Rheumatoid arthritis

Rheumatoid arthritis (RA) is an autoimmune disease in which the immune system attacks the lining of the joints, most often in the hands and feet, causing pain, swelling and stiffness. It is treated with disease-modifying antirheumatic drugs (DMARDs), such as methotrexate, which slow the disease and the joint damage it causes. Trials of acupuncture in RA measure pain, the number of tender and swollen joints, blood markers of inflammation such as C-reactive protein (CRP) and the erythrocyte sedimentation rate (ESR), and the Disease Activity Score in 28 joints (DAS28), which combines several of these into one figure. No trial on this page tested acupuncture as a replacement for DMARDs: most of the later trials gave it alongside them. Two kinds of comparison carry different weight. Against sham acupuncture, a procedure made to resemble acupuncture, a trial can separate the needling from the attention and expectation that come with it; Sham acupuncture describes the kinds in use. Against drugs alone, it cannot.

The Cochrane review

The Cochrane review by Casimiro and colleagues, published in October 2005, updated a first version from 2002. It searched to May 2005, analysed only trials published in English or French, and included 2 trials with 84 people. In the first, acupuncture made no significant difference against placebo to pain, swollen or tender joints, CRP, ESR, disease activity, general health or the use of painkillers. In the second, people given electroacupuncture at the knee reported less knee pain than those given placebo, 24 hours and 4 months after treatment, but the reviewers judged that trial of poor quality and too small to support a recommendation. Side effects were not measured in either trial. The review predates the routine use of GRADE, the scale reviews now use to rate how far a result can be trusted, and gives no rating in that form. It was edited in 2010 with no change to its conclusions, and no later version had been published by October 2026, so the only Cochrane assessment of acupuncture for RA rests on searches more than 20 years old.

Later reviews

Seven later reviews pooled trials the Cochrane review did not reach, several of them searching Chinese databases. None of their abstracts reports a GRADE rating.

The two 2008 reviews separated sham from active controls. Wang and colleagues found some favourable results in trials against active treatment and conflicting evidence in those against placebo. Lee and colleagues pooled the 4 sham-controlled trials, which held 88 people, and found no significant difference in pain on a 10 cm scale; 3 trials of acupuncture combined with moxibustion, the burning of a herb on or near the skin, showed no advantage over conventional drugs in response rate or pain.

Seca and colleagues excluded 9 of 22 eligible trials after appraising their quality; 10 of the 13 that remained were published in China and reported favourable effects against their controls. The authors concluded that acupuncture may have a positive effect on pain, function and quality of life, but that the trials were too varied and too flawed to support a guideline. Lu and colleagues, in Taipei, pooled only trials that added acupuncture to drug treatment and found greater reductions in CRP, ESR, pain (1.1 points on a 10-point scale), DAS28 (0.6 points) and swollen and tender joints when acupuncture, electroacupuncture or moxibustion was added to standard drugs. Li and colleagues, in Beijing, found pain lower by 1.0 point across all controls, but in the subgroup against sham the difference of 0.94 points was not statistically significant, and nor was the difference in ESR. Wan and colleagues, in a network meta-analysis of the same kind of add-on trials, which compares several treatments at once through shared comparators, ranked electroacupuncture with DMARDs first for DAS28, and found no kind of acupuncture added to DMARDs better than DMARDs alone for morning stiffness.

Wang and colleagues' 2026 network meta-analysis separated two kinds of sham. It found pain lower with electroacupuncture and with manual acupuncture than with standard medication or with sham needling away from acupuncture points. The authors argued that sham needling at the same points as the real treatment may understate its effect and should not be used as a placebo control, and that needling away from the points is the more valid control.

Trials

The sham-controlled trials in RA are small. The 1999 trial by David and colleagues randomised 64 people already taking second-line drugs, of whom 56 were suitable for analysis, to needling at a single point, LV03 (Taichong), or placebo, crossing each person over to the other treatment after a 6-week break. It found no significant effect on any outcome, including pain, joint counts and inflammatory markers, and no adverse effects. The authors concluded that acupuncture of this type could not be considered a useful addition to treatment. A 2007 pilot trial by Tam and colleagues gave 36 people 20 sessions over 10 weeks of electroacupuncture, traditional acupuncture or sham, 12 to each. Pain did not change in any group at week 10; tender joints fell in the two active groups. By week 20, 7 of the 12 in the sham group and all 12 in the electroacupuncture group had finished, and all but one of those who left gave lack of effect as the reason.

A 2019 trial led by the first author of the 2019 review chose points by a Chinese medicine diagnosis and randomised 105 people with RA of the hand to acupuncture at those points, the same needling at points off the meridians, or a waiting list, 35 to each. People were assessed five minutes after needling and followed for 4 weeks. The abstract reports changes within each group: pain, grip strength, quality of life and joint counts improved with real acupuncture, pain alone improved with the control needling, and the waiting-list group got worse. A 2024 trial by Luo and colleagues randomised 132 people with RA and a Chinese medicine diagnosis of liver and kidney deficiency to intradermal acupuncture, in which short needles are left under the skin, or a sham version. DAS28 improved by 0.11 points more and the Health Assessment Questionnaire, a measure of daily function, by more with intradermal acupuncture, while CRP did not differ between groups. No adverse effects were reported.

What guidelines say

The American College of Rheumatology's 2022 guideline on exercise, rehabilitation, diet and additional integrative interventions, by England and colleagues and published in 2023, was the College's first on these treatments, to be used alongside DMARDs. It conditionally recommends acupuncture over no acupuncture, on low-certainty evidence of inconsistent improvements in pain and function. Its panel of people with RA rated acupuncture of lower value than the other interventions it considered, and the guideline names burden, cost, access and invasiveness as factors in the choice. The same guideline conditionally recommends against electrotherapy, such as stimulation of nerves through the skin, and against chiropractic.

The European Alliance of Associations for Rheumatology's (EULAR) 2025 update of its RA management recommendations, published in 2026, covers DMARDs and glucocorticoids only, so it does not weigh acupuncture. NICE's guideline NG100, published in July 2018 and last updated in October 2020, does not name acupuncture. It advises telling adults with RA who want to try complementary therapies that some may give short-term relief of symptoms but there is little or no evidence of long-term benefit, and that such therapies should not replace conventional treatment; those recommendations last had an evidence review in 2009. Clinical guidelines compares other bodies' positions.

What the reviews could not settle

The Cochrane review has not been updated since 2005, and the later reviews report no GRADE rating, so the only graded judgement of the evidence is the ACR guideline's, at low certainty. The largest pooled results compare acupuncture added to drugs with drugs alone, which cannot separate the needling from the attention and expectation that come with it, and most of the trials in Seca's review were published in China; Where the trials come from discusses what the country of a trial tells a reader. The sham-controlled trials enrolled tens of people each and disagree with one another, and the reviews disagree on which sham is a fair control. Follow-up ran from minutes to months, and the outcomes pooled were pain, disease-activity scores and blood markers, over periods too short to show whether acupuncture changes the course of the disease. Knee osteoarthritis covers the trials in osteoarthritis of the knee, and Chronic pain the individual-patient-data meta-analyses across painful conditions.

AcuiQ's rheumatoid arthritis page lists the protocols individual studies prescribed, each with its citation.