Article Text
PDF
Abstract
Objectives This study aims to evaluate (1) the effect and safety of acupuncture in patients with knee osteoarthritis (KOA) and explore (2) whether the effect of acupuncture differed according to acupuncture type, acupuncture dose and follow-up time.
Design Systematic review and pairwise and exploratory network meta-analysis.
Setting PubMed, Embase, Cochrane Central Register of Controlled Trials, Web of Science, China National Knowledge Infrastructure, Chinese Biomedical Literature Database, VIP Database for Chinese Technical Periodicals and Wanfang from inception to 13 November 2023.
Participants Randomised controlled trials comparing acupuncture with sham acupuncture, non-steroidal anti-inflammatory drugs (NSAIDs), usual care or waiting list groups, intra-articular (IA) injection and blank groups in patients with KOA.
Interventions Eligible interventions included manual acupuncture (MA) and electroacupuncture (EA).
Main outcomes measures The primary outcome was pain intensity at the end of treatment.
Results 80 trials (9933 participants) were included. Very low certainty evidence suggested that acupuncture may reduce pain intensity compared with sham acupuncture (standardised mean difference, SMD −0.74, 95% CI −1.08 to −0.39, corresponded to a difference in Visual Analogue Scale of −18.50 mm, −27.00 to −9.75), NSAIDs (SMD −0.86 –1.26 to −0.46, corresponded to −21.50 mm, −31.50 to −11.50), usual care or waiting list groups (SMD −1.01, –1.47 to −0.54, corresponded to −25.25 mm, −36.75 to −13.50) and blank groups (SMD −1.65, –1.99 to −1.32, corresponded to −41.25 mm, −49.75 to −33.00), but not IA injection. Similar results were also found in other outcomes. For most of the subgroup analyses, acupuncture type, acupuncture dose and follow-up time did not show a significant relative effect. Only when compared with NSAIDs, a higher dose of acupuncture may provide greater pain relief (interaction p<0.001). The network meta-analysis revealed that electroacupuncture (SMD −0.75, 95% CI −1.34 to −0.17) had a greater effect on pain relief in patients with KOA compared with manual acupuncture.
Conclusions The findings suggest that acupuncture may provide clinically important effects in reducing pain and improving physical function in patients with KOA, but the certainty of evidence was very low. Electroacupuncture and higher dose of acupuncture probably are two potential contributing factors.
PROSPERO registration number CRD42021232177.
- Acupuncture
- Knee
- PAIN MANAGEMENT
Data availability statement
All data relevant to the study are included in the article or uploaded as online supplemental information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Acupuncture has long been recognised as a potentially effective non-pharmacological therapy for the treatment of knee osteoarthritis (KOA).
Different guidelines for treatment of KOA provide conflicting recommendations on use of acupuncture.
The optimal type and dose of acupuncture for KOA remain unknown.
WHAT THIS STUDY ADDS
Acupuncture may provide clinically important effects in reducing pain and improving physical function in patients with KOA, but the certainty of evidence was very low.
Electroacupuncture and higher dose of acupuncture probably are two potential contributing factors, and research is needed to better uncover dose–effect relationship for acupuncture.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Based on the evidence, the guideline recommendations on the treatment of KOA with acupuncture should be reconsidered.
Electroacupuncture and higher dose of acupuncture may provide greater benefits for patients with KOA.
Introduction
Osteoarthritis (OA) is emerging as a common degenerative disease and one of the leading causes of disability worldwide.1 Knee OA (KOA) is the most prevalent OA location and contributes the approximately 85% of the overall burden, affecting 3.8% of the global population.2–4 Pain symptoms associated with KOA result in reduced physical function and quality of life (QOL), which in turn increase the risk of all-cause mortality.5 6
Non-steroidal anti-inflammatory drugs (NSAIDs) are the most commonly prescribed pharmacological agents for treating KOA.7–9 However, side effects associated with NSAIDs, such as cardiovascular and gastrointestinal bleeding, limit their use in clinical practice.10 11 Besides, the healthcare systems are overstretched because of the global opioid crisis aggravated by opioid prescriptions for arthritis pain12 13 and the increasing joint replacement requirements.14 Two effective non-pharmacological therapies, exercise and weight loss, are strongly recommended for KOA patients by updating guidelines.7–9 Also, a real-world cohort study suggested early initiation of physical treatment care could reduce utilisation of opioids in patients with KOA.15 However, adherence to long-term exercise or weight loss is challenging for some patients with KOA. In this context, identifying safe and effective treatments is urgently needed to provide immediate benefits to patients.
Acupuncture has long been recognised as a potentially effective non-pharmacological therapy for the treatment of KOA. Following several randomised controlled trials (RCTs) suggesting that acupuncture showed benefits for KOA,16 17 some trials with contrary evidence have been published.18 19 Moreover, current guidelines differ substantially in whether acupuncture should be recommended for patients with KOA,7 9 20 21 posing a challenge to clinical decision-making. Although some systematic reviews were conducted to evaluate the effect and safety of acupuncture to treat KOA,22–26 few drew a definitive conclusion.25 26 These systematic reviews did not include several published trials,27 28 considered only one type of acupuncture or comparators22 23 or failed to assess the certainty of evidence.24 Moreover, the inconsistent findings may be associated with various acupuncture regimens and parameters including acupuncture type, acupuncture dose, etc.29 A previous RCT found that intensive electroacupuncture (EA) but not intensive manual acupuncture (MA) showed benefits for KOA compared with sham acupuncture at week 8.30 Also, recent studies showed that there was a positive correlation between higher-dose acupuncture and better treatment outcomes.31 32 However, there is still a lack of comprehensive studies exploring whether these factors contribute to any potential acupuncture effect. To identify potentially important parameters, we systematically evaluated (1) the effect and safety of acupuncture in patients with KOA and explored (2) whether the effect of acupuncture differed according to acupuncture type, acupuncture dose and follow-up time.
Methods
This is a systematic review and meta-analysis (PROSPERO registration CRD42021232177) focused on RCTs involving acupuncture for KOA patients. We followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines to organise this article.33 The study protocol was published in a peer-reviewed journal.34
Eligibility criteria
We included RCTs that compared acupuncture with sham acupuncture, NSAIDs, usual care or waiting list groups, intra-articular (IA) injection and blank control in patients with KOA. Eligible interventions included MA and EA regardless of the sessions, needling manipulation (eg, lifting, thrusting and twirling needles to acupoints), types of filiform needles, stimulation methods and stimulation areas. We included RCTs published in peer-reviewed journals as well as unpublished data as a supplement. No restrictions were placed on the publication date, but we limited the publication language to English and Chinese. Besides, we only included RCTs that were judged as ‘low risk’ of bias for random sequence generation.
We excluded studies that included patients with knee pain but were not diagnosed with KOA and patients who received previous total knee arthroplasty. RCTs comparing acupuncture with other traditional Chinese medicine therapies were excluded. RCTs that reported only improvement rates were also excluded.
Usual care or waiting list groups were classified as one of the following: (1) patients were asked to continue their existing general care; (2) patients were not given any specific intervention and (3) waiting list groups did not receive any specific therapy during the trial period; they received an active intervention after the trial period. The blank groups were defined as acupuncture plus one or more therapies versus the same therapies alone.
Data sources and searches
We searched four English databases (PubMed, Embase, Cochrane Central Register of Controlled Trials and Web of Science) and four Chinese databases (China National Knowledge Infrastructure, Chinese Biomedical Literature Database, VIP Database for Chinese Technical Periodicals and Wanfang) from database inception to 1 September 2021 and updated the searches on 13 November 2023 (see online supplemental file 1). We also manually searched the WHO International Clinical Trials Registry Platform, ClinicalTrials.gov and the Chinese Clinical Registry.
Supplemental material
Selection of studies and data extraction
Four reviewers (CL, YD, HZ and YW) independently screened all records by titles and abstracts, and read full texts of potentially relevant studies to establish eligibility (see online supplemental file 2 for a list of excluded studies with reasons). Any disagreements were settled by discussion among reviewers or consultation with another reviewer (LW). The characteristics of study populations, interventions and outcome were extracted using a piloted data extraction form. For quantitative outcome, the post-treatment means, SDs and the number of patients in each group at each time point were extracted. We classified follow-up times into 26 weeks and 52 weeks after randomisation. When CIs were reported, we used the means, CIs and the number of patients in each group to obtain SDs. When SEs were reported, we used the SEs and the number of patients in each group to estimate SDs. If SDfinal was not reported, we calculated the missing SDs according to the following formula recommended by Cochrane handbook: SD change = square root [SD2 baseline+SD2 final−(2×Corr×SDbaseline×SDfinal).35
Outcomes
The primary outcome was pain. Pain intensity was selected as the targeted outcome because it plays an important role in the pain management of KOA. When a study reported more than one pain outcome measure, the outcomes were prioritised according to the following hierarchical list:36 (1) global knee pain evaluated by Numerical Rating Scales or Visual Analogue Scale (VAS); (2) Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain subscale; (3) other pain scales such as McGill pain questionnaire and Lequesne pain subscale; (4) WOMAC global score and (5) Lequesne OA index score. We extracted results from each time point after the last acupuncture.
Secondary outcomes were physical function and QOL. When a study reported more than one outcome measure, the following hierarchical list was used to extract data.36 For physical function: (1) WOMAC physical function subscale; (2) WOMAC global score; (3) other scales for example Lequesne OA index score. For QOL: (1) 36-item Short Form Health Survey (SF-36) physical, mental and total and (2) 12-item Short Form Health Survey (SF-12) physical, mental and total. We extracted data on these outcomes for the same time points as those described for pain.
Our safety outcomes were adverse events (participants experiencing any adverse event or serious adverse events), irrespective of whether the adverse events were associated with acupuncture intervention.
Risk of bias and certainty of evidence
Four reviewers (CL, YD, HZ and YW) judged the risk of bias in trials using the Cochrane Collaboration risk of bias tool for seven domains: random sequence generation, allocation concealment, blinding of patients and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other bias (see online supplemental file 3).37 We assessed the certainty of evidence using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE).38 The factors considered for downgrading were as follows: risk of bias, inconsistency, indirectness, imprecision and publication bias (see online supplemental file 4 for a description of the GRADE framework used). The certainty of evidence was classified four levels: high, moderate, low or very low. The summary of findings tables were prepared as a summary of the certainty of evidence using GRADEpro V.3.6.
Assessment of acupuncture dose
We assessed the acupuncture dose by using a semiquantitative scoring instrument our group proposed before.32 The scoring instrument includes the following four parameters: (1) number of acupoints; (2) De Qi response; (3) frequency of treatment per week and (4) duration of treatment. Based on the sum of scores, we determined three doses of the overall acupuncture treatment: high dose, moderate dose and low dose (see online supplemental file 5).
Data synthesis and analysis
When three-arm studies were included, the sample size of the common group was divided by 2 to be able to make 2 comparisons while avoiding statistical dependence. For dichotomous outcomes, we divided the number of events and the sample size by the number of treatment groups.35
For pain, physical function and QOL outcomes, the treatment effects were presented as the standardised mean difference (SMD) or mean difference (MD) with 95% CI, and relative risk (RR) with 95% CI was used for adverse events. A random effects model was used across all comparisons because of the clinical heterogeneity. Statistical heterogeneity in each meta-analysis was identified by the χ2 test and reported as I2. Prespecified subgroup analyses included the type of acupuncture and acupuncture dose, and the follow-up time was decided post hoc. We used funnel plots to assess the publication bias when there were 10 or more trials within a comparison (online supplemental file 6). For SMDs, Egger’s tests were used to estimate small study effects. In order to facilitate the interpretation of our results, we defined a between-group minimum clinically important difference (MCID) of 0.37 SD units based on the median MCID from recent studies in patients with OA.36 39 40 We used the median SD of 25 mm which was previous studies reported to transform SMD to VAS (in mm). Therefore, this effect approximately corresponds to −9 mm on a 100 mm VAS. Of note, this threshold of 0.37 SD units only applies to the comparison of acupuncture versus sham acupuncture.
Analyses were conducted by using Review Manager (Revman V.5.3) and Stata V.12.0 and V.16.0. A two-sided p<0.05 was considered statistical significance for all tests.
Exploratory network meta-analysis and sensitivity analysis
We performed a network meta-analysis (NMA) to compare the efficacy of EA and MA. Global inconsistency was evaluated, and the local inconsistency assessment was performed using the node-splitting method to check whether the estimated effects from the direct comparisons were consistent with those from the indirect comparisons. A p>0.05 indicates that there were no significant differences in estimated effects between direct and indirect comparisons, thus the consistency model was used; otherwise, the inconsistency model was used. To rank the efficacy of the interventions, we calculated the probabilities of the surface under the cumulative ranking curve (SUCRA) between all interventions for the primary outcome.
We also conducted sensitivity analysis by removing one study at a time and changing different models and types of effect sizes to confirm the robustness of the results.
Patient and public involvement
No patients or public were involved in the design, implementation and reporting of the study. Only data already existent in the literature and the aforementioned sources were used for this study.
Results
Characteristics of included trials
A total of 15 664 references were retrieved, from which 7293 duplicate records were removed and 7750 were excluded based on titles and abstracts. Ultimately, of 593 potentially eligible articles, we included 84 publications describing 80 trials (figure 1). These trials involved 9933 patients with a median sample size of 90 (IQR 61–120). The mean age of patients was 61.5 years (75 trials reported, n=9342), and 63.3% (77 trials reported, n=9438) were female. Of the 80 trials, 2 acupuncture types were evaluated, including MA (51 trials, n=6729) and EA (36 trials, n=4075). 69 trials (n=8754) reported pain-related outcomes and 59 trials (n=7674) reported function-related outcomes. The details of the included trials are shown in online supplemental file 7.
Study flow diagram. NSAIDs, non-steroidal anti-inflammatory drugs.
Online supplemental file 8 shows the risk of bias in eligible trials: 10% of trials had a low risk of bias for all domains and 85% were categorised as high risk in at least one domain. The random sequence generation was adequate in all trials because of eligibility criteria (n=80, 100%). Most trials were at unclear risk of allocation concealment (n=54, 68%), blinding of outcome assessment (n=53, 66%) and incomplete outcome data (n=37, 46%). 59 (74%) trials were classified as high risk of performance bias due to inadequate blinding of patients. 44 (55%) trials had a low-risk bias for selective reporting and 76 (95%) for other bias.
Acupuncture dose
Acupuncture regimens and parameters differed in acupoints, De Qi response, frequency and duration of treatment (see online supplemental file 9). The number of acupoints ranged from 2 to 14, whereas most of the trials had the number of acupoints of 4–9 (85% of 66 trials). 65 trials clearly required De Qi response; however, 1 trial did not seek for De Qi response and 13 trials failed to report relevant details. Frequency ranged from once weekly to daily, whereas most trials reported that frequencies were 3, 4 or 5 times weekly (69% of 75 trials). The duration ranged from 5 days to 26 weeks; however, most trials had durations of 2, 4, 5 or 8 weeks (71% of 79 trials). Ultimately, the acupuncture total doses of included trials were classified as high dose (32% of 79 trials), moderate dose (49% of 79 trials) and low dose (16% of 79 trials). Only two trials had two groups of moderate and low doses.
Pain
Very low certainty evidence showed that acupuncture provided clinically relevant improvements compared with sham acupuncture at the end of treatment (SMD −0.74, 95% CI −1.08 to −0.39, n=2488, 14 RCTs). This effect corresponds to a pain intensity reduction of −18.50 mm (95% CI −27.00 to −9.75) on a 100 mm VAS. Very low certainty evidence showed that the analgesic effect of acupuncture was superior to NSAIDs (SMD −0.86, 95% CI −1.26 to −0.46, n=1878, 24 RCTs), usual care or waiting list groups (SMD −1.01, 95% CI −1.47 to −0.54, n=1313, 14 RCTs) and blank groups (SMD −1.65, 95% CI −1.99 to −1.32, n=2679, 28 RCTs). Very low certainty evidence showed equivalent effects of acupuncture in pain reduction compared with IA injection (SMD 0.29, 95% CI −0.73 to 1.32, n=319, 5 RCTs) (figure 2, table 1 and online supplemental file 10, 11).
Forest plot for pain reduction at the end of treatment. (A) Acupuncture versus sham acupuncture; (B) Acupuncture versus NSAIDs; (C) Acupuncture versus usual care/waiting list; (D) Acupuncture versus blank. NSAIDs, non-steroidal anti-inflammatory drugs.
Summary of findings
Physical function
For physical function, very low certainty evidence showed that acupuncture provided clinically relevant improvements compared with sham acupuncture (SMD −0.77, 95% CI −1.21 to −0.34, n=2487, 14 RCTs). Very low certainty showed that acupuncture was associated with functional improvements compared with NSAIDs (SMD −1.06, 95% CI −1.52 to −0.60, n=1622, 20 RCTs), usual care or waiting list groups (SMD −0.61, 95% CI −1.01 to −0.21, n=1109, 11 RCTs) and blank groups (SMD −1.74, 95% CI −2.19 to −1.30, n=2271, 22 RCTs) at the end of treatment. Very low certainty evidence showed that there was no statistical difference between acupuncture and IA injection in physical function (SMD 0.92, 95% CI −1.02 to 2.86, n=163, 3 RCTs) (table 1 and online supplemental file 10, 11).
Quality of life
For physical health, evidence ranging from moderate to very low certainty showed acupuncture was more effective than NSAIDs (MD 2.02, 95% CI 0.07 to 3.96, n=208, 1 RCT), usual care or waiting list groups (MD 4.85, 95% CI 2.95 to 6.75, n=462, 2 RCTs) and blank groups (MD 8.05, 95% CI 3.70 to 12.40, n=185, 3 RCTs) but not sham acupuncture. On the other hand, evidence ranging from high to very low certainty showed that acupuncture was associated with an improvement in mental health compared with usual care or waiting list groups (MD 2.90, 95% CI 0.51 to 5.29, n=212, 1 RCT) and NSAIDs (MD 4.44, 95% CI 1.17 to 7.70, n=208, 1 RCT), but not sham acupuncture and blank groups. For global health, evidence ranging from high to low certainty suggested acupuncture showed benefits when compared with blank groups (SMD 0.70, 95% CI 0.34 to 1.06, n=127, 2 RCTs), but not sham acupuncture, NSAIDs and usual care or waiting list groups (table 1 and online supplemental file 10, 11).
Safety
37 trials evaluated the safety of acupuncture in patients with KOA. Of these, three trials reported serious adverse events and nine trials reported no adverse responses to treatment occurred in any of the groups (online supplemental file 12). For any adverse events, evidence ranging from moderate to low certainty showed no significant difference between the acupuncture and control groups in all comparisons. In comparison of acupuncture and usual care and a waiting list, no studies reported the adverse events in control groups so we did not merge the insufficient data (12.34% in acupuncture group, n=397, 8 RCTs). Besides, evidence ranging from high to low certainty indicated that acupuncture was associated with a significant risk of serious adverse events when compared with sham acupuncture (6.59% vs 3.07%, RR 2.13, 95% CI 1.16 to 3.92, n=972, 2 RCTs), but not usual care or waiting list groups and blank groups (table 1 and online supplemental file 10, 11).
Subgroup analyses
For most of the subgroup analyses, acupuncture type, acupuncture dose and follow-up time did not show a significant relative effect (table 2 and online supplemental file 13). Only when compared with NSAIDs, a high-dose acupuncture (SMD −2.30, 95% CI −3.68 to −0.93, n=435, 4 RCTs) and moderate dose acupuncture (SMD −0.62, 95% CI −1.02 to −0.22, n=1232, 17 RCTs) had a greater effect on pain relief than low dose acupuncture (SMD 0.32, 95% CI −0.22 to 0.86, n=211, 3 RCTs) (interaction p<0.001).
Subgroup analyses for pain
Exploratory NMA and sensitivity analysis
A total of 19 RCTs explored the efficacy of EA, MA or sham acupuncture on pain relief in patients with KOA, and the network plots are presented in figure 3. Both tests showed that there was no significant inconsistency between the direct comparisons and indirect comparisons (p>0.05); thus, the consistency model was used. The NMA revealed that EA (SMD −0.75, 95% CI −1.34 to −0.17) had a greater effect on pain relief in patients with KOA compared with MA. The pooled results also found EA (SMD −1.22, 95% CI −1.80 to −0.63) was associated with improvements in pain reduction but not MA (SMD −0.46, 95% CI −0.98 to 0.06) compared with sham acupuncture. The SUCRA analysis and league table showed that EA had the best rank, MA ranked second and sham acupuncture ranked last (online supplemental file 14). Sensitivity analysis testified the robustness of our results (online supplemental file 15).
(A) Structure of network formed by interventions for pain relief at the end of treatment. The direct comparison between electroacupuncture and manual acupuncture using a network diagram, in which the size of the nodes represents the sample size of each intervention, and the thickness of the continuous lines connecting the nodes indicates the number of studies directly comparing the two interventions. (B) Forest plots depicting estimates from direct and indirect comparison for electroacupuncture to manual acupuncture based on network meta-analysis. Electric, electroacupuncture; Manual, manual acupuncture; Sham, sham acupuncture.
Discussion
Principal findings
To our knowledge, this is an up-to-date and comprehensive systematic review and meta-analysis evaluating the effect and safety of acupuncture in patients with KOA considering the type of comparisons, type of acupuncture, acupuncture dose and follow-up time. Our findings showed that acupuncture provided clinically important effects in reducing pain and improving physical function but not QOL compared with sham acupuncture, though the upper limit of the CI for function was less than the MCID. We also found that acupuncture was superior to NSAIDs and was indeed effective (better than usual care or waiting list and blank) for pain, related dysfunction and impaired QOL in some domains. Besides, no difference was found between acupuncture and IA injection in reducing pain and improving function. The overall risk of bias in the included studies was high risk (85%) and the evidence is very uncertain, though we only included trials that implemented randomisation properly. In our review, acupuncture slightly increased the risk of serious adverse events which are regarded as unrelated to the intervention only when compared with sham acupuncture and, therefore, was considered as generally safe. However, the number of studies evaluating the safety was small and the certainty of evidence ranged from high to low.
Although there was no statistical difference between subgroups of EA and MA in different comparisons, the point estimate on pain relief and functional improvement of the EA is better than MA in most of the subgroup analyses. Interestingly, we conducted an exploratory NMA and found that EA resulted in a statistically significant reduction in pain intensity compared with MA. For acupuncture dose, we also found the higher dose had a better point estimate on pain relief and functional improvement, though there was still no significant difference between subgroups in most comparisons. For follow-up time, current evidence did not detect the benefit of acupuncture for KOA at weeks 26 and 52 compared with sham acupuncture.
Strengths and limitations
Our review has several strengths. First, we used the MCID of 0.37 SD units which have been generally accepted in previous studies on OA39 41 42 to better interpret our results. Then we transformed the minimal clinically important SMD to VAS (−9 mm in 100 mm).39 40 Our results found the effect of acupuncture corresponds to a difference in VAS of −18.50 mm (95% CI −27.00 to −9.75 mm) compared with sham acupuncture. Second, we assessed the acupuncture dose using a scoring instrument our group proposed before.32 To our knowledge, this is the up-to-date meta-analysis to explore the impact of acupuncture dose on treatment effect in patients with KOA through subgroup analyses. Third, we pooled direct and indirect evidence in mixed treatment comparisons to assess the comparative effect of MA and EA vs sham acupuncture in pain reduction. The exploratory NMA also confirmed our assumption from subgroup analyses.
Our review also has limitations. First, although we applied more rigorous eligibility criteria (ie, ‘low risk’ of bias for random sequence generation) to ensure the quality of RCTs included, high risk of bias and low certainty of evidence remained evident. Notably, trials were classified as high risk of performance bias for lack of blinding among open-label studies (74%). Second, the included trials showed substantial heterogeneity, which reduced evidence grade to low or very low. We conducted subgroup analyses to explore the possible sources of heterogeneity, but heterogeneity still persisted (table 2). Other acupuncture parameters (eg, needle retaining time, depth and angle of insertion), patients’ baseline characteristics, different scales and uncertainty risk of bias are also likely to contribute to high heterogeneity. Although we observed the impact of each study on results through sensitivity analysis, caution is still needed in interpreting our findings. Third, the scoring instrument of acupuncture dose is an unweighted composite of a number of acupoints, De Qi response, frequency and duration of treatment. We explored the impact of acupuncture total dose on treatment effect using the scoring instrument but did not investigate single dose component and possible interaction. Besides, the scoring instrument was based on standard clinical practice in China, so there may be limitations in evaluating those trials conducted in other countries using this method. Fourth, 45 potentially eligible trials were excluded because of insufficient data available and the authors did not reply to our request for more details (online supplemental file 2). Therefore, we could not entirely rule out selective reporting and publication bias. Fifth, we limited the publication language to English and Chinese and excluded other languages (eg, one trial was excluded due to writing in German), which suggested the potential risk of selection bias introduced by missing some eligible studies.
Comparisons with previous studies
In the past 5 years, there have been two pairwise meta-analyses and two NMA investigating the effect of acupuncture on KOA.22 23 43 44 A recent pairwise meta-analysis included 77 RCTs with 9422 patients, which found acupuncture had a beneficial effect on KOA compared with no treatment and usual care.44 However, in this review, no treatment group was defined as sham acupuncture, health education, and function rehabilitation and usual care group included physiotherapy, NSAIDs and glucosamine. These combined groups did not provide enough granular evidence to guide clinicians on whether should choose acupuncture when compared with other therapies. Our study supported that the effects of acupuncture differ depending on different controls providing enough comprehensive evidence for clinical practice. Moreover, our findings suggested that acupuncture type is one of the contributing factors of acupuncture’s treatment effects through exploratory NMA; however, Mei et al 44 found no difference between MA and EA through p value for interaction in subgroup analyses. Our findings aligned with the systematic review by Kwak et al,22 which revealed the benefits of acupuncture combined with oral medication compared with oral medication alone.
One NMA investigated the comparative effect of different acupuncture methods in the treatment of KOA to confirm that EA is better than MA.43 Another NMA suggested acupuncture in sham-controlled trials with or without sham devices has different effect sizes for KOA, which was not considered in this review. We compared EA with MA directly within a trial or indirectly through a common comparator (sham acupuncture) between trials. The pooled results found EA showed superior analgesic compared with MA and sham acupuncture, but no differences between MA and sham acupuncture was observed.
Implications for clinical practice and research
Different guidelines for treatment of KOA provide conflicting recommendations on use of acupuncture.7 9 20 21 Also, few studies focused on optimising the acupuncture regimens and parameters selection with respect to acupuncture type, acupuncture dose and follow-up time. According to our results, EA may be associated with a better effect on pain relief and function improvement. This is not surprising as EA provides a stronger treatment dose to acupoints with electrical stimulation on the basis of needle acupuncture. However, the evidence from systematic reviews on the comparative effectiveness of EA versus MA is difficult to interpret due to underpowered subgroup analysis.45 Therefore, conducting larger adequately powered RCTs and identifying populations who may benefit from EA or MA are needed to provide clarity.
Regarding the dose of acupuncture applied, the higher dose showed an association with the larger or more stable effects. Due to the challenge of adequacy of dose, there is an urgent need for research to determine the appropriate acupuncture dose to achieve the optimal treatment effect.46–48 Our findings explained why some clinical trials with inadequate dose acupuncture got negative results and supported the use of adequate dose acupuncture for the treatment of KOA. In the future, research including clinical trials and meta-regression analyses are needed to uncover dose–effect relationship for acupuncture. Also, the description of relevant details for acupuncture dose should be well documented in RCTs.
For follow-up time, despite the lack of difference in pain reduction and function improvement between the acupuncture and sham acupuncture on weeks 26 and 52, the acupuncture exhibited a trend of reduced pain and improved function on 26 weeks. Previous studies49–51 concluded that the benefit of acupuncture relative to controls would be sustained at 12 months. However, due to the limited number of trials and short-term outcomes, it remains unclear whether the factor was truly not associated with outcomes or whether our results were prone to type II error. Further research in this area may help to measure the long-term effect of acupuncture. We encouraged clinicians to provide all this information to patients and engage in shared decision-making.
Conclusions
Although these findings show that acupuncture may provide clinically important effects in reducing pain and improving physical function in patients with KOA, but the evidence is very uncertain. Meanwhile, we identified two potential contributing factors and found EA or a higher dose of acupuncture probably provide greater and more stable effects. However, caution is needed in interpreting our findings because the certainty of evidence was very low. More clinical trials with a low risk of bias and head-to-head comparative studies are required to confirm our findings and provide further direct evidence.
Differences between the protocol and the systematic review
First, we decided to use MCID and the back-transformation method to make better interpretations of our results after we read some articles published recently.36 39 40 The advantage of MCID is making results comparable and highlighting clinical sense. Although MCID was not prespecified, it did not affect the interpretation of the main conclusions. Second, exploratory NMA (described in the published protocol34 but not PROSPERO) is conducted if EA and MA are effective for KOA compared with sham acupuncture. Third, there were insufficient data to perform subgroup analyses according to the stage and Traditional Chinese Medicine type of KOA. Fourth, regarding the meta-regression, we explored the possible sources of heterogeneity via subgroup analyses and decided to report the results of meta-regression in other articles. Fifth, as we mentioned in the protocol, stringent eligibility criteria will be applied to ensure the quality of the included RCTs. We only include true RCTs (‘low risk’ of bias for random sequence generation).
Data availability statement
All data relevant to the study are included in the article or uploaded as online supplemental information.
Ethics statements
Patient consent for publication
Acknowledgments
The authors thank Dr. Mei Han for her valuable guidance during the development of the search strategies. Further, we are grateful to Dr. Xu Wang, Dr. Lu-lu Lin, Dr. Bin Nie, Dr. Chun-sheng Jia, Dr. John T. Farrar, Dr. Jian Kong and Dr. Jun Tang for replying to our author queries and providing us with valuable information in their studies. Some of the articles have previously been presented as part of an abstract at SAR/RCMI PolyU International Research Conference (2024, Hong Kong, China).
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Correction notice Since this article first published, reference 52 has been deleted.
Contributors CL and LW designed the review protocol. CL, YD, HZ, YW and LW developed the search strategy and selected studies. CL, YD, HZ and YW extracted data. CL and LW analysed the data. YD, HZ and XB contributed to the implementation of the study. CL drafted the manuscript. JT, JY and LW revised the manuscript critically for important intellectual content. All authors approved the final version of the article. All authors had access to all the data in the study and could take responsibility for the integrity of the data and the accuracy of the data analysis. CL is the guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding This study was supported in part by the National Natural Science Foundation of China (grant number 82004223).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.