The effectiveness of acupuncture as an adjunctive therapy to oral pharmacological medication in patient with knee osteoarthritis: A systematic review and meta-analysis

Abstract

Background: 

We aimed to find out whether the combined treatment of acupuncture and oral medication is more effective than sole oral medication in reducing pain and improving knee function at the end of treatment and after short-term period (4–6 weeks after treatment). Second, if it is effective, we investigated whether the effect surpasses the minimal clinically important difference.

Methods: 

Articles published between January 1, 1992, and August 31, 2022, were searched in PubMed, Cochrane, and Embase. The PICO (population, intervention, comparison, and outcome) of this study are as follows: Population: knee osteoarthritis patients; Intervention: acupuncture (non-sham acupuncture) + oral medication (analgesic or non-steroidal anti-inflammatory drugs); Comparison: oral medication (analgesic or non-steroidal anti-inflammatory drugs); Outcome: visual analog scale (VAS) or Western Ontario and McMaster University (WOMAC) osteoarthritis index

Results: 

The combined treatment of oral medication and adjuvant acupuncture showed statistically significant improvement in VAS and WOMAC scores at the end of acupuncture treatment and short-term follow-up time (between 4 and 6 weeks after acupuncture). In addition, the degree of improvement of VAS and WOMAC index showed effects beyond minimal clinically important differences compared to pretreatment at both the end of acupuncture treatment and the short-term follow-up of acupuncture treatment.

Conclusion: 

The existing evidence suggests that adjuvant acupuncture may play a role in the treatment of knee osteoarthritis. However, physicians should be aware of adverse effects such as hematoma in adjuvant acupuncture treatment.

1. Introduction

Osteoarthritis of the knee is a representative degenerative disease,^[1] and its prevalence is increasing with the aging of the population.^[2] Therefore, treatment methods for osteoarthritis of the knee are diversifying.^[3] Initial treatments include non-pharmaceutical treatments such as weight control and lifestyle modification,^[4–7] but patients who do not respond are typically prescribed medications like symptomatic slow-acting drugs for osteoarthritis, non-steroidal anti-inflammatory drugs (NSAID), and pain killers.^[8] The goal of medication is to improve knee function along with reducing pain. Pain can lead to chronic pain if it is not treated properly in the early stages, which can eventually lead to a vicious cycle.^[9] Therefore, if the effect of pain relief during initial treatment is small, various mechanisms of drugs are mixed and prescribed to further alleviate pain.^[10] However, persistent, high-dose medication intake has many adverse effects.^[11] In particular, the Food and Drug Administration mandates the attachment of warnings to all NSAIDs that can cause cardiovascular complications.^[12] Since high doses and long-term prescriptions of oral medication are burdensome for both patients and prescription doctors, oral medication and alternative treatment methods are combined for pain relief.

Acupuncture has already been actively used as a pain treatment in East Asia,^[13] and many studies have shown its efficacy and safety.^[13,14] In a meta-analysis comparing and analyzing the effects of acupuncture treatment with other single treatments, the main results are that acupuncture treatment is not inferior to other treatments.^[15,16] In addition, acupuncture is recognized as one of the symptomatic treatments in the guideline of many academic societies.^[17] Therefore, a study analyzing the advantages of combining acupuncture with oral medication, which is the main treatment for arthritis, is an important study that can give clinicians a new treatment option. However, meta-analysis has not been done so far.

Therefore, the authors assumed that in the treatment of patients with knee osteoarthritis, the combination treatment of oral medication (analgesic or NSAID) and acupuncture treatment may be more effective in terms of pain reduction and functional improvement than sole oral medication treatment. The purpose of this study is as follows. First, when treating patients with knee osteoarthritis, we would like to find out whether the combined treatment of acupuncture and oral medication is more effective than sole oral medication in reducing pain and improving knee function at the end of treatment and after short-term period (4–6 weeks after treatment). Second, if it is effective, we investigated whether the effect surpasses the minimal clinically important difference (MCID).^[18–20] This study was conducted without Institutional Review Board approval as a meta-analysis study.

2. Methods

2.1. Search strategy

The PICO (population, intervention, comparison, and outcome) of this study are as follows.

1. Population: knee osteoarthritis patients.
2. Intervention: acupuncture (non-sham acupuncture) + oral medication (analgesic or NSAID).
3. Comparison: oral medication(analgesic or NSAID).
4. Outcome: visual analog scale (VAS) or Western Ontario and McMaster University (WOMAC) osteoarthritis index.

Articles published between January 1, 1992, and August 31, 2022, were searched in PubMed, Cochrane, and Embase using the following key phrases (Table 1).

Table 1 - Articles were searched in PubMed using the following key phrases.

Pubmed; search on August 31, 2022
Search	Query
#1	“knee osteoarthritis” [Ti/Ab] or “knee pain” [Ti/Ab] or “knee arthritis” [Ti/Ab] or “osteoarthritis of the knee” [Ti/Ab] or “knee joint osteoarthritis” [Ti/Ab] or “degenerative knee arthritis” [Ti/Ab]
#2	“acupuncture” [Ti/Ab] or “electroacupuncture” [Ti/Ab]
#3	#1 and #2

2.2. Inclusion and exclusion criteria

The following studies were included in this study:

1. A study with an experimental group where oral medication-adjuvant acupuncture combination treatment was performed to treat knee osteoarthritis.
2. A study with a control group where sole oral medication treatment was performed to treat knee osteoarthritis.
3. Studies written in English.

The exclusion criteria were as follows:

1. Review articles, case reports, protocol, and conference presentations.
2. A study with an experimental group where sham acupuncture was performed instead of real acupuncture.

2.2.1. Data extraction.

Data for meta-analysis were independently investigated by 2 researchers (W.K.C. and S.G.K.). Duplicate studies were excluded, and studies that met the eligibility criteria were selected. Studies were evaluated for eligibility by reviewing the title and abstract. After reading the full text, studies were finally selected for inclusion in the meta-analysis, and discrepancies were resolved through discussion. Research characteristics, study design, number of patients (intervention and control groups), schedules of acupuncture treatment, outcome measurements, time, and number of measurements were investigated.

The outcome variables were VAS and WOMAC index. When the scale of VAS ranged from 0 to 10 was used, the value presented in the paper was multiplied by 10 to standardize the range from 0 to 100. A low value of VAS means no pain, and a high value means severe pain. We used a Korean-translated validated version of WOMAC. The WOMAC consists of a total of 24 items, consisting of 5 items for pain, 2 items for stiffness, and 17 items for body function.^[21] When the scale of the WOMAC total ranged from 0 to 2400 was used, the value presented in the paper was divided by 100 and multiplied by 4 to standardize the range from 0 to 96. When a scale ranging from 0 to 10 was used for the WOMAC total, the value presented in the paper was multiplied by 9.6 for standardization. In a similar way, WOMAC pain was standardized on a range of 0 to 20, WOMAC stiffness was standardized on a range of 0 to 8, and WOMAC function was standardized on a range of 0 to 68.

We collected the values of the evaluation variables such as VAS and WOMAC index presented in the paper according to the evaluation time points. Then, the amount of change from baseline to just-after acupuncture and the amount of change from baseline to follow-up were calculated. Since this corresponds to the amount of change in the dependent group, the mean value and standard deviation for change between baseline and post-time (just-after acupuncture or follow-up) were calculated using the following formulas, and the correlation was derived from a different study that presented the standard deviation for change.

${\displaystyle \begin{array}{l}{\displaystyle Mea{n}_{change}=Mea{n}_{baseline}-Mea{n}_{posttime}.}\end{array}}$

${\displaystyle \begin{array}{l}{\displaystyle S{D}_{change}=\sqrt{\begin{array}{c}S{D}_{baseline}^2+S{D}_{posttime}^2\\ -2\times Correlation\times S{D}_{baseline}\times S{D}_{posttime}\end{array}}.}\end{array}}$

2.2.2. Quality assessment.

The quality assessment of the risk of bias for the included studies was assessed using the Cochrane Collaboration’s Handbook. It consisted of random sequence generation, allocation concealment, blinding of participants, incomplete outcome data, selective reporting, and other potential sources of bias. The judgments of bias were expressed as “low risk,” “high risk,” or “unclear risk.”

2.2.3. Statistical analysis.

RevMan v.5.3 software (http://tech.cochrane.org/revman) was used for statistical analysis of the pooled data. For each analysis, a heterogeneity test was performed using I² statistics, which measures the extent of inconsistency among results. P values of <.05 were considered as having substantial heterogeneity, and the random-effects model was used for the analysis of the data. In contrast, when P values were ≥.05, pooled data was homogenous, and the fixed effects model was applied. We analyzed the mean difference between the only medication (analgesic or NSAID) and medication plus adjuvant acupuncture treatment. Further, the 95% confidence interval (CI) was used in the analysis. For the evaluation of publication bias, the funnel plot and the Egger test were used. P value < .05 was considered statistically significant.

For sensitivity analysis, root mean square error (RMSE) was calculated by using the effect size and total effect size calculated for each evaluation variable except for the i-th study, i = 1, ..., n (total number of studies). VAS ranged from 0 to 100 and the WOMAC index ranged from 0 to 96 for each evaluation variable, so the effect size was calculated using the standardized mean difference value. For RMSE calculation, the following formula was used.

${\displaystyle \begin{array}{l}{\displaystyle RMSE=\sqrt{\frac{1}{k}{\displaystyle \underset{i=1}{{\displaystyle \sum ^k}}}{(Totaleffectsize-Totaleffectsiz{e}_{(i)})}^2}}\end{array}}$

For RMSE, closer to 0 means that the overall effect size is not sensitive to the study, and a larger value means that the overall effect size is sensitive to the study.

3. Results

In the databases, 749 articles were searched, and 193 duplicated articles were removed (Fig. 1). After screening for eligibility based on a review of the title and abstract, 8 articles were included for full-text reading. After a detailed assessment, 2 articles were excluded: 1 study used only sham acupuncture, and 1 study was not written in English. Accordingly, 6 studies were finally included in our meta-analysis.^[22–27] The characteristics of the studies included in the research are also described in Table 2.

Table 2 - Six studies were finally included in our meta-analysis.

Reference	Study design	Intervention	Control	Number of patients (intervention/control)	Schedules of acupuncture treatment	Outcome measurements	Time and number of measurements
Berman et al (1999)	RCT	Acupuncture + oral medication	Oral medication	29/29	Biweekly for 8 wk	WOMAC	Baselines 4 wk 8 wk 12 wk
Kwak et al (2022)	RCT	Acupuncture + oral NSAID	Oral NSAID (COX-2 inhibitor)	12/14	12 times during first 6 wk	VAS WOMAC SF 36	Baselines 6 wk 12 wk
Scharf et al (2006)	RCT	Acupuncture + oral NSAID	Oral NSAID (diclofenac 150 mg/d or rofecoxib 25 mg/d)	326/316	10 times (+5)* Between 2 wk and 8 wk after the start of the study	WOMAC SF 12	Baselines 13 wk 26 wk
Mavrommatis et al (2012)	RCT	Acupuncture + oral NSAID	Oral NSAID (etoricoxib 60 mg/d)	40/40 (4 wk) 39/38 (8 wk, 12 wk)	Biweekly for 8 wk	VAS WOMAC SF 36	Baselines 4 wk 8 wk 12 wk
Sangdee et al (2002)	RCT	Acupuncture + oral NSAID	Oral NSAID (dicolofenac)	49/49 (initial) 46/49	3 times a wk for 4 wk	VAS WOMAC	Baselines 4 wk
Tukmachi et al (2004)	RCT	Acupuncture + oral medication	Oral medication (NSAID or analgesic)	10/10	Biweekly for 5 wk	VAS WOMAC	Baselines 5 wk

NSAID = non-steroidal anti-inflammatory drugs, RCT = randomized controlled trial, VAS = visual analog scale, WOMAC = Western Ontario and McMaster University.

^*+5 indicates 5 additional sessions or visits if treatment was graded as partially successful after 6 weeks.

Figure 1.: Flowchart showing the search results.

3.1. Study characteristics

As a result of reviewing titles and abstracts, a total of 6 RCT studies were included. In the 6 studies, 466 participants were for only oral medication (analgesic or NSAID) and 473 participants for medication plus adjuvant acupuncture treatment. In all 6 studies, measurement of treatment effect was performed at the end of acupuncture treatment and short-term follow-up (between 4 and 6 weeks after termination). There were 4 studies measuring VAS as a result of treatment, and a total of 5 studies measuring WOMAC. With reference to previous studies, MCIDs of VAS and WOMAC were considered to be 20%.^[19] In this study, the MCID of VAS was 11.50 points based on 100 points, and the MCID was 9.58 points out of 96 points for the WOMAC total. The MCIDs for WOMAC function, WOMAC pain, and WOMAC stiffness were 7.10, 1.96, and 0.86 points, respectively.

3.2. Risk of bias

The results of a qualitative assessment conducted using the Cochrane Collaboration’s Handbook are shown in Figure 2. These studies were evaluated for low risk for random sequence generation, allocation concealment, blinding of participants, incomplete outcome data, and selective reporting. One study was evaluated for unclear risk for allocation concealment and low risk for random sequence generation, blinding of participants, incomplete outcome data, and selective reporting.

Figure 2.: Risk of bias summary.

3.3. Meta-analysis results

In performing meta-analysis, a fixed effect model was used when the P value was >.05 as a result of the homogeneity test, and a random effect model was used when the P value was <.05. Mean difference and 95% CI were presented in the forest plot for VAS and WOMAC score of change from baseline to just-after acupuncture between only medication group and medication plus adjuvant acupuncture treatment group (Fig. 3) and forest plots for VAS and WOMAC score of change from baseline to follow-up between only medication group and medication plus adjuvant acupuncture treatment group (Fig. 4).

Figure 3.: Forest plots for VAS and WOMAC score of change from baseline to just-after acupuncture between only medication group and medication + acupuncture. VAS = visual analog scale, WOMAC = Western Ontario and McMaster University.

Figure 4.: Forest plots for VAS and WOMAC score of change from baseline to follow-up between only medication group and medication + acupuncture. VAS = visual analog scale, WOMAC = Western Ontario and McMaster University.

There was a statistically significant difference between the 2 groups in the amount of change from baseline to just-after acupuncture in all variables such as VAS, WOMAC, WOMAC pain, WOMAC stiffness, and WOMAC function. For the effect size of VAS, the mean difference was 21.30 (95% CI: 6.13–36.47), and for the effect size of WOMAC, the mean difference was 12.69 (95% CI: 7.61–17.77). This means that the amount of change in VAS and WOMAC scores was significantly reduced in the medication plus adjuvant acupuncture treatment group just-after acupuncture time than in the only medication group.

There was a statistically significant difference between the 2 groups in the amount of change from baseline to follow-up in all variables such as VAS, WOMAC, WOMAC pain, WOMAC stiffness, and WOMAC function. For the effect size of VAS, the mean difference was 23.05 (95% CI: 19.67–26.43), and for the effect size of WOMAC, the mean difference was 13.67 (95% CI: 10.16–17.18). This means that the amount of change in VAS and WOMAC scores was significantly reduced in the medication plus adjuvant acupuncture treatment group follow-up time than in the only medication group.

3.4. Sensitivity analysis

For the change from baseline to just-after acupuncture, the RMSE values for the effect sizes of VAS, WOMAC total, WOMAC pain, WOMAC stiffness, and WOMAC function were calculated as 0.302, 0.185, 0.138, 0.166, and 0.496, respectively. For the change from baseline to follow-up, the RMSE values for the effect sizes of WOMAC total, WOMAC pain, WOMAC stiffness, and WOMAC function at the time of follow-up were calculated as 0.200, 0.212, and 0.239, respectively. RMSE was not calculated when there were only 2 studies. RMSE values in all cases were calculated to be <0.5 except for the RMSE for VAS. It can be seen that the overall effect size is not sensitive depending on the study.

3.5. Publication bias

On the basis of a few distinct methods, 2 of the authors (W.K.C. and S.G.K.) individually assessed the risk of bias. The risk of publication bias was determined using a funnel plot and the Egger test. A funnel plot was produced to investigate the risk of publication bias, as seen in Figures 5 and 6. Performing a visual inspection suggested some funnel plot asymmetry. In addition, the Egger test was not statistically significant in all cases, indicating that the observation of asymmetry was not supported and that there may not be a risk of publication bias.

Figure 5.: Funnel plots for VAS and WOMAC score of change from baseline to just-after acupuncture. VAS = visual analog scale, WOMAC = Western Ontario and McMaster University.

Figure 6.: Funnel plots for VAS and WOMAC score of change from baseline to follow-up. VAS = visual analog scale, WOMAC = Western Ontario and McMaster University.

4. Discussion

As a result of this meta-analysis, the combined treatment of oral medication and adjuvant acupuncture, which are commonly prescribed for knee osteoarthritis patients, showed statistically significant improvement in VAS and WOMAC scores at the end of acupuncture treatment and short-term follow-up time (between 4 and 6 weeks after acupuncture). In addition, the degree of improvement of VAS and WOMAC index showed effects beyond MCID compared to pretreatment at both the end of acupuncture treatment and the short-term follow-up of acupuncture treatment.

Many comparative studies have already been conducted on the effectiveness of single acupuncture treatment and other single treatments.^[15,16,28] However, acupuncture is classified as a symptomatic treatment rather than a primary treatment in academies associated with knee osteoarthritis with various recommendation levels.^[6,7,29,30] Instead, most academies recommend non-pharmaceutical treatment as first-line therapy. When non-pharmaceutical treatment is not effective, oral medication is most commonly used. Various types of oral medication are used, ranging from symptomatic slow-acting drugs for osteoarthritis to narcotic painkillers.^[8] Among them, the most commonly used ones are NSAIDs and analgesics. However, NSAIDs typically have gastrointestinal and cardiovascular risk problems, which depend on the dose and duration of the drug. Due to this problem, various adjuvant treatments are being used simultaneously with oral medication to reduce the amount of time and dose of oral medications by accelerating pain reduction and functional recovery.^[11,31]

Based on this background, this meta-analysis was conducted. The results show that when acupuncture is used in combination with drug treatment, the pain reduction effect and knee function improvement surpass MCID at the end of acupuncture treatment and short-term follow-up period.

Regarding the safety of adjuvant acupuncture treatment, 3 RCT studies showed that there was no special adverse effect, and there was no statistical difference in liver and kidney function tests compared to the oral medication group. In a Scharf study with the largest number of enrolls, the incidence of hematoma in patients treated with adjuvant acupuncture was higher than in the medication group.^[23]

There are several limitations to this study.

First, in all 6 studies, compliance with the use of oral medication was not investigated. However, since oral medication generally stops when pain decreases and function improves, compliance is not considered to be a clinically important factor.^[32] Second, all 6 studies studied the short-term treatment time, and no analysis was conducted on the results of the mid-to-long-term study >6 weeks after treatment. Third, we searched only for articles written in English, not including papers written in other languages.

5. Conclusion

The existing evidence suggests that adjuvant acupuncture may play a role in the treatment of knee osteoarthritis. However, physicians should be aware of adverse effects such as hematoma in adjuvant acupuncture treatment.

Author contributions

Conceptualization: Sang Gyu Kwak, Won-Kee Choi.

Data curation: Sang Gyu Kwak, Jae Bum Kwon, Young Woo Seo, Won-Kee Choi.

Formal analysis: Sang Gyu Kwak, Jae Bum Kwon, Young Woo Seo, Won-Kee Choi.

Funding acquisition: Sang Gyu Kwak.

Investigation: Sang Gyu Kwak, Jae Bum Kwon, Young Woo Seo, Won-Kee Choi.

Methodology: Sang Gyu Kwak.

Project administration: Sang Gyu Kwak.

Resources: Sang Gyu Kwak, Won-Kee Choi.

Software: Sang Gyu Kwak, Won-Kee Choi.

Supervision: Sang Gyu Kwak, Won-Kee Choi.

Validation: Sang Gyu Kwak, Won-Kee Choi.

Visualization: Sang Gyu Kwak, Won-Kee Choi.

Writing – original draft: Sang Gyu Kwak, Won-Kee Choi.

Writing – review & editing: Sang Gyu Kwak, Won-Kee Choi.