Acupuncture for pain relief in patients with rheumatoid arthritis: A systematic review

Abstract

Objective

To systematically review the efficacy of acupuncture on pain relief in patients with rheumatoid arthritis (RA).

Methods

We performed a comprehensive search of 12 western and Chinese databases and reference lists through March 2008. We included randomized controlled trials with pain as an end point, measured by tender joint count (TJC) or a pain scale. We also reviewed the effect of acupuncture on morning stiffness, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) level. Study quality was assessed by Jadad score. Differences between treatment groups were pooled as mean or median change (P value).

Results

Eight studies met eligibility criteria with a total of 536 subjects. There were 4 placebo-controlled trials and 4 active-controlled trials. Average study duration was 11 weeks. Mean ± SD acupuncture points and sessions were 11 ± 8 and 42 ± 62, respectively. Average duration of needle insertion was 24 minutes. Six studies reported a decrease in pain for acupuncture versus controls; the mean or median changes of acupuncture-decreased TJC pain ranged from 1.5 to 6.5. In addition, 4 studies reported a significant reduction in morning stiffness (mean change −29 minutes), but the difference was nonsignificant versus controls. With regard to inflammatory markers, 5 studies observed a reduction in ESR (mean change −3.9 mm/hour) and 3 observed a CRP level reduction (mean change −2.9 mg/dl); only 1 study showed a significant difference for both ESR and CRP.

Conclusion

Despite some favorable results in active-controlled trials, conflicting evidence exists in placebo-controlled trials concerning the efficacy of acupuncture for RA. Rigorous and well-controlled randomized trials are warranted.

INTRODUCTION

Pain and functional impairment are hallmarks of rheumatoid arthritis (RA), a chronic, progressive, inflammatory disease. RA is characterized by synovial inflammation that leads to joint destruction, resulting in substantial long-term disability and a significantly shorter life expectancy (1, 2). Given the complexity of the therapeutic armamentarium used in RA, nonpharmacologic therapies are increasingly attractive. Recently, complementary and alternative medicine therapies for arthritis have been heavily advertised with broad, attractive claims.

Acupuncture, originating in China more than 3,000 years ago, is one of the most popular sensory stimulation therapies. It is an ancient technique of inserting and manipulating fine needles to stimulate specific anatomic points, also known as acupuncture points or meridian points. Acupuncture is often used for the treatment of conditions associated with pain and is supported by evidence of biologic mechanisms (3, 4). However, despite the fact that acupuncture is widely used for pain relief in a number of conditions including severe knee osteoarthritis (5), acute postoperative pain (6), musculoskeletal disorders (7), and other health conditions (8), evidence of the effect of acupuncture for pain relief in RA is scarce. A recent systematic review of 2 trials of acupuncture on measures of disease activity in RA reported that acupuncture had no effect, but this review was limited by the methodologic weakness inherent in the studies (9). Another narrative review including 63 Chinese studies with a variation of acupuncture therapies in patients with RA concluded that acupuncture is helpful for RA (10). However, in addition to the complexities revealed by evaluation of this type of intervention, most prior studies may have methodologic concerns that limit their interpretation. Therefore, as a first step to determine whether the English/Chinese literature on this question might support a robust meta-analysis, we performed an updated systematic review of all currently available data that included Chinese publications. Our aim was to determine the efficacy of acupuncture for pain relief in patients with RA.

MATERIALS AND METHODS

We performed comprehensive computerized searches of the medical literature in 8 English databases: Medline (from 1950); PsycINFO (from 1806); CAB (from 1910); Health Star (from 1966); EMB Reviewers–Cochrane Database of Systematic Review, ACP Journal Club, and DARE (from 1991); CINAHL (from 1982); Global Health (from 1910); and Alt HealthWatch (from 1969). We also searched 4 Chinese databases: China Hospital Knowledge Database (from 1994), China National Knowledge Infrastructure (from 1915), WanFang Data (from 1980), and Traditional Chinese Medical Database System (from 1984) through March 2008. In addition, we performed hand searches at the Traditional Chinese Medicine libraries of Beijing including the Beijing Chinese Academy of Traditional Medicine and Nanjing Medical Universities (CW). We also manually searched reference lists of selected articles and review articles for additional references. The search terms used were acupuncture, electroacupuncture, rheumatoid arthritis, arthritis, rheumatism, randomized controlled trial, and clinical trial.

Study selection.

We considered randomized controlled trials (RCTs) published in English and Chinese that reported original data comparing acupuncture with placebo or other active therapies for RA. Individuals with RA were classified according to the American College of Rheumatology (ACR; formerly the American Rheumatism Association) criteria (11). The primary outcome measure was assessed by a tender joint count (TJC). The secondary measures included the effect of acupuncture on a pain scale, duration of morning stiffness, and serum concentrations of inflammatory markers such as erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) level.

Data abstraction and study quality.

We extracted data according to prespecified criteria including information on study design, duration, intervention, population characteristics, outcome assessment, and results. Three independent reviewers, one for English (PdP), one for Chinese publications (XC), and one for both (CW), performed the data extraction in duplicate. Any discrepancies were discussed until the authors reached consensus. We described placebo-controlled trials in detail whereas summaries sufficed for the active-controlled trials due to study quality concern.

The methodologic quality of the studies was assessed with the Jadad instrument (12), which includes scoring for randomization (0–2 points), double blinding (0–2 points), and withdrawals (1 point), with scores ranging from 0 to 5 points.

The differences in pain outcomes between treatment groups were pooled as a mean or median change (P value) across the studies. We also provided a qualitative discussion of the findings. The heterogeneity of the studies, including varying outcome measures, active intervention dose, duration, comparison methods, and methodologic limitations, precluded a formal meta-analysis.

RESULTS

We identified 936 abstracts from Chinese and western databases. After initially screening 95 potentially relevant abstracts, we excluded 67 because they did not meet the inclusion criteria (i.e., participants did not have RA, pain was not included as an end point, case reports or duplicate publications). We retrieved and reviewed 28 full articles, 15 of which were published in Chinese. Twenty were excluded due to lack of randomization or absence of a control group (n = 17), major methodologic flaws, and/or insufficient data (n = 3). The study selection process is summarized in Figure 1.

Figure 1. Study selection process (for a more detailed description and discussion of the methods used to search and review the extant literature on this topic, contact the authors). ** = Publication language: 4 in Chinese and 4 in English; publication years: 1974–2007.

Of 8 eligible studies, 4 were published in English (placebo-controlled trials) and 4 in Chinese (active-controlled trials) with a total of 536 subjects with RA. The disease duration ranged between 1 month and >10 years. The studies were conducted in 4 countries (Canada, UK, Brazil, and China) between 1974 and 2007. The characteristics of all studies are summarized in Table 1. The quality of all studies is presented in Table 2.

Table 1. Randomized controlled trials evaluating the effect of acupuncture on RA*

Reference, country	Study design and participants	N	Age†	Duration, weeks	Outcome assessments	Acupuncture intervention‡	Control group intervention	Effect on TJC (mean or median change)§	P¶
* RA = rheumatoid arthritis; TJC = tender joint count; RCTs = randomized controlled trials; yrs = years; VAS = visual analog scale; ACR = American College of Rheumatology; SJC = swollen joint count; GA = global assessment; HAQ = Health Assessment Questionnaire; DAS28 = Disease Activity Score in 28 joints; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; EA = electroacupuncture; TA = traditional acupuncture; AP = acupuncture points; ↓ = decrease; ↑ = increase; GHQ = General Health Questionnaire; min = minutes; ND = no data; RF = rheumatoid factor; MTX = methotrexate; IM = intramuscular; tid = three times a day; bid = twice a day. † Age reported in years as a mean or range. ‡ Acupuncture intervention/dose: number of acupuncture points/number of sessions. § Mean or median difference or improvement was calculated between groups; confidence interval could not be calculated from published data. ¶ P value for comparisons between treatment groups and control group. # Sham acupuncture: needles inserted up to 2 mm, shorter insertion duration, and minimal needle stimulation. ** Median difference. †† Votalin ointment components not reported.
Placebo-controlled RCTs
Tam (13), Hong Kong	Double-blind, randomized, placebo-controlled trial; patients with active RA, mean ± SD disease duration 9.3 ± 6.4 yrs	36	58	10	Baseline and weeks 5, 10, 15, 20: pain VAS, ACR core disease measures (TJC, SJC, patient and physician GA, HAQ), ACR20 at week 10, DAS28, ESR, CRP	Group 1: EA, group 2: TA (6 AP/20 sessions); needle insertion: 30 min	Sham acupuncture#; placebo needles (6 AP/20 sessions)	EA: ↓5.5, TA: ↓4, control group: ↓0.5; difference between groups at week 10**: EA vs. control: ↓5.0, TA vs. control: ↓3.5	0.03 0.01
Zanette (14), Brazil	Double-blind, randomized, placebo-controlled trial; patients with RA for at least 6 months with stable pharmacologic treatment for at least 1 month before inclusion	40	≈50	9	Baseline and weeks 3, 5, 9: ACR20, morning stiffness, DAS, pain VAS, HAQ, TJC, SJC, patient and physician GA, ESR, CRP	TA (16 AP/10 sessions); needle insertion: 20 min	Superficial acupuncture at nonacupuncture points#; (fewer AP/10 sessions)	TA: ↓8.35, control group: ↓2.45; difference between groups: ↓5.9	0.145
David (19), UK	Double-blind, randomized, placebo-controlled crossover trial with a 6-week washout period; patients with RA, mean disease duration 10 yrs	64	18–75	22	Weeks 1 (A), 6 (B), 12 (C), 17 (D), and 22 (E): ESR, CRP, pain VAS, TJC, SJC, tablets/day, GHQ, modified DAS	TA (2 AP/5 sessions); needle insertion: 4 min	Placebo needles (2 AP/5 sessions)	Treatment effect: ↑0.5 95% confidence interval (−1, 1.5)**	0.6
Man (20), Canada	Patients with seropositive RA for ≥5 yrs, for whom bilateral knee pain was a major problem; the study knee was selected at random	20	ND	16	Baseline, 24 hours posttreatment, weekly for 4 weeks, and monthly for 3 months: knee pain reduction (scale 0–4): 0 = no pain, 1 = marked decrease in pain, 2 = moderate decrease in pain, 3 = minimal decrease in pain, 4 = no decrease in pain; knee swelling	EA (3 AP/1 session); needle insertion: 15 min	Placebo needles (3 AP/1 session)	EA (24 hours): 51% improvement, EA (3 months): 23% improvement, control group: 0% improvement in pain scale (0–4)	< 0.05
RCTs compared with active comparators
Liu (15), China	Patients with RA (1987 ACR criteria [11]) with mean disease duration 3.6 yrs	240	42	12	Baseline and 12 weeks: TJC, SJC, grip strength, ESR, CRP, RF	TA (8 AP/180 sessions); needle insertion: 30 min	MTX IM injection: week 1: 5 mg, week 2: 10 mg, week 3: 15 mg, + diclofenac (20 mg/day)	TA: ↓16.6, control group: ↓10.1; difference between groups: ↓6.5	< 0.05
Jiang (16), China	Functional class 1 and 2 patients with RA (1987 ACR criteria [11]) with mean disease duration 4.5 yrs	60	45	4	Baseline and 4 weeks: TJC, SJC, morning stiffness, ESR, CRP, RF	TA (20 AP/15 sessions); needle insertion: 30 min	Indomethacin (25 mg tid)	TA: ↓5.1, control group: ↑0.6; difference between groups: ↓4.5	< 0.05
Cui (17), China	Patients with RA (1987 ACR criteria [11]) and no data for disease duration	62	ND	12	Baseline and 12 weeks: TJC, pain VAS, SJC, morning stiffness	TA (9 AP/90 sessions); needle insertion: 20–30 min	Votalin ointment (bid) ††	TA: ↓8.3, control group: ↓6.8; difference between groups: ↓1.5	< 0.01
Zhou (18), China	Functional class 1 and 2 patients with RA (1987 ACR criteria [11]) with mean ± SD disease duration 3 ± 5 yrs	45	18–65	4	Baseline and 4 weeks: TJC, SJC, grip strength, RF, ESR, CRP, morning stiffness	TA (24 AP/15 sessions); needle insertion: 40 min	Indomethacin (25 mg tid)	TA: ↓7.1, control group: 4.0; difference between groups: ↓3.1	< 0.01

Table 2. Jadad scores of included studies*

Reference, country	Described as randomized? (scoring: +1)†	Randomization described and appropriate? (scoring: −1, 0, +1)‡	Described as double blind? (scoring: +1)§	Double blinding described and appropriate? (scoring: −1, 0, +1)¶	No. and reasons for withdrawals described in each group? (scoring: +1)#	Total Jadad score (maximum 5)
* Items on the Jadad scale are scored with either 1 point for each “yes” or 0 points for each “no” (range 0–5). RCTs = randomized controlled trials. † Was the study described as randomized (this includes the use of words such as randomly, random, and randomization)? ‡ Add 1 point if the method to generate the sequence of randomization was described and it was appropriate (table of random numbers, computer generated, etc.) or deduct 1 point if the method to generate the sequence of randomization was described and it was inappropriate (patients were allocated alternately, or according to date of birth, hospital number, etc.). § Was the study described as double blind? ¶ Add 1 point if, for question 2, the method of double blinding was described and it was appropriate (identical placebo, active placebo, dummy, etc.) or deduct 1 point if, for question 2, the study was described as double blind but the method of blinding was inappropriate (e.g., comparison of tablet versus injection with no double dummy). # Was there a description of withdrawals and dropouts?
Placebo-controlled RCTs
Tam (13), Hong Kong	+1	+1	+1	+1	+1	5
Zanette (14), Brazil	+1	+1	+1	+1	+1	5
David (19), UK	+1	+1	+1	+1	+1	5
Man (20), Canada	+1	0	+1	+1	0	3
RCTs compared with active drug interventions (nonplacebo-controlled)
Liu (15), China	+1	+1	0	0	+1	3
Jiang (16), China	+1	0	0	0	+1	3
Cui (17), China	+1	0	0	0	+1	2
Zhou (18), China	+1	0	0	0	+1	2

Seven trials used traditional acupuncture (TA) (13–19), 2 used electroacupuncture (EA) (13, 20), and 1 used both (13). Four trials used placebo needles (sham acupuncture or incorrectly placed needles) as controls (13, 14, 19, 20). The other 4 studies published in China used a variety of active interventions in the control groups, including methotrexate (MTX) (15), topical Votalin ointment (Pui Industrial, China) (17), and nonsteroidal antiinflammatory drugs (NSAIDs) (16, 18).

Seven studies reported TJC as the main end point, and 1 study used a pain scale (20). The mean ± SD study duration was 11 ± 6 weeks (range 4–22 weeks). The mean ± SD number of acupuncture points used was 11 ± 8 and the average duration of needle insertion was 24 minutes (range 4–40 minutes). The mean number of acupuncture sessions was 42 ± 62.

Collectively, 6 studies found a decrease in pain for acupuncture compared with controls, and 5 showed a statistically significant improvement (13, 15–18). Compared with controls, the mean or median changes in acupuncture-decreased TJC pain ranged from 1.5 to 6.5. In addition, 4 studies reported a significant reduction in morning stiffness (mean change −29 minutes), but the difference was nonsignificant compared with controls (14–16, 18). With regard to inflammatory markers, 5 studies observed a reduction in ESR (mean change –3.9 mm/hour) (13–16, 18) and 3 noted a reduction in CRP level (mean change −2.9 mg/dl) (14, 16, 18), but only 1 study showed a significant difference for both ESR and CRP (16).

Placebo-controlled trials.

The 4 placebo-controlled trials involved 160 participants. Of these trials, 3 had optimal quality and 1 had moderate quality (13, 14, 19, 20). Two high-quality, double-blind, randomized, placebo-controlled trials evaluated the effects of either TA or EA versus sham acupuncture (13, 14). Tam et al (13) conducted a trial in Hong Kong among 36 patients with RA with a mean disease duration of 9.3 years. Patients were randomly assigned to 3 groups: TA, EA, and sham (placebo needles). Patients received a total of 20 sessions for 10 weeks using 6 acupuncture points. The authors found that TJC improved for both the EA group (a TJC decrease of 5.5 versus a decrease of 0.5; P < 0.05) and TA group (a TJC decrease of 4 versus a decrease of 0.5; P < 0.05) compared with the control group. Physician's global score was also significantly reduced for the EA group and patient's global score was significantly reduced for the TA group. Of note, the pain score remained unchanged in all 3 groups and the ACR core disease measures and Disease Activity Score in 28 joints (DAS28) were not achieved at week 10.

In a contemporaneous study conducted by Zanette et al (14), 40 patients with longstanding RA were randomized to a TA or a control group using superficial acupuncture at nonacupuncture points. All participants received a total of 10 sessions, twice a week for 5 consecutive weeks. TJC improved for the TA group compared with the control group (a TJC decrease of 8.35 versus a decrease of 2.45; P = 0.145). There was no significant difference between treatment groups regarding the ACR 20% improvement criteria (ACR20; primary outcome) after the 5th and 10th sessions. At the last visit, there was a trend in favor of the acupuncture intervention compared with the control group (40% versus 10% achieved ACR20 criteria; P = 0.07). In addition, there was a significant difference favoring the TA group for physician's global assessment of the treatment (P = 0.012) and patient's and physician's global assessments of disease activity (P = 0.003 versus P = 0.011), but there was no difference for other end points. The authors concluded that the negative result could be related to the small sample size, selection of patients, type of acupuncture protocol applied, and difficulties in establishing an appropriate control group.

A double-blind, randomized, placebo-controlled trial conducted by David et al used a 6-week crossover design comparing TA with sham acupuncture in 64 patients (19). There were a total of 5 weekly sessions. After a washout period of 6 weeks, participants were crossed over into the other intervention arm (acupuncture or control) for an additional 5 weeks. TJC, swollen joint count, pain scale, patient's and physician's global assessments, DAS28, ESR, and CRP level were assessed at baseline and at the end of both intervention periods. The authors found no significant differences between the intervention and control groups for any of the end points at the end of both intervention periods and a followup assessment.

An early RCT conducted by Man and Baragar (20) used a parallel design for 20 participants with seropositive RA. Patients were randomly assigned to either the EA or control group. Pain was assessed with a pain scale ranging from 0 to 4 at 24 hours after treatment. The authors reported that EA resulted in a significant moderate and marked decrease in knee pain in 80% of the participants (60% and 20%, respectively) compared with no pain reduction in the controls. At 3 months, 70% of the participants in the EA group reported a significant minimal or moderate decrease in pain compared with no pain reduction in the control group. However, because no baseline data were reported, we estimated the percentage improvement on the pain scale from the published figure (51% and 23% at 24 hours and 3 months, respectively) for the EA group compared with no change in the control group.

Active-controlled trials.

Four active drug-controlled Chinese studies with modest quality including 407 subjects were conducted in China since 2000 (15–18). The mean study duration was 8 weeks with 15–180 sessions. The number of acupuncture points varied from 8 to 24. In the control groups, 2 studies used indomethacin (25 mg 3 times a day for 4 weeks) (16, 18) and 1 study used MTX (5 mg during week 1, 10 mg during week 2, 15 mg during week 3) and diclofenac (20 mg twice a day for 3 months) (15). One study compared acupuncture with topical Votalin ointment (twice a day) (17), which may be considered as an NSAID. Every Chinese publication consistently reported that acupuncture treatment was associated with a significant decrease in pain (TJC mean change −3.9) compared with controls. In addition, 3 studies found a significant reduction in morning stiffness (mean change −29 minutes) compared with controls (15, 16, 18). Three studies observed a reduction in ESR (mean change −5.5 mm/hour) (15, 16, 18) and 2 found a reduction in CRP level (mean change −3.0 mg/dl) (16, 18), but only 1 showed a significant difference for ESR and CRP levels (16). No dropouts were reported. Although these trials concluded that acupuncture was effective in relieving symptoms of RA, the long-term benefits remain unknown.

DISCUSSION

We evaluated 3 decades' worth of the cumulative literature on acupuncture for the treatment of RA in both Chinese and western populations. Despite some favorable results in 1 placebo-controlled trial and 4 active-controlled trials, conflicting evidence remains regarding the efficacy of acupuncture for RA in the placebo-controlled trials. While an early RCT reported that 1 EA session, compared with sham acupuncture, significantly improved knee pain in 20 patients with seropositive RA (20), a later randomized crossover study of 56 patients with RA reported no significant differences between traditional and sham acupuncture in pain relief or inflammation (19). A recent systematic review that utilized these 2 trials in the analysis concluded that the evidence is limited due to methodologic considerations, such as the type of acupuncture (acupuncture versus EA), the site of intervention, and the small sample size of the included studies (9). In the 2 most recent and well-designed double-blind, randomized, placebo-controlled trials (13, 14), one concluded that traditional acupuncture and EA significantly improved TJC compared with placebo (13), while the other, in contradistinction, found that TA was not superior to placebo (14).

All 4 active comparator trials published in China concluded that acupuncture treatment was associated with a significant decrease in TJC and was effective in relieving other symptoms of RA compared with controls. However, the methodologic limitations of the Chinese studies included inappropriate control interventions (noncomparable), no double-blind interventions, inadequate description of the randomization process, and scarce use of validated outcome measures. Notwithstanding these discrepancies in methods, the Chinese studies also used active drugs in place of placebo acupuncture.

None of the 8 trials in this review have formally adopted the acupuncture-specific Standards for Reporting Interventions in Controlled Trials of Acupuncture (STRICTA) guidelines (21). Some parts of the STRICTA checklist are ambiguous (22), and the authors may fail to appreciate the underlying need for reporting some items.

In contrast to those RCTs published in English that, for the most part, proved negative, every Chinese study we reviewed (rejected or retained) reported positive results. Of note, the Chinese studies included here tended to use acupuncture interventions with higher doses (intensity and duration) in active-controlled RCTs than in placebo-controlled RCTs, which might be one source of heterogeneity. Different levels of expertise in acupuncture interventions as well as differences in methodologic rigor between western and Chinese studies may be other potential sources of heterogeneity. In addition, publication bias may vary across countries and cultures (23).

In addition, the comparison methods (e.g., mean, median, and percentage improvements) prohibited a formal meta-analysis. It is difficult to obtain overall accurate quantitative estimates of treatment effects due to the heterogeneity of the studies, including acupuncture dose/intensity and duration variations. For instance, the number of acupuncture points ranged from 1 to 24, the duration of needle insertion ranged from 4 to 40 minutes, and the number of sessions varied from 1 to 180. The time elapsed between sessions also fluctuated. For all of these reasons, evidence for the usefulness of acupuncture in RA is very modest and uncertain, and the long-term benefits remain unknown.

Because it is difficult to compare such divergent trials, and due to the lack of standardized treatment protocol, future studies should focus on the optimum dose (effective evidence-based dose/intensity including number of acupuncture points, frequency of acupuncture sessions, and intervention duration) of acupuncture treatments for RA. It is possible that some studies, although designed correctly from a western scientific approach, do not have the correct Chinese medical approach (i.e., inappropriate dose/intensity and duration of acupuncture, insufficient expertise of acupuncturist, inappropriate needle insertion and manipulation techniques). This might explain why no specific effects were observed in western studies.

With regard to other rheumatic conditions, 2 recent publications on the effect of acupuncture in osteoarthritis noted different observations. A systematic review concluded that acupuncture is significantly superior to sham acupuncture in improving pain and function in patients with osteoarthritis (24). In contrast, a meta-analysis observed that sham-controlled trials demonstrated clinically irrelevant short-term benefits of acupuncture, while waiting list–controlled trials demonstrated clinically relevant short-term benefits in pain and function, suggesting that placebo or expectation effects may be involved. Participants who received acupuncture also reported clinically relevant short- and long-term improvements in pain and function compared with patients in usual care groups (25).

Explanatory mechanisms from eastern and western biologic theory provide a supposed rationale for the effectiveness of acupuncture in treating the chronic inflammatory nature of RA (26–29). However, with the exception of knee osteoarthritis as mentioned above, our findings are similar to most acupuncture trials in chronic pain. Indeed, there are troublesome findings of nonsuperiority of acupuncture over shams; for example, the larger literature on clinical trials of acupuncture in pain has failed to show a significant improvement over sham acupuncture (30–32).

The field can be improved by applying the STRICTA guidelines (21), in addition to using the following recommendations. Future large-scale studies should have double-blind, randomized, placebo-controlled trial designs, and use appropriate control groups (such as physiologically inactive yet credible shams) (25) and validated longer-term outcome measures for RA to further understand its long-term effects. The level of qualifications and experience of acupuncturists in both eastern and western countries should be compared and recorded. Standardized, generalizable, and optimized treatment protocols should be developed and, importantly, dose/intensity superiority studies should be carried out in RA. There also is the need for registration of trials for both eastern and western studies so that potential reporting/publication bias can be addressed. In addition, the explication mechanisms of the physiologic and biologic effects of acupuncture in RA need to be further determined. This first comprehensive review of acupuncture for RA provides guidance for future research.

AUTHOR CONTRIBUTIONS

Dr. Wang 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.

Study design. Wang, de Pablo, McAlindon.

Acquisition of data. Wang, de Pablo, Chen, Schmid.

Analysis and interpretation of data. Wang, de Pablo, Chen, Schmid, McAlindon.

Manuscript preparation. Wang, de Pablo, Chen, Schmid, McAlindon.

Statistical analysis. Wang, de Pablo, Schmid.

Acknowledgements

The authors thank Drs. Yannan Sun and Qunhao Zhang for retrieving Chinese articles and Bruce Kupelnick and Aghogho Okparavero for assistance in reviewing this manuscript.