Accepted Article

DR. CUN-ZHI LIU (Orcid ID : 0000-0001-8031-5667)

Article type : Full Length

DR. Cun-Zhi Liu (Orcid ID: 0000-0001-8031-5667)

Article type: Full Length

Title: Efficacy of intensive acupuncture versus sham acupuncture in knee osteoarthritis: A

randomized controlled trial

Jian-Feng Tu, MD PhD,1,2 Jing-Wen Yang, MD PhD,1 Guang-Xia Shi, MD PhD,1 Zhang-Sheng
Yu, PhD,3,4 Jin-Ling Li, MD PhD,1 Lu-Lu Lin, MD PhD,1 Yu-Zheng Du, MD,5 Xiao-Gang Yu,
MD PhD,6 Hui Hu, MD PhD,7 Zhi-Shun Liu, MD PhD,8 Chun-Sheng Jia, MD,9 Li-Qiong Wang,
PhD,1 Jing-Jie Zhao, MD,10 Jun Wang, MD PhD,11 Tong Wang, MD,12 Yang Wang, PhD,13
Tian-Qi Wang, MD PhD,1 Na Zhang, MD PhD,1 Xuan Zou, MD,1 Yu Wang, MD,2 Jia-Kai Shao
MD,2 Cun-Zhi Liu, MD PhD1,2

Affiliations:
1 Acupuncture Research Center, School of Acupuncture-Moxibustion and Tuina, Beijing
University of Chinese Medicine, Beijing 100029, China
2 Department of Acupuncture and Moxibustion, Beijing Hospital of Traditional Chinese Medicine
Affiliated to Capital Medical University, Beijing 100010, China
3 School of Life Sciences and Biotechnology and SJTU-Yale Joint Center for Biostatistics,
Shanghai Jiao Tong University, Shanghai 200240, China
This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1002/ART.41584
This article is protected by copyright. All rights reserved
 4 Centre for Biomedical Data Science, Shanghai Jiao Tong University, Shanghai 200240, China
Accepted Article
5 Department of Acupuncture and Moxibustion, First Teaching Hospital of Tianjin University of
Traditional Chinese Medicine, Tianjin 300112, China
6 Department of Acupuncture and Moxibustion, Beijing Hospital of Traditional Chinese and
Western Medicine, Beijing 100039, China
7 Department of Acupuncture and Moxibustion, Dongfang Hospital, Beijing University of
Chinese Medicine, Beijing 100078, China
8 Department of Acupuncture and Moxibustion, Guang'an Men Hospital, China Academy of
Chinese Medical Sciences, Beijing 100053, China
9 Hebei University of Chinese Medicine, Heibei 050091, China
10 Department of Traditional Chinese Medicine, Beijing Friendship Hospital, Capital Medical
University, Beijing 100050, China
11 Department of Acupuncture and Moxibustion, Dongzhimen Hospital, Beijing University of
Chinese Medicine, Beijing 100700, China
12 Department of orthopedics, Institute of Acupuncture and Moxibustion, China Academy of
Chinese Medicine Sciences, Beijing 100700, China
13 School of Mathematical Sciences and SJTU-Yale Joint Center for Biostatistics, Shanghai Jiao
Tong University 200240, Shanghai, China

Running head: Acupuncture for knee osteoarthritis

Corresponding Author: Cun-Zhi Liu, MD, Acupuncture Research Center, School of
Acupuncture-Moxibustion and Tuina, Beijing University of Chinese Medicine, No. 11, Bei San
Huan Dong Lu, Chaoyang District, Beijing 100029, China. E-mail: [email protected]

Acknowledgments
We are grateful for all the patients who consented to participate in the trial. We acknowledge
Tian-Hong Cui, MD, Beijing QiHuang Medicine Clinical Research Centre, for monitoring the
trial, Yang Wang, PhD, State Key Laboratory of Translational Cardiovascular Medicine, for

This article is protected by copyright. All rights reserved

 generating the randomization sequence. We thank Jia-Li Jiao, MAS, Department of Statistics,
Accepted Article
Shanghai Jiao Tong University, for the guidance of statistical analyses. We acknowledge Marc
Fisher (Harvard Medical School, USA) and Ke-Hua Zhou (State University of New York at
Buffalo, USA) for proofreading the manuscript. They were not compensated for their
contributions.

Funding
This trial was funded by Beijing Municipal Science & Technology Commission
(D171100003217003) and Beijing Municipal Administration of Hospitals (XMLX201607). The
funders of the study had no role in study design, data collection, data analysis, data interpretation,
or writing of the report. The corresponding author (Cun-Zhi Liu) had full access to all the data in
the study and had final responsibility for the decision to submit for publication.

This article is protected by copyright. All rights reserved

 ABSTRACT
Accepted Article
Objective: To assess the efficacy of intensive acupuncture (3 times weekly for 8 weeks) versus
sham acupuncture for knee osteoarthritis (KOA).
Methods: In this multicenter randomized sham-controlled trial, participants with KOA were
randomly assigned to receive electro-acupuncture (EA), manual acupuncture (MA) or sham
acupuncture (SA) 3 times weekly for 8 weeks. Participants, outcome assessors and statisticians
were masked to treatment group assignment. The primary outcome was the response rate, which is
the proportion of participants who simultaneously achieved minimal clinically important
improvement in pain and function at week 8. The primary analysis was analyzed by the Z-test for
proportions with the modified intention-to-treat population, which included all randomized
participants who have at least one post-baseline measurement.
Results: Out of 480 participants recruited in the trial, 442 were evaluated for efficacy. The
response rates at week 8 were 60.3% (91/151), 58.6% (85/145), and 47.3% (69/146) in the EA,
MA, and SA groups, respectively. The between-group differences were 13.0% (97.5%CI, 0.2% to
25.9%; P=0.0234) for EA vs SA and 11.3% (97.5%CI, -1.6% to 24.4%; P=0.0507) for MA vs SA.
The response rates in EA and MA groups were both significantly higher than the SA group at
weeks 16 and 26.
Conclusion: Among patients with KOA, compared with SA, intensive EA resulted in less pain
and better function at week 8 and these effects persisted though week 26. Intensive MA had no
benefit for KOA at week 8, although it showed benefits during follow-up.
Trial registration: ClinicalTrials.gov NCT03366363

This article is protected by copyright. All rights reserved

 INTRODUCTION
Accepted Article
Knee osteoarthritis (KOA) is one of the leading causes of chronic pain and disability in older
adults (1). The prevalence of KOA was found to be 16% among the general population in the late
20th to early 21st centuries and had doubled since the mid-20th century (2). The socioeconomic
burden of KOA is high, amounting to between 1.0% and 2.5% of gross domestic product for
westernized countries (3). Worse still, this situation will be exacerbated with increased longevity
and the burgeoning obesity epidemic, especially in the United States and other developed
countries.

Since no disease-modifying pharmaceutical agents have been approved, current KOA treatments
are mainly symptomatic. Exercise therapy is considered as one of the key elements for the
management of KOA; however, wide implementation and long-term adherence of exercise therapy
remain challenges (1). Non-steroidal anti-inflammatory drugs (NSAIDs) are widely prescribed for
KOA; however, side effects of NSAIDs, like upper gastrointestinal tract bleed and worsening
kidney functions, limit their use on many occasions (4, 5). Additional treatments may include
eventual total knee replacement, but it has minimal effects on quality adjusted life years at the
group level (6). Identifying novel therapies for KOA, outside the common pharmacological agents
and eventual surgery, remains an important priority for research.

Although the number of acupuncture research studies on KOA has grown markedly (7), its
efficacy remains a subject of controversy (8-10). Factors contributing to this controversy likely
include the dose of acupuncture in previous trials. A systematic review suggests that higher
dosages of acupuncture are related to better treatment outcomes in KOA (11). Frequency of
acupuncture sessions is an important determinant of acupuncture dosage (12), and unfortunately, it
was inadequate in many of prior studies (less than 2 sessions per week) (13-15). Our recent trial
suggests that 3 sessions per week of acupuncture immediately improved knee pain and
dysfunction compared with 1 session per week and the benefits of 3 sessions per week persist
throughout the 16 weeks’ follow-up (16). Electro-acupuncture (EA) and manual acupuncture

This article is protected by copyright. All rights reserved

 (MA) are both frequently used in clinical practices. EA combines MA with additional electric
Accepted Article
stimuli (17). The present trial was designed to evaluate the effects of intensive acupuncture (EA
and MA), compared with sham acupuncture (SA), for pain and joint function in participants with
KOA.

PATIENTS AND METHODS

Study design
This multicenter randomized controlled trial was approved by the institutional review boards of
Beijing Hospital of Traditional Chinese Medicine Affiliated to Capital Medical University (Ref.
No 2017BL-077-01) and 8 other study hospitals before participant enrollment, and was conducted
in accordance with the Declaration of Helsinki. The protocol (NCT03366363) (18) and the
statistical analysis plan (19) were published previously. Protocol changes were summarized in
Supplementary (Methods section). And there was no interim analysis conducted prior to these
changes being made. Written informed consent was provided by all participants before
randomization. Participants did not receive any financial incentive during the trial, but all
acupuncture treatments and X-ray examinations were free of charge.

Participants
Participants were recruited through a multi-modal strategy, including a social media networking
tool (WeChat), newspaper, patient database and posters at community service centers.
Participants, diagnosed with KOA according to the American College of Rheumatology clinical
criteria (20), were eligible if they were 45 to 75 years old; reported knee pain for longer than 6
months; had radiologic confirmation of osteoarthritis (Kellgren-Lawrence score II or III) (21); and
had a pain score of greater than 4 (range, 0-10; higher scores indicate greater pain) on the numeric
rating scale (NRS) (22). The exclusion criteria were history of knee arthroplasty for the most
painful knee or waiting for any knee surgery for either knee, knee pain caused by other diseases,
arthroscopy in the last 12 months or intra-articular injection within the previous 6 months,
acupuncture treatment in the last 3 months, serious acute or chronic organic diseases or psychiatric

This article is protected by copyright. All rights reserved

 disorders, blood coagulation disorders, cardiac pacemaker, metal allergy or needle phobia,
Accepted Article
pregnancy or breastfeeding, or participated in other clinical trials in the past 3 months.

Randomization and masking

Eligible participants were randomized to three groups by a central stratified randomization system
at a 1:1:1 ratio. The randomization sequence was generated using SAS version 9.3 (SAS Institute,
Cary, NC, USA) by an independent statistician and was stratified by enrollment hospital with a
randomized block size of 6 or 9. Acupuncturists were not blinded for the responsibility of
delivering acupuncture treatment. However, participants, outcome assessors, and statisticians were
all blinded to group assignment.

Treatments
Each participant was treated in a single treatment room during the trial. Both knees were needled
for patients with bilateral KOA, whereas only the affected knee was acupunctured for those with
unilateral osteoarthritis symptoms (23). Thirty-minute sessions of treatment were delivered three
times weekly for 8 weeks, with 24 sessions in total for all groups. Disposable sterile needles (0.25
mm × 25-40 mm, Hwato, Suzhou, China), and HANS-200 electro-acupuncture devices (Nanjing
Jisheng Medical Co, Ltd) were used. Twenty-three registered acupuncturists (11.7 ± 4.9 years of
experience) performed the procedures. All acupuncturists were trained in standardized operating
procedures prior to the start of the study; this train included locations of acupoints and
non-acupoints and manipulation of needles. Paracetamol (Tylenol, Shanghai Johnson & Johnson
Pharmaceuticals, Ltd) was provided as needed, except during 48 hours before outcome
measurements.

The acupuncture prescription was based on traditional Chinese medicine and was developed from
clinical practices and expert consensus (23). Five obligatory acupoints and three adjunct acupoints
were used in both EA and MA groups. The obligatory acupoints included Dubi (ST35), Neixiyan
(EX-LE5), Ququan (LR8), Xiyangguan (GB33) and an Ashi point (the point where the participant

This article is protected by copyright. All rights reserved

 felt the most pain). Adjunct acupoints were selected from the acupoint pool (Supplementary Table
Accepted Article
1) at the acupuncturist’s discretion depending on which meridian was affected. De qi was required
for both EA and MA groups; it was a composite of sensations including soreness, numbness,
distention and heaviness. In the SA group, eight non-acupoints (Supplementary Table 2), which
were away from the conventional acupoints or meridians, were superficially penetrated (2-3 mm in
depth) without needle manipulation for de qi. The locations of acupoints and non-acupoints are
presented in Supplementary Figure 1.

Electrodes from the EA apparatus were attached to the handles of needles at LR8, GB33 and two
adjunct acupoints in both EA and MA groups or four non-acupoints in the SA group by a research
assistant. In the EA group, a dilatational wave of 2/100 Hz was chosen and the electric current was
gradually increased until the needles began to vibrate slightly. During treatment, the power light of
the EA apparatus was switched on in the MA and SA groups but with no electric current output.
For blinding participants, all participants were told that the electricity may be under the threshold
that human could sense. Details on the interventions were summarized in Supplementary Table 3.

Outcomes
The more painful knee defined at baseline was assessed throughout the entire study for
participants with bilateral KOA, whereas only the affected knee was assessed for participants with
unilateral KOA (24). Participants completed questionnaires at baseline, weeks 4, 8, 16 and 26 after
randomization.

The primary outcome was the response rate at week 8. The response rate (25) was defined as the
proportion of participants who simultaneously achieved minimal clinically important improvement
(MCII) on the NRS (22) and Western Ontario and McMaster Universities Osteoarthritis Index
(WOMAC) function subscale (26). The MCII of the 11-point NRS was set at 2 points, which was
extrapolated from a 19.9-mm MCII reported for the 100-mm Visual Analogue Scales (14, 27). The
MCII of the 68-point WOMAC function subscale (Likert version 3.1) was set at 6 points, which

This article is protected by copyright. All rights reserved

 was extrapolated from a 9.1-point MCII reported for the standardized 100-point WOMAC
Accepted Article
function subscale (14, 27). The procedures to define a responder were presented in Supplementary
Figure 2.

Secondary outcomes included average knee pain over the previous week using the NRS (22)
(score range, 0-10) and WOMAC (26) pain subscale (score range, 0-20); average knee function
over the previous week using the WOMAC (26) function subscale (score range, 0-68); average
knee stiffness over the previous week using the WOMAC (26) stiffness subscale (score range,
0-8); quality of life using the 12-item Short Form Health Survey (SF-12) (28) (score range, 0-100);
paracetamol use; patient’s global assessment (PGA) using a 5-point ordinal scale (ranging from
none improved to extremely improved) (29).

To evaluate the success of blinding, all participants were asked to guess which type of acupuncture
they received at weeks 4 and 8 after randomization. The credibility and expectancy of participants
was measured using the Credibility/Expectancy Questionnaire (30) within 5 min after the first
acupuncture session. All adverse events were appropriately managed and documented throughout
the trial. Based on the potential relationship with acupuncture, adverse events were categorized as
treatment-related or non-treatment-related.

Statistical Analysis
The following two null hypotheses were tested simultaneously for the primary outcome:
H1: There is no difference in the response rate between the EA and SA groups.
H2: There is no difference in the response rate between the MA and SA groups.
Based on our previous pilot study (16), the response rates of the EA, MA and SA groups at week 8
were expected to be 70%, 60%, and 40%, respectively. To detect a between-group difference of
20% in response rate, 128 participants per group were required for 80% power (2-sided test;
adjusted α = 0.025 for two comparisons). With an estimated dropout rate of 20%, 160 participants
per group were required (480 participants in total).

This article is protected by copyright. All rights reserved

Accepted Article
The primary analysis was based on the modified intention-to-treat principle, which included all
randomized participants who have at least one post-baseline measurement. Missing data of the
NRS and WOMAC function scores at week 8 were imputed using the baseline values for the
primary outcome. Two comparisons of the response rate (EA vs SA, MA vs SA, respectively)
were analyzed by the Z-test for proportions. The comparison of response rates between EA and
MA was not the aim of this study.

For the NRS, a comparison among three groups was assessed by a mixed-effect model with
repeated measurement (MMRM) analysis using NRS scores at all follow up time points as the
dependent variable, treatment as the main factor, baseline value as a covariate, and a random
intercept to model within-subject correlation. The same approach was used to analyze WOMAC
subscales and SF-12. Chi-square test was used for the group comparisons of patient global
assessments and prevalence of acupuncture-related adverse events. All participants with at least
one acupuncture session were included in the safety analyses. The James blinding index (range
0-1) was used to assess the blinding (0, total absence of blinding; 1, complete blinding; 0.5,
completely random blinding).

Sensitivity analyses were conducted for the response rates in the per-protocol set which included
all participants who completed the treatment and follow-up without major violations. To examine
the robustness of the conclusion in sensitivity analyses, three different schemes were used to deal
with missing data for the response rates: the last observation carried forward approach, the listwise
deletion, and the multiple imputation (Monte Carlo Markov chain) (31). A preplanned subgroup
analysis was conducted to test the interactive effect between the Kellgren-Lawrence grade and
treatment using a logistic model.

A post hoc analysis of response rate between groups at week 8 was performed in the (not
modified) intention-to-treat set which included all randomized participants. An additional post hoc

This article is protected by copyright. All rights reserved

 subgroup analysis of primary outcome was conducted based on history of previous acupuncture
Accepted Article
use. History of previous acupuncture use was dichotomized, and an interaction term with
treatment group was fitted to the analysis models to obtain the P value for interaction.

All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA) and
R software 3.6.0. For the primary outcome, Bonferroni adjustment was used for multiple
comparisons (setting α = 0.025) for each independent comparison of EA vs SA and MA vs SA.
For multiple comparisons of secondary outcomes, no adjustment was made.

RESULTS
Participants
Between December 25, 2017 and October 10, 2018, 1243 patients were screened (Supplementary
Figure 3). A total of 480 patients were included and randomly assigned to the three study groups
(Figure 1). Among these participants, 156 in the EA group, 155 in the MA group, and 157 in the
SA group received at least one session of treatment and were included in the safety analysis; 151
participants in the EA group, 145 in the MA group, and 146 in the SA group had at least one
post-baseline measurement and were included in the primary analysis. At week 26, 407 (84.8%)
participants completed the study. The baseline characteristics and results of the
Credibility/Expectancy Questionnaire were shown in Table 1.

Efficacy
For the primary outcome, the response rates at week 8 were 60.3% (91/151 participants) in the EA
group, 58.6% (85/145 participants) in the MA group, and 47.3% (69/146 participants) in the SA
group. The between-group differences were 13.0% (97.5% CI, 0.2% to 25.9%; P = 0.0234) for EA
vs SA and 11.3% (97.5% CI, -1.6% to 24.4%; P = 0.0507) for MA vs SA (Table 2). Similar
results were observed in the sensitivity analyses (Supplementary Table 4). The response rates in
the EA and MA groups were both significantly higher than the SA group at weeks 16 and 26
(Table 2). The trajectory of response rates over time was presented in Figure 2.

This article is protected by copyright. All rights reserved

Accepted Article
Both EA and MA significantly decreased the NRS and WOMAC pain score, compared with SA,
during 26 weeks. The between-group differences of NRS at week 8 were -0.79 (95% CI, -1.2 to
-0.37) for EA vs SA and -0.51 (95% CI, -0.93 to -0.09) for MA vs SA. EA significantly improved
the WOMAC function and stiffness scores, compared with SA, during 26 weeks. However, no
significant difference was found in either WOMAC function or WOMAC stiffness scores between
the MA and the SA groups. The between-group differences of WOMAC function score at week 8
were -2.51 (95% CI, -4.30 to -0.72) for EA vs SA and -0.95 (95% CI, -2.79 to 0.88) for MA vs
SA. There was no significant difference at each time point among three groups, except at week 16
for SF-12 physical health and at week 26 for SF-12 mental health (Table 2). Similar results were
observed in the modified MMRM model, which included center effect (Supplementary Table 5).
Patient global assessments at weeks 4, 8 and 16 in the EA and MA groups were better than the SA
group (Supplementary Table 6). Four participants in the EA group, 2 in the MA group, and 10 in
the SA group used paracetamol during the study. In subgroup analyses, neither EA nor MA had
interactive effects between radiologic grade and treatment on the response rate (P=0.85 and 0.57,
respectively) (Supplementary Figure 4).

Post hoc analysis

The results in the (not modified) intention-to-treat analyses were similar with the primary analysis
(Supplementary Table 4). Neither EA nor MA had interactive effect between history of previous
acupuncture use and treatment on the response rate (P=0.71 and 0.83, respectively)
(Supplementary Figure 5) After excluding patients who had received acupuncture in the past year,
the between-group differences were 14.0% (97.5% CI, 0.8% to 27.2%; P = 0.018) for EA vs SA
and 12.3% (97.5% CI, -3.0% to 23.0%; P = 0.040) for MA vs SA (Supplementary Table 7).

Adverse events
Acupuncture-related adverse events, including subcutaneous hematoma, post-needling pain, and
pantalgia occurred in 11.5% (18/156) of the participants in the EA group, 14.2% (22/155) in the

This article is protected by copyright. All rights reserved

 MA group, and 10.8% (17/157) in the SA group (Table 3). All acupuncture-related adverse events
Accepted Article
were mild and resolved spontaneously in the subsequent week. Adverse events unrelated to
acupuncture were infrequent (Supplementary Table 8).

Blinding
Participants were unaware of assigned treatments at week 4 (James blinding index 0.64, 95% CI
0.50 to 0.78). The success of blinding was maintained at week 8 (James blinding index 0.63, 0.46
to 0.79; Table 4).

DISCUSSION
This multicenter, randomized sham-controlled trial demonstrated that participants with KOA who
received the intensive EA had a significantly higher response rate than SA. The therapeutic
benefits of EA for pain and function in KOA were maintained through week 26. The difference
between MA and SA groups did not meet the statistical significance level at week 8. Interestingly,
by week 16, MA group also showed benefits against SA in pain and function of KOA, sustaining
through week 26.

In this trial, an expected between-group difference of 20% in response rate was used for sample
size calculation, but the actual differences were 13.0% (EA vs SA) and 11.3% (MA vs SA). The
study power might be lower than anticipated and the confidence intervals were thus wide in the
present trial. An addition of 10% of response rate over control group was suggested to be
clinically meaningful for renal colic (32). Although a clinically meaningful response rate for KOA
is not available in the literature, the difference of 11.3%, which indicates the number
needed-to-treat (NNT) (33) of 9, is acceptable in clinical practices. Moreover, the expected 20% in
response rate was achieved during the follow-up period because of sustaining effects of
acupuncture in both EA and MA groups. These persistent effects are in line with recent
meta-analyses, which concluded that acupuncture was effective for chronic pain without
importantly decrease in efficacy over 12 months (34, 35).

This article is protected by copyright. All rights reserved

Accepted Article
Nonetheless, Hinman et al (14) reported that acupuncture did not confer benefits over SA for pain
or function for KOA. The reason for these seemly contradictory findings may be inadequate
dosage of acupuncture in the previous trials (36). Acupuncture has a dose-dependent effect. The
dosage of acupuncture can be measured with the frequency and total number of acupuncture
sessions, the number of needles, retention time and mode of stimulation, location of needles and
treatment timing (12). Hence, the intensive acupuncture program was designed in this trial, and it
is also in accordance with the expert consensus on standardized acupuncture treatment for KOA
(23). EA provides continuous electrical stimulation and may provide additional dosage of
acupuncture. Not surprisingly, a network meta-analysis suggested that EA might be one of the best
choices in all acupuncture types for KOA (37). Moreover, a recent trial indicated that
strong-electricity EA was better than weak-electricity EA in alleviating knee pain (38). In this
trial, however, we managed the electric stimulation as imperceptible as possible in the EA group
for blinding purpose, which may account for the actual response rate (60.3%) less than the
expected 70%.

The effect size of EA for KOA in this trial was similar to topical NSAIDs which are recommended
prior to oral NSAIDs for KOA according to the guidelines (8). In the present trial, 60.3% of
participants in the EA group achieved significant improvement in pain and function; this was
almost the same as the use of topical diclofenac or ketoprofen for pain relief in chronic
musculoskeletal pain (60%) (39). The response rate of 60.3% with EA in this trial also
collaborated with the findings on acupuncture for joint pain from aromatase inhibitors. In women
with early stage breast cancer who developed joint pain with aromatase inhibitors, 58% reported
significantly decreased joint pain at 6 weeks (40).

Experimental evidence suggests that adenosine, cannabinoids, opioids, dopamine D2 receptors,

and muscarinic cholinergic receptor are involved in analgesia effect of acupuncture for
inflammatory pain (41). Neuroimaging study indicate that acupuncture may achieve its therapeutic

This article is protected by copyright. All rights reserved

 effects on KOA pain by preventing cortical thinning and decrease in functional connectivity in
Accepted Article
major pain related areas, and therefore modulating pain in the descending pain modulatory
pathway (42). The mechanism of the persistent effects of acupuncture is not well understood.
Laboratory study indicates that repeated non-nociceptive stimuli, such as EA, reduce nociception
through generalized habituation from the non-nociceptive stimulus-response pathway to the
nociceptive pathway where endocannabinoid-mediated neuromodulation is required (43). Apart
from the time-dependent responses, persistent activities after acupuncture were also identified in
the anterior insula and prefrontal cortices through functional magnetic resonance imaging method
(44). Nonetheless, the therapeutic mechanisms of acupuncture for KOA as an intervention in
general deserve further investigation.

The strengths of this trial included the adequate dosage of acupuncture, the use of the primary
outcome at an individual level, and the rigorous methodology. In clinical trials, results are usually
reported as means of outcomes at the group level, which is difficult to interpret for patients and
policy makers. The concept of MCII, which was defined as the smallest improvement in the
domain of interest that patients perceive as beneficial, was put forward to present the effect of
intervention at the individual level. An advisable design using MCII is based on a “responder
analysis,” namely, comparing the proportion of patients with each intervention who experience a
change greater than MCII. This type of data presentation can provide patients and policy makers
with more straightforward information to decide whether a treatment should be used.

This trial has several limitations. First, three sessions per week may be burdensome for patients in
the United States and other developed countries, even though five sessions per week is common in
China (45). Furthermore, the noninvasive acupoint stimulator, which can be manipulated by the
patients themselves at home after the training by acupuncturists, may be an alternative option for
chronic diseases, although further research and development are required. Second, the study
population represented a highly selected group of patients whose radiologic grade of KOA was II
or III. Hence, it is uncertain whether acupuncture will be beneficial for the patients with radiologic

This article is protected by copyright. All rights reserved

 grade of IV. Third, both primary outcome and secondary outcomes were subjective and vulnerable
Accepted Article
to potential biases from self-reporting, although blinding method was adopted and the James
blinding index indicated that the blinding was successful. Fourth, the significance level was not
adjusted for multiple comparisons for secondary outcomes and these analyses should be
interpreted as exploratory. Fifth, no blinding test for outcome assessors was performed; influences
from outcome assessors remain unknown. Sixth, acupuncture was suggested as a cost-effective
treatment in patients with chronic osteoarthritis pain (46) and dysmenorrhea (47), however, there
is not enough information to substantiate a cost effectiveness analysis in present trial.

CONCLUSIONS
In summary, the intensive EA, compared with SA, resulted in less pain and better function among
patients with KOA. The effects of EA persisted through week 26. The intensive MA had no
benefit for KOA at week 8, but its effect became statistically significant by week 16 and was
maintained at week 26.

This article is protected by copyright. All rights reserved

 Contributors
Accepted Article
All authors were involved in drafting the article or revising it critically for important intellectual
content, and all authors approved the final version to be published. Dr. Cun-Zhi Liu had full
access to all of the data in the study and takes responsibility for the integrity of the data and the
accuracy of the data analysis.
Jian-Feng Tu: 1a, 2, 3.
Jing-Wen Yang: 1a, 2, 3.
Guang-Xia Shi: 1a, 2, 3.
Zhang-Sheng Yu: 1c, 2, 3.
Jin-Ling Li: 1b, 2, 3.
Lu-Lu Lin: 1b, 2, 3.
Yu-Zheng Du: 1c, 2, 3.
Xiao-Gang Yu: 1c, 2, 3.
Hui Hu, MD: 1c, 2, 3.
Zhi-Shun Liu: 1c, 2, 3.
Chun-Sheng Jia: 1c, 2, 3.
Li-Qiong Wang: 1c, 2, 3.
Jing-Jie Zhao: 1c, 2, 3.
Jun Wang: 1c, 2, 3.
Tong Wang: 1c, 2, 3.
Yang Wang: 1c, 2, 3.
Tian-Qi Wang: 1b, 2, 3.
Na Zhang: 1b, 2, 3.
Xuan Zou: 1b, 2, 3.
Yu Wang: 1b, 2, 3.
Jia-Kai Shao: 1b, 2, 3.
Cun-Zhi Liu: 1a, 2, 3.
Competing interests

This article is protected by copyright. All rights reserved

 None declared.
Accepted Article
Patient consent for publication
Not required.
Ethics approval
Ethics approval was received from the institution review board at each of the 9 study hospitals
before participant enrollment. The study protocol was registered in the ClinicalTrials register
(NCT03366363).
Data availability statement
Individual anonymous participant data, protocol or statistical analysis plan will be shared for
researchers whose proposed use of the data has been approved. For inquiries about data sharing,
please send request at [email protected].

This article is protected by copyright. All rights reserved

 References
Accepted Article
1. Hunter DJ, Bierma-Zeinstra S. Osteoarthritis. Lancet 2019; 393: 1745-59.
2. Wallace IJ, Worthington S, Felson DT, Jurmain RD, Wren KT, Maijanen H, et al. Knee
osteoarthritis has doubled in prevalence since the mid-20th century. Proc Natl Acad Sci U
S A. 2017;114:9332-6.
3. Hiligsmann M, Cooper C, Arden N, Boers M, Branco JC, Luisa Brandi M, et al. Health
economics in the field of osteoarthritis: an expert's consensus paper from the European
Society for Clinical and Economic Aspects of Osteoporosis and Osteoarthritis (ESCEO).
Semin Arthritis Rheum. 2013;43:303-13.
4. da Costa BR, Reichenbach S, Keller N, Nartey L, Wandel S, Jüni P, et al. Effectiveness of
non-steroidal anti-inflammatory drugs for the treatment of pain in knee and hip
osteoarthritis: a network meta-analysis. Lancet. 2017;390:e21-21e33.
5. Gregori D, Giacovelli G, Minto C, Barbetta B, Gualtieri F, Azzolina D, et al. Association
of Pharmacological Treatments With Long-term Pain Control in Patients With Knee
Osteoarthritis: A Systematic Review and Meta-analysis. JAMA. 2018;320:2564-79.
6. Ferket BS, Feldman Z, Zhou J, Oei EH, Bierma-Zeinstra SM, Mazumdar M. Impact of
total knee replacement practice: cost effectiveness analysis of data from the Osteoarthritis
Initiative. BMJ. 2017;356:j1131.
7. Ma Y, Dong M, Zhou K, Mita C, Liu J, Wayne PM. Publication Trends in Acupuncture
Research: A 20-Year Bibliometric Analysis Based on PubMed. PLoS One.
2016;11:e0168123.
8. Osteoarthritis: Care and Management in Adults. 2014. London. National Institute for
Health and Care Excellence (UK).
9. Bannuru RR, Osani MC, Vaysbrot EE, Arden NK, Bennell K, Bierma-Zeinstra S, et al.
OARSI guidelines for the non-surgical management of knee, hip, and polyarticular
osteoarthritis. Osteoarthritis Cartilage 2019;27:1578-89.
10. Kolasinski SL, Neogi T, Hochberg MC, Oatis C, Guyatt G, Block J, et al. 2019 American
College of Rheumatology/Arthritis Foundation Guideline for the Management of

This article is protected by copyright. All rights reserved

 Osteoarthritis of the Hand, Hip, and Knee. Arthritis Rheumatol 2020;72:220-33.
Accepted Article
11. Sun N, Tu JF, Lin LL, Li YT, Yang JW, Shi GX, et al. Correlation between acupuncture
dose and effectiveness in the treatment of knee osteoarthritis: a systematic review.
Acupunct Med 2019;37:261-7.
12. Armour M, Smith CA. Treating primary dysmenorrhoea with acupuncture: a narrative
review of the relationship between acupuncture 'dose' and menstrual pain outcomes.
Acupunct Med. 2016;34:416-424.
13. Scharf HP, Mansmann U, Streitberger K, Witte S, Krämer J, Maier C, et al. Acupuncture
and knee osteoarthritis: a three-armed randomized trial. Ann Intern Med 2006;145:12-20.
14. Hinman RS, McCrory P, Pirotta M, Relf I, Forbes A, Crossley KM, et al. Acupuncture for
chronic knee pain: a randomized clinical trial. JAMA. 2014;312:1313-22.
15. Vas J, Méndez C, Perea-Milla E, Vega E, Panadero MD, León JM, et al. Acupuncture as a
complementary therapy to the pharmacological treatment of osteoarthritis of the knee:
randomised controlled trial. BMJ 2004;329:1216.
16. Lin LL, Tu JF, Wang LQ, Yang JW, Shi GX, Li JL, et al. Acupuncture of Different
Treatment Frequency in Knee Osteoarthritis: A Pilot Randomised Controlled Trial. Pain
2020; doi: 10.1097/j.pain.0000000000001940.
17. Noordergraaf A, Silage D. Electroacupuncture. IEEE Trans Biomed Eng. 1973;20:364-6.
18. Tu JF, Yang JW, Lin LL, Wang TQ, Du YZ, Liu ZS, et al. Efficacy of electro-acupuncture
and manual acupuncture versus sham acupuncture for knee osteoarthritis: study protocol
for a randomised controlled trial. Trials. 2019;20:79.
19. Tu JF, Yang JW, Wang LQ, Wang Y, Li JL, Zhang N, et al. Efficacy of
electro-acupuncture and manual acupuncture versus sham acupuncture for knee
osteoarthritis: statistical analysis plan for a randomized controlled trial. Trials.
2019;20:394.
20. Hochberg MC, Altman RD, Brandt KD, Clark BM, Dieppe PA, Griffin MR, et al.
Guidelines for the medical management of osteoarthritis. Part I. Osteoarthritis of the hip.
American College of Rheumatology. Arthritis Rheum 1995;38:1535-40.

This article is protected by copyright. All rights reserved

 21. KELLGREN JH, LAWRENCE JS. Radiological assessment of osteo-arthrosis. Ann
Accepted Article Rheum Dis. 1957;16:494-502.
22. Hawker GA, Mian S, Kendzerska T, French M. Measures of adult pain: Visual Analog
Scale for Pain (VAS Pain), Numeric Rating Scale for Pain (NRS Pain), McGill Pain
Questionnaire (MPQ), Short-Form McGill Pain Questionnaire (SF-MPQ), Chronic Pain
Grade Scale (CPGS), Short Form-36 Bodily Pain Scale (SF-36 BPS), and Measure of
Intermittent and Constant Osteoarthritis Pain (ICOAP). Arthritis Care Res (Hoboken).
2011;63:S240-52.
23. Sun N, Wang LQ, Shao JK, Zhang N, Zhou P, Fang SN, et al. An expert consensus to
standardize acupuncture treatment for knee osteoarthritis. Acupunct Med.
2020;38.doi:10.1177/0964528419900789.
24. Witt C, Brinkhaus B, Jena S, et al. Acupuncture in patients with osteoarthritis of the knee:
a randomised trial. Lancet 2005; 366: 136-43.
25. Forestier R, Desfour H, Tessier JM, Françon A, Foote AM, Genty C, et al. Spa therapy in
the treatment of knee osteoarthritis: a large randomised multicentre trial. Ann Rheum Dis.
2010;69:660-5.
26. Xie F, Li SC, Goeree R, Tarride JE, O'Reilly D, Lo NN, et al. Validation of Chinese
Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) in patients
scheduled for total knee replacement. Qual Life Res. 2008;17:595-601.
27. Tubach F, Ravaud P, Baron G, Falissard B, Logeart I, Bellamy N, et al. Evaluation of
clinically relevant changes in patient reported outcomes in knee and hip osteoarthritis: the
minimal clinically important improvement. Ann Rheum Dis. 2005;64:29-33.
28. Ware J Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of
scales and preliminary tests of reliability and validity. Med Care. 1996;34:220-33.
29. Collins SL, Edwards J, Moore RA, Smith LA, McQuay HJ. Seeking a simple measure of
analgesia for mega-trials: is a single global assessment good enough. Pain.
2001;91:189-94.
30. Devilly GJ, Borkovec TD. Psychometric properties of the credibility/expectancy

This article is protected by copyright. All rights reserved

 questionnaire. J Behav Ther Exp Psychiatry. 2000;31:73-86.
Accepted Article
31. Li P, Stuart EA, Allison DB. Multiple Imputation: A Flexible Tool for Handling Missing
Data. JAMA. 2015;314:1966-7.
32. Pathan SA, Mitra B, Straney LD, et al. Delivering safe and effective analgesia for
management of renal colic in the emergency department: a double-blind, multigroup,
randomised controlled trial. Lancet 2016;387: 1999-2007.
33. Laupacis A, Sackett DL, Roberts RS. An assessment of clinically useful measures of the
consequences of treatment. N Engl J Med 1988;318:1728-33.
34. Vickers AJ, Vertosick EA, Lewith G, MacPherson H, Foster NE, Sherman KJ, et al.
Acupuncture for Chronic Pain: Update of an Individual Patient Data Meta-Analysis. J Pain
2018; 19: 455-74.
35. MacPherson H, Vertosick EA, Foster NE, Lewith G, Linde K, Sherman KJ, et al. The
persistence of the effects of acupuncture after a course of treatment: a meta-analysis of
patients with chronic pain. Pain 2017;158:784-93.
36. Zhang Q, Yue J, Lu Y. Acupuncture treatment for chronic knee pain: study by Hinman et
al underestimates acupuncture efficacy. Acupunct Med. 2015;33:170.
37. Li S, Xie P, Liang Z, Huang W, Huang Z, Ou J, et al. Efficacy Comparison of Five
Different Acupuncture Methods on Pain, Stiffness, and Function in Osteoarthritis of the
Knee: A Network Meta-Analysis. Evid Based Complement Alternat Med.
2018;2018:1638904.
38. Lv ZT, Shen LL, Zhu B, Zhang ZQ, Ma CY, Huang GF, et al. Effects of intensity of
electroacupuncture on chronic pain in patients with knee osteoarthritis: a randomized
controlled trial. Arthritis Res Ther. 2019;21:120.
39. Derry S, Conaghan P, Da Silva JA, Wiffen PJ, Moore RA. Topical NSAIDs for chronic
musculoskeletal pain in adults. Cochrane Database Syst Rev. 2016;4:CD007400.
40. Hershman DL, Unger JM, Greenlee H, Capodice JL, Lew DL, Darke AK, et al. Effect of
Acupuncture vs Sham Acupuncture or Waitlist Control on Joint Pain Related to Aromatase
Inhibitors Among Women With Early-Stage Breast Cancer: A Randomized Clinical Trial.

This article is protected by copyright. All rights reserved

 JAMA. 2018;320:167-176.
Accepted Article
41. Zhang R, Lao L, Ren K, Berman BM. Mechanisms of acupuncture-electroacupuncture on
persistent pain. Anesthesiology. 2014;120:482-503.
42. Chen X, Spaeth RB, Retzepi K, Ott D, Kong J. Acupuncture modulates cortical thickness
and functional connectivity in knee osteoarthritis patients. Sci Rep. 2014;4:6482.
43. Hanson A, Burrell BD. Are the persistent effects of "gate control" stimulation on
nociception a form of generalization of habituation that is endocannabinoid-dependent.
Neurobiol Learn Mem. 2018;155:361-370.
44. Bai L, Tian J, Zhong C, et al. Acupuncture modulates temporal neural responses in wide
brain networks: evidence from fMRI study. Mol Pain. 2010;6:73. doi:
10.1186/1744-8069-6-73.
45. He W, Zhu B, Yu X, Liu B, Xu N, Jing X. Compairison between western and Chinese
acupuncture and its enlightenment [in Chinese]. Zhongguo Zhen Jiu. 2015;35:105-8.
46. Reinhold T, Witt CM, Jena S, Brinkhaus B, Willich SN. Quality of life and
cost-effectiveness of acupuncture treatment in patients with osteoarthritis pain. Eur J
Health Econ. 2008;9:209-19.
47. Witt CM, Reinhold T, Brinkhaus B, Roll S, Jena S, Willich SN. Acupuncture in patients
with dysmenorrhea: a randomized study on clinical effectiveness and cost-effectiveness in
usual care. Am J Obstet Gynecol. 2008;198:166.e1-8.

This article is protected by copyright. All rights reserved

 Figure legends
Accepted Article
Figure 1. Flow diagram
EA, electro-acupuncture; MA, manual acupuncture; SA, sham acupuncture
* Participants, who received randomization and had at least one post-baseline measurement, were
included in the modified intention-to-treat population.

Figure 2. Trajectory of response rates over time

Data presented as percentage of patients and error bars were 97.5% confidence interval. All p
values are versus sham acupuncture.
Red line: Electro-acupuncture group
Blue line: Manual acupuncture group
Green line: Sham acupuncture group
* represent p<0.025.

This article is protected by copyright. All rights reserved

 Table 1. Baseline Characteristics of Participants in the Modified Intention-to-Treat Analysis
Accepted Article Characteristic EA (N = 151) MA (N = 145) SA (N = 146) All (N = 442)
Age, mean ± SD years 62.7 ± 6.6 63.0 ± 7.2 62.8 ±7.6 62.8 ± 7.1
Female, no. (%) 119 (78.8) 111 (76.6) 106 (72.6) 336 (76.0)
Race, no. (%)
Han 146 (96.7) 139 (95.9) 145 (99.3) 430 (97.3)
Minorities 5 (3.3) 6 (4.1) 1 (0.7) 12 (2.7)
BMI, mean ± SD kg/m2 25.4 ± 2.9 25.1 ± 3.4 26.0 ± 3.5 25.5± 3.3
Duration, mean ± SD years 6.0 ± 5.3 6.3 ± 5.6 7.5 ± 6.1 6.6 ± 5.7
Radiological grade, no. (%)
Grade Ⅱ 91 (60.3) 83 (57.2) 91 (62.3) 265 (60.0)
Grade Ⅲ 60 (39.7) 62 (42.8) 55 (37.7) 177 (40.0)
Affected knee, no. (%)
Unilateral knee 14 (9.3) 17 (11.7) 17 (11.6) 48 (10.9)
Bilateral knees 137 (90.7) 128 (88.3) 129 (88.4) 394 (89.1)
Treatment in the past, no. (%) a
Medication 73 (48.3) 85 (58.6) 60 (41.1) 218 (49.3)
Physical therapy 42 (27.8) 42 (29.0) 42 (28.8) 126 (28.5)
Acupuncture 14 (9.3) 10 (6.9) 15 (10.2) 39 (8.8)
Injections 7 (4.6) 8 (5.5) 7 (4.8) 22 (5.0)
Concomitant diseases, no. (%)
0 78 (51.7) 70 (48.3) 78 (53.4) 226 (51.1)
1 47 (31.1) 46 (31.7) 52 (35.6) 145 (32.8)
2 22 (14.6) 23 (15.9) 14 (9.6) 59 (13.3)
≥3 4 (2.6) 6 (4.1) 2 (1.4) 12 (2.7)
History of acupuncture
Yes 62(41.1) 61(39.6) 59(40.4) 182(41.2)
No 89(58.9) 84(60.4) 87(59.6) 260(58.8)
NRS, mean (SD) 6.1 (1.3) 6.1 (1.3) 5.8 (1.3) 6.0 (1.3)
WOMAC, mean (SD)
Function subscale 21.1 (9.0) 20.7 (8.9) 20.8 (8.3) 20.9 (8.7)
Pain subscale 6.7 (2.9) 6.6 (2.8) 6.4 (2.7) 6.5 (2.8)
Stiffness subscale 2.1 (1.5) 2.1 (1.5) 2.2 (1.7) 2.2 (1.6)
SF-12, mean (SD)
Physical health 30.9 (8.0) 31.6 (8.1) 30.9 (7.8) 31.1 (7.9)
Mental health 51.5 (11.3) 51.1 (11.2) 51.3 (10.9) 51.3 (11.1)
CEQ score, mean (SD) b

This article is protected by copyright. All rights reserved

Accepted Article Credibility 0.2 (2.7) -0.1 (2.6) -0.1 (2.8) 0 (1.0)
Expectancy 0.3 (2.8) -0.2 (2.1) -0.2 (2.2) 0 (1.0)
Abbreviations: BMI, Body Mass Index (calculated as weight in kilograms divided by height in meters squared); CEQ,
credibility expectancy questionnaire; NRS, Numerical Rating Scale; SF-12, 12-item Short Form Health Survey;
WOMAC, Western Ontario and McMaster Universities osteoarthritis index.
a Defined as treatments sought for knee pain previously. More than 1 subcategory type is allowed.
b Scale has M = 0.0 (SD = 1.0) because items were converted to z-scores before averaging.

This article is protected by copyright. All rights reserved

Accepted Article
Table 2. Primary and Secondary Outcomes in the Modified Intention-to-Treat Analysis
Pairwise Comparison
Outcomes EA (n=151) MA (n=145) SA (n=146) Overall P value a
EA vs SA MA vs SA
Response rate, No. (%) Difference (97.5% CI) b P value Difference (97.5% CI) b P value
Week 4 68 (45.0) 60 (41.4) 45 (30.8) NA 14.2 (1.7 to 26.7) 0.011 10.6 (-2.0 to 23.1) 0.059
Week 8 c 91 (60.3) 85 (58.6) 69 (47.3) NA 13.0 (0.2 to 25.9) 0.023 11.3 (-1.6 to 24.4) 0.051
Week 16 91 (60.3) 82 (56.6) 53 (36.3) NA 24.0 (11.4 to 36.6) ＜0.001 20.3 (7.4 to 33.1) ＜0.001
Week 26 86 (57.0) 76 (52.4) 50 (34.2) NA 22.7 (10.1 to 35.3) ＜0.001 18.2 (5.4 to 31.0) 0.002
Numerical Rating Scale, mean (SD) Difference (95% CI) Overall P value d Difference (95% CI) Overall P value d
Baseline 6.11 (1.32) 6.06 (1.33) 5.83 (1.30) 0.28 (-0.02 to 0.58) 0.23 (-0.07 to 0.54)
Week 4 3.76 (1.64) 3.91 (1.58) 4.27 (1.74) -0.51 (-0.90 to -0.13) -0.36 (-0.75 to -0.02)
Week 8 2.59 (1.61) 2.87 (1.62) 3.38 (1.91) <0.001 -0.79 (-1.20 to -0.37) <0.001 -0.51 (-0.93 to -0.09) <0.001
Week 16 2.78 (1.64) 3.07 (1.72) 3.73 (1.97) -0.94 (-1.38 to -0.51) -0.65 (-1.10 to -0.21)
Week 26 2.94 (1.67) 3.14 (1.64) 3.72 (1.89) -0.78 (-1.21 to -0.35) -0.58 (-1.01 to -0.15)
WOMAC function subscale, mean (SD) Difference (95% CI) Overall P value d Difference (95% CI) Overall P value d
Baseline 21.09 (9.00) 20.70 (8.92) 20.77 (8.27) 0.33 (-1.65 to 2.30) -0.06 (-2.05 to 1.92)
Week 4 12.44 (7.24) 13.99 (7.76) 14.45 (8.34) -2.00 (-3.79 to -0.22) -0.45 (-2.31 to 1.41)
Week 8 9.26 (7.03) 10.82 (7.32) 11.78 (8.17) <0.001 -2.51 (-4.30 to -0.72) <0.001 -0.95 (-2.79 to 0.88) 0.060
Week 16 9.39 (7.01) 10.86 (7.74) 12.77 (8.26) -3.38 (-5.20 to -1.57) -1.92 (-3.82 to -0.01)
Week 26 9.87 (7.04) 11.29 (7.70) 13.21 (9.09) -3.33 (-5.29 to -1.38) -1.92 (-3.96 to 0.12)
WOMAC pain subscale, mean (SD) Difference (95% CI) Overall P value d Difference (95% CI) Overall P value d
Baseline 6.68 (2.93) 6.55 (2.75) 6.40 (2.73) 0.28 (-0.37 to 0.92) 0.15 (-0.48 to 0.79)
<0.001 <0.001 0.001
Week 4 3.66 (2.47) 4.00 (2.06) 4.29 (2.54) -0.63 (-1.21 to -0.06) -0.29 (-0.83 to 0.24)

This article is protected by copyright. All rights reserved

Accepted Article
Week 8 2.50 (1.95) 2.79 (1.91) 3.57 (2.59) -1.07 (-1.60 to -0.53) -0.78 (-1.32 to -0.24)
Week 16 2.49 (1.98) 2.98 (2.21) 3.64 (2.48) -1.14 (-1.68 to -0.61) -0.66 (-1.22 to -0.10)
Week 26 2.79 (2.41) 3.16 (2.03) 3.94 (2.70) -1.15 (-1.76 to -0.53) -0.78 (-1.36 to -0.21)
WOMAC stiffness subscale, mean (SD) Difference (95% CI) Overall P value d Difference (95% CI) Overall P value d
Baseline 2.13 (1.49) 2.09 (1.54) 2.23 (1.65) -0.09 (-0.45 to 0.27) -0.14 (-0.50 to 0.23)
Week 4 1.54 (1.14) 1.50 (1.25) 1.62 (1.33) -0.08 (-0.36 to 0.20) -0.12 (-0.42 to 0.18)
Week 8 1.05 (1.04) 1.22 (1.14) 1.36 (1.20) 0.068 -0.31 (-0.58 to -0.05) 0.021 -0.13 (-0.41 to 0.14) 0.174
Week 16 1.15 (1.03) 1.30 (1.15) 1.39 (1.16) -0.23 (-0.49 to 0.03) -0.08 (-0.36 to 0.19)
Week 26 1.19 (1.06) 1.30 (1.20) 1.59 (1.34) -0.40 (-0.69 to -0.11) -0.29 (-0.60 to 0.02)
SF-12 Physical health, mean (SD) Difference (95% CI) P value e Difference (95% CI) P value e
Baseline 30.89 (8.01) 31.60 (8.08) 30.94 (7.77) -0.06 (-1.86 to 1.75) 0.95 0.66 (-1.17 to 2.49) 0.48
Week 4 35.92 (8.22) 36.28 (7.95) 34.93 (8.82) 1.90 (-0.03 to 3.83) 0.054 0.99 (-0.61 to 3.31) 0.32
Week 8 38.97 (8.33) 39.22 (8.26) 38.42 (9.13) NA -0.17 (-1.95 to 2.30) 0.87 0.55 (-1.52 to 2.61) 0.60
Week 16 39.92 (8.20) 39.98 (7.98) 37.62 (8.74) 2.80 (0.71 to 4.89) 0.009 2.29 (0.27 to 4.31) 0.026
Week 26 39.18 (8.79) 39.20 (8.61) 38.05 (8.11) 1.59 (-0.42 to 3.59) 0.120 1.13 (-0.92 to 3.17) 0.28
SF-12 Mental health, mean (SD) Difference (95% CI) P value e Difference (95% CI) P value e
Baseline 51.47 (11.30) 51.11 (11.21) 51.35 (10.93) 0.12 (-2.43 to 2.66) 0.93 -0.24 (-2.80 to 2.32) 0.85
Week 4 52.64 (9.93) 52.68 (9.65) 52.86 (9.09) -0.22 (-2.39 to 1.96) 0.85 -0.35 (-2.34 to 1.99) 0.87
Week 8 53.73 (9.23) 54.67 (8.51) 53.21 (8.95) NA 0.53 (-1.59 to 2.64) 0.63 1.46 (-0.60 to 3.53) 0.164
Week 16 54.33 (8.42) 54.61 (7.93) 52.25 (9.23) 2.08 (-0.01 to 4.17) 0.051 2.36 (0.31 to 4.41) 0.024
Week 26 54.11 (8.04) 54.92 (8.22) 51.34 (9.37) 2.77 (0.67 to 4.90) 0.010 3.57 (1.44 to 5.71) 0.001
Abbreviations: ITT set, intention-to-treat set; EA, electro-acupuncture; MA, manual acupuncture; SA, sham acupuncture; SF-12, 12-item Short Form Health Survey; WOMAC,
Western Ontario and McMaster Universities osteoarthritis index.

This article is protected by copyright. All rights reserved

Accepted Article
a NRS and WOMAC subscales were not significantly influenced by the interaction of treatment and time, thus the data during 26 weeks were aggregated in the mixed-effect model
with repeated measurement and no between-group comparisons were performed at each time-point.
b The significance level for primary outcome was adjusted at 0.025 for multiple comparisons using the Bonferroni method. Thus, the 97.5% CI were used for primary outcome.
c The response rate at weeks 8 was the primary outcome; the response rates at weeks 4, 16, and 26, and other questionnaires were the secondary outcomes.
d The overall P value is for the dimensionality of all time points, not for the dimensionality of comparison among three groups.
e Both SF-12 physical health and mental health were significantly influenced by the interaction of treatment and time. Thus, the P value at each time-point was presented instead of
overall P value.

This article is protected by copyright. All rights reserved

 Table 3. Adverse Events Related to Treatment a
Accepted Article
Adverse Event EA (n = 156) b MA (n = 155) b SA (n = 157) b
Overall, no. (%) 18 (11.5) 22 (14.2) 17 (10.8)
Severe adverse events 0 0 0
Subcutaneous hematoma 7 (4.5) 10 (6.5) 9 (5.7)
Post-needling pain 10 (6.4) 13 (8.4) 10 (6.4)
Pantalgia 1 (0.6) 0 0
Abbreviations: EA, electro-acupuncture; MA, manual acupuncture; SA, sham acupuncture.
a Adverse events were counted by type rather than frequency in the same participant. Adverse events with different
types occurring in a single participant were defined as independent adverse events. An adverse event with multiple
occurrences in a single participant was defined as 1 adverse event.
b Four participants in the electro-acupuncture group, 5 in the manual acupuncture group and 3 in the sham
acupuncture group did not receive treatment.
c Pantalgia means pain in the whole body (joints and muscles).

This article is protected by copyright. All rights reserved

 Table 4. Participant-Blinding Assessment
Accepted Article
Treatment guess, n(%) EA (n=151) MA (n=145) SA (n=146) James Blinding Index (95%CI) a
Weeks 4
‘Electro-acupuncture’ 11 (7.3) 30 (20.7) 29 (19.9)
‘Manual acupuncture’ 103 (68.2) 59 (40.7) 49 (33.6)
0.64 (0.50 to 0.78)
‘Sham acupuncture’ 8 (5.3) 16 (11.0) 22 (15.1)
‘Unsure’ 29 (19.2) 40 (27.6) 46 (31.5)
Weeks 8 b
‘Electro-acupuncture’ 11 (7.7) 31 (22.3) 33 (23.9)
‘Manual acupuncture’ 98 (68.5) 59 (42.4) 50 (36.2)
0.63 (0.46 to 0.79)
‘Sham acupuncture’ 8 (5.6) 14 (10.1) 21 (15.2)
‘Unsure’ 26 (18.2) 35 (25.2) 34 (24.6)
Abbreviations: EA, electro-acupuncture; MA, manual acupuncture; SA, sham acupuncture.
a James blinding index range, 0-1; 0 indicates total absence of blinding, 1 indicates complete blinding, and 0.5
indicates completely random blinding.
b Eight participants in the electro-acupuncture group, 6 in the manual acupuncture group, and 8 in the sham
acupuncture group did not complete the blinding assessment at weeks 8.

This article is protected by copyright. All rights reserved

Accepted Article art_41584_f1.eps

This article is protected by copyright. All rights reserved

Accepted Article 100
Response rate (%)

80
* ** **
60 *
40

20

0
0 4 8 16 26
Weeks
art_41584_f2.eps

This article is protected by copyright. All rights reserved