (pain* OR ach* OR discomfort*))
3. (knee*[tiab] or joint*[tiab]) and stiff*[tiab]
OR ((knee*OR joint*) near/3 stiff*))
4. 1 or 2 or 3
#2 MeSH descriptor Osteoarthritis explode all trees
5. randomized[tiab]
#3 #1 or #2
6. placebo[tiab]
#4 short wave therapy in Clinical Trials
7. controlled[tiab]
#5 ultrasonograph in Clinical Trials
8. random*[tiab]
#6 (ultrasound* or ultrasonic*) in Clinical Trials
9. trial*[tiab]
#7MeSH descriptor Ultrasonography explode all trees
10. groups[tiab]
#8 MeSH descriptor Short-Wave Therapy explode all trees
11. ((singl*[tiab] or doubl*[tiab] or tripl*[tiab]) and (mask*[tiab]
#9 #4 or #5 or #6 or #7 or #8
or blind*[tiab]))
#10 #3 and #9
12. Or/5-11
13. Ultrasonography[tiab]
14. Ultrasonic[tiab]
15. ultrasound*[tiab] or ultrasonic*[tiab] Ovid/EMBASE
16. short[tiab] and wave[tiab]
17. ultrasonograph*[tiab] 1. osteoarthriti$.ti,ab,sh.
18. Ultraso*[tiab] 2. osteoarthro$.ti,ab,sh.
19. Or 13-18 3. osteo?arthritic.ti,ab,sh.
20. 4 and 12 and 19 4. gonarthriti$.ti,ab,sh.
5. gonarthro$.ti,ab,sh.
6. coxarthriti$.ti,ab,sh.
Ovid/MEDLINE 7. coxarthro$.ti,ab,sh.
8. arthros$.ti,ab.
1. osteoarthriti$.ti,ab,sh. 9. arthrot$.ti,ab.
2. osteoarthro$.ti,ab,sh. 10. ((knee$ or joint$) adj3 (pain$ or ach$ or discomfort$)).ti,ab.
3. osteo?arthritic.ti,ab,sh. 11. ((knee$ or joint$) adj3 stiff$).ti,ab
4. gonarthriti$.ti,ab,sh. 12. Or/1-11
5. gonarthro$.ti,ab,sh. 13. randomized controlled trial.pt.
6. coxarthriti$.ti,ab,sh. 14. controlled clinical trial.pt.
 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099 1097

15. randomi?ed.ab.
16. placebo.ab,tw.
17. controlled.ti,ab
18. randomly.ti,ab.
19. trial.ti,ab.
20. groups.ti,ab
21. ((randomized controlled trials) or (random$ allocation) or
(double blind) or (single blind)).tw
22. ((singl$ or doubl$ or tripl$) and (mask$ or blind$)).tw
23. Or/13-22
24. exp ultrasonography/
25. exp Ultrasonic Therapy/
26. (ultrasound$ or ultrasonic$).tw.
27. short wave therapy.tw.
28. ultrasonograph$.tw.
29. Ultraso$.ti,ab,sh
30. Or/24-29
31. 12 and 23 and 30

Appendix 2. WinBUGS codes of random effect models for

multi-arm trials
1098 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099

Appendix 3. Results of the studies not included in the network meta-analysis

Study Groups Pain scale/function scale Outcome of pain Outcome of function

Original data Significance of Original data Significance of

difference difference

Yang PF 2011 G1: PUS VAS/LI Effect index P ¼ 0.000 Effect index P ¼ 0.000
G2: placebo G1: 0.36 ± 0.28 G1: 0.31 ± 0.42
G2: 0.10 ± 0.19 G2: 0.03 ± 0.11
Falconer J 1992 G1: CUS VAS/gait velocity NR NS NR NS
G2: placebo
Külcü 2009 G1: CUS VAS/WOMAC Median (range) P < 0.0001 Median (range) P < 0.0001
G2: blank G1: 2 (0e6) G1: 11.5 (0e26)
G2: 5 (2e10) G2: 24 (18e30)
Kapidzic 2011 G1: CUS Likert's scale/WOMAC Difference of before P < 0.05 Difference of before P < 0.05
G2: placebo function and LI and after treatment and after treatment
G1: 0.83 ± 0.51 G1: 0.61 ± 0.55
G2: 0.53 ± 0.40 G2: 0.38 ± 0.70

NS: not significant, NR: not report.

References 14. Lumley T. Network meta-analysis for indirect treatment

1. Lawrence RC, Felson DT, Helmick CG, Arnold LM, Choi H,
Deyo RA, et al, National Arthritis Data Workgroup. Estimates of
the prevalence of arthritis and other rheumatic conditions in
the United States. Arthritis Rheum 2008;58:26e35.
2. Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG,
Jordan JM, et al. Osteoarthritis: new insights, part 1: the dis-
ease and its risk factors. Ann Intern Med 2000;133:635e46.
3. Johns LD. Nonthermal effects of therapeutic ultrasound: the
frequency resonance hypothesis. J Athl Train 2002;37:293e9.
4. Rand SE, Goerlich C, Marchand K, Jablecki N. The physical
therapy prescription. Am Fam Physician 2007;76:1661e6.
5. Baker KG, Robertson VJ, Duck FA. A review of therapeutic ul-
trasound: biophysical effects. Phys Ther 2001;81:1351e8.
6. McAlindon TE, Bannuru RR, Sullivan MC, Arden NK,
Berenbaum F, Bierma-Zeinstra SM, et al. OARSI guidelines for
the non-surgical management of knee osteoarthritis. Osteo-
arthritis Cartilage 2014;22:363e88.
7. Loyola-Sanchez A, Richardson J, MacIntyre NJ. Efficacy of ul-
trasound therapy for the management of knee osteoarthritis: a
systematic review with meta-analysis. Osteoarthritis Cartilage
2010;18(9):1117e26.
8. Rutjes AW, Nuesch E, Sterchi R, Juni P. Therapeutic ultrasound
for osteoarthritis of the knee or hip. Cochrane Database Syst
Rev 2010;(1):CD003132.
9. Ulus Y, Tander B, Akyol Y, Durmus D, Buyukakıncak O, Gul U,
et al. Therapeutic ultrasound versus sham ultrasound for the
management of patients with knee osteoarthritis: a random-
ized double-blind controlled clinical study. Int J Rheum Dis
2012;15:197e206.
10. Cipriani A, Higgins JP, Geddes JR, Salanti G. Conceptual and
technical challenges in network meta-analysis. Ann Intern
Med 2013;159:130e7.
11. Bucher HC, Guyatt GH, Griffith LE, Walter SD. The results of
direct and indirect treatment comparisons in meta-analysis of
randomized controlled trials. J Clin Epidemiol 1997;50:
683e91.
12. Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of
multiple treatments: combining direct and indirect evidence.
BMJ 2005;331:897e900.
13. Caldwell DM, Welton NJ, Ades AE. Mixed treatment compari-
son analysis provides internally coherent treatment effect es-
timates based on overviews of reviews and can reveal
inconsistency. J Clin Epidemiol 2010;63:875e82.
comparisons. Stat Med 2002;21:2313e24.
15. Lu G, Ades AE. Combination of direct and indirect evidence
in mixed treatment comparisons. Stat Med 2004;23:
3105e24.
16. Elia N, Tramer MR. Ketamine and postoperative pain-a quan-
titative systematic review of randomized trials. Pain 2005;113:
61e70.
17. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ,
Gavaghan DJ, et al. Assessing the quality of reports of ran-
domized clinical trials: is blinding necessary? Control Clin
Trials 1996;17:1e12.
18. Jüni P, Reichenbach S, Dieppe P. Osteoarthritis: rational
approach to treating the individual. Best Pract Res Clin Rheu-
matol 2006;20:721e40.
19. Veroniki A, Salanti G. Graphical exploration of inconsistency in
a network of randomized trials in R., http://www.mtm.uoi.gr/
index.php/how-to-do-an-mtm/10-how-to-do-an-mtm/18-
inconsistency; 2012.
20. Spiegelhalter DJ, Best NG. Bayesian approaches to multiple
sources of evidence and uncertainty in complex cost-
effectiveness modelling. Stat Med 2003;22:3687e709.
21. Salanti G, Ades AE, Ioannidis JP. Graphical methods and nu-
merical summaries for presenting results from multiple-
treatment meta-analysis: an overview and tutorial. J Clin
Epidemiol 2011;64:163e71.
22. Begg CB, Mazumdar M. Operating characteristics of a rank
correlation test for publication bias. Biometrics 1994;50:
1088e101.
23. Cakir S, Hepguler S, Ozturk C, Korkmaz M, Isleten B, Atamaz FC.
Efficacy of therapeutic ultrasound for the management of knee
osteoarthritis: a randomized, controlled, and double-blind
study. Am J Phys Med Rehabil 2014;93:405e12.
24. Loyola-S
anchez A, Richardson J, Beattie KA, Otero-Fuentes C,
Adachi JD, MacIntyre NJ. Effect of low-intensity pulsed ul-
trasound on the cartilage repair in people with mild to
moderate knee osteoarthritis: a double-blinded, randomized,
placebo-controlled pilot study. Arch Phys Med Rehabil
2012;93:35e42.
25. Tascioglu F, Kuzgun S, Armagan O, Ogutler G. Short-term
effectiveness of ultrasound therapy in knee osteoarthritis. J Int
Med Res 2010;38:1233e42.
26. Ozgo € nenel L, Aytekin E, Durmuşoglu G. A double-blind trial of
clinical effects of therapeutic ultrasound in knee osteoarthritis.
Ultrasound Med Biol 2009;35:44e9.
 C. Zeng et al. / Osteoarthritis and Cartilage 22 (2014) 1090e1099
27. Huang MH, Yang RC, Lee CL, Chen TW, Wang MC. Preliminary
results of integrated therapy for patients with knee osteoar-
thritis. Arthritis Rheum 2005;15:812e20.
28. Huang MH, Lin YS, Lee CL, Yang RC. Use of ultrasound to in-
crease effectiveness of isokinetic exercise for knee osteoar-
thritis. Arch Phys Med Rehabil 2005;86:1545e51.
29. Cetin N, Aytar A, Atalay A, Akman MN. Comparing hot pack,
short-wave diathermy, ultrasound, and TENS on isokinetic
strength, pain, and functional status of women with osteoar-
thritic knees: a single-blind, randomized, controlled trial. Am J
Phys Med Rehabil 2008;87:443e51.
30. Kapidzic S. Measurement of therapeutic effect of ultrasound
on knee osteoarthritis; double blind study. Ann Phys Rehabil
Med 2011;54S:e180e3.
31. Yang PF, Li D, Zhang SM, Wu Q, Tang J, Huang LK, et al. Efficacy
of ultrasound in the treatment of osteoarthritis of the knee.
Orthop Surg 2011;3:181e7.
32. Falconer J, Hayes KW, Chang RW. Effect of ultrasound on
mobility in osteoarthritis of the knee. A randomized clinical
trial. Arthritis Care Res 1992;5:29e35.
33. Külcü DG, Gülşen G, Altunok EÇ. Short-term efficacy of pulsed
electromagnetic field therapy on pain and functional level in
knee osteoarthritis: a randomized controlled study. Turk J
Rheumatol 2009;24:144e8.
34. MacIntyre NJ, Busse JW, Bhandari M. Physical therapists in
primary care are interested in high quality evidence regarding
efficacy of therapeutic ultrasound for knee osteoarthritis: a
provincial survey. ScientificWorldJournal 2013;2013:348014.
35. Conaghan P, Birrell F, Burke M, Cumming J, Dickson J, Dieppe P,
et al. Osteoarthritis: National Clinical Guideline for Care and
Management in Adults. London, UK: Royal College of Physi-
cians; 2008.
36. The Royal Australian College of General Practitioners. Guide-
line for the Non-surgical Management of Hip and Knee Oste-
oarthritis, http://www.racgp.org.au/guidelines/osteoarthritis;
2009.
37. Vogels E, Hendricks H, Van Baar M, Dekker J, Hopman-Rock M,
Oostendorp R, et al. Clinical Practice Guidelines for Physical
Therapy Inpatients with Osteoarthritis of the Hip or Knee,
http://www.cebp.nl/media/m1; 2003.
38. Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD,
Arden N, et al. OARSI recommendations for the management of
hip and knee osteoarthritis, part II: OARSI evidence-based,
expert consensus guidelines. Osteoarthritis Cartilage
2008;16:137e62.
39. Jordan KM, Arden NK, Doherty M, Bannwarth B, Bijlsma JW,
Dieppe P, et al, Standing Committee for International Clinical
Studies Including Therapeutic Trials ESCISIT. EULAR Recom-
mendations 2003: an evidence based approach to the man-
agement of knee osteoarthritis: report of a Task Force of the
standing Committee for International Clinical Studies
Including Therapeutic Trials (ESCISIT). Ann Rheum Dis
2003;62:1145e55.
40. Philadelphia Panel. Philadelphia panel evidence-based clinical
practice guidelines on selected rehabilitation interventions for
knee pain. Phys Ther 2001;81:1675e700.
41. Hochberg MC, Altman RD, April KT, Benkhalti M, Guyatt G,
McGowan J, et al, American College of Rheumatology. Amer-
ican College of Rheumatology 2012 recommendations for the
1099
use of nonpharmacologic and pharmacologic therapies in
osteoarthritis of the hand, hip, and knee. Arthritis Care Res
(Hoboken) 2012;64:465e74.
42. Ottawa Panel. Ottawa panel evidence-based clinical practice
guidelines for therapeutic exercises and manual therapy in the
management of osteoarthritis. Phys Ther 2005;85:907e71.
43. Richmond J, Hunter D, Irrgang J, Jones MH, Snyder-Mackler L,
Van Durme D, et al, American Academy of Orthopaedic Sur-
geons. American academy of orthopaedic surgeons clinical
practice guideline on the treatment of osteoarthritis (OA) of
the knee. J Bone Joint Surg Am 2010;92:990e3.
44. The Arthritis Society. Getting a Grip on Arthritis Best Practice
Guidelines, http://www.arthritis.ca/gettingagrip; 2004.
45. Nolte PA, Klein-Nulend J, Albers GH, Marti RK, Semeins CM,
Goei SW, et al. Low intensity ultrasound stimulates endo-
chondral ossification in vitro. J Orthop Res 2001;19:301e7.
46. Korstjens CM, Nolte PA, Burger EH, Albers GH, Semeins CM,
Aartman IH, et al. Stimulation of bone cell differentiation by
low-intensity ultrasound e a histomorphometric in vitro
study. J Orthop Res 2004;22:495e500.
47. Ingber DE. Cellular mechanotransduction: putting all the
pieces together again. FASEB J 2006;20:811e27.
48. Parvizi J, Parpura V, Greenleaf JF, Bolander ME. Calcium
signaling is required for ultrasound-stimulated aggrecan syn-
thesis by rat chondrocytes. J Orthop Res 2002;20:51e7.
49. Parvizi J, Wu CC, Lewallen DG, Greenleaf JF, Bolander ME. Low-
intensity ultrasound stimulates proteoglycan synthesis in rat
chondrocytes by increasing aggrecan gene expression.
J Orthop Res 1999;17:488e94.
50. Cook SD, Salkeld SL, Popich-Patron LS, Ryaby JP, Jones DG,
Barrack RL. Improved cartilage repair after treatment with low
intensity pulsed ultrasound. Clin Orthop Relat Res
2001;391(Suppl l):S231e43.
51. Cui JH, Park K, Park SR, Min BH. Effects of low-intensity ul-
trasound on chondrogenic differentiation of mesenchymal
stem cells embedded in polyglycolic acid: an in vivo study.
Tissue Eng 2006;12:75e82.
52. Huang MH, Ding HJ, Chai CY, Huang YF, Yang RC. Effects of
sonication on articular cartilage in experimental osteoarthritis.
J Rheumatol 1997;24:1978e84.
53. Huang MH, Yang RC, Ding HJ, Chai CY. Ultrasound effect on
level of stress proteins and arthritic histology in experimental
arthritis. Arch Phys Med Rehabil 1999;80:551e6.
54. Gurkan I, Ranganathan A, Yang X, Horton Jr WE, Todman M,
Huckle J, et al. Modification of osteoarthritis in the guinea pig
with pulsed low-intensity ultrasound treatment. Osteoar-
thritis Cartilage 2010;18:724e33.
55. Naito K, Watari T, Muta T, Furuhata A, Iwase H, Igarashi M,
et al. Low-intensity pulsed ultrasound (LIPUS) increases the
articular cartilage type II collagen in a rat osteoarthritis model.
J Orthop Res 2010;28:361e9.
56. Ebrahimi S, Abbasnia K, Motealleh A, Kooroshfard N, Kamali F,
Ghaffarinezhad F. Effect of lidocaine phonophoresis on sensory
blockade: pulsed or continuous mode of therapeutic ultra-
sound? Physiotherapy 2012;98:57e63.
57. Srbely JZ, Dickey JP, Lowerison M, Edwards AM, Nolet PS,
Wong LL. Stimulation of myofascial trigger points with ultra-
sound induces segmental antinociceptive effects: a random-
ized controlled study. Pain 2008;139:260e6.