Carpal tunnel syndrome is compression of the median nerve at the wrist, which causes pain and tingling in the hand. A 2017 trial in Brain calls it the most common entrapment neuropathy, a nerve disorder caused by compression. The acupuncture trials for it compare needling, electroacupuncture or laser acupuncture with sham procedures, with drugs and steroid injections, or with a wrist splint worn at night, and the reviews below keep those comparisons apart because they answer different questions.
The Cochrane review
A 2018 Cochrane review by Choi and colleagues searched English-language databases to November 2017 and Korean and Chinese databases to April 2018. It included 12 randomised or quasi-randomised trials with 869 adults, covering needle acupuncture, electroacupuncture and laser acupuncture, in which a low-power laser beam stimulates points without a needle. All had an unclear or high overall risk of bias, and most could not be pooled because they differed too much. GRADE, the scale Cochrane uses to say how much confidence a result supports, rated each comparison low or very low.
| Comparison | Trials (people) | Result | Certainty |
|---|---|---|---|
| Acupuncture vs sham | 1 (41) | No clear difference in symptom severity or function three months after treatment | Low |
| Laser acupuncture vs sham laser | 1 (60) | Global symptom score 7.5 points better on a 0 to 50 scale four weeks after treatment; more responders (risk ratio 1.59) | Low |
| Acupuncture vs oral corticosteroids | 1 (77) | No clear difference at two or four weeks; global symptom score 8.3 points better at 13 months | Very low |
| Acupuncture vs vitamin B12 | 2 (100) | Responder rate risk ratio 1.16, confidence interval crossing no difference | Very low |
| Acupuncture vs ibuprofen, all wearing night splints | 1 (50) | Lower symptom severity score one month after treatment | Very low |
| Electroacupuncture vs night splints | 1 (60) | No clear difference in symptom severity | Very low |
| Electroacupuncture plus night splints vs splints alone | 1 (181) | No difference in symptom severity at 17 weeks | Low |
A second sham-controlled acupuncture trial, of 111 people, reported no data the review could use. Seven of the 12 trials reported adverse events: bruising with electroacupuncture and local pain at the needle site, and none serious. The authors concluded that acupuncture and laser acupuncture may have little or no short-term effect on symptoms compared with sham, and that it is uncertain whether they work better or worse than steroid nerve blocks, oral steroids, vitamin B12, ibuprofen or splints. Laser acupuncture describes the method, and Sham acupuncture the kinds of control in use.
Other systematic reviews
The first review, a 2011 analysis by Sim, Ernst and colleagues, found six randomised trials of generally low quality, two of them sham-controlled. Pooling two trials with 144 people, it found more responders with acupuncture than with steroid block therapy (risk ratio 1.28), and called the evidence encouraging but not convincing.
A 2020 review by Wu and colleagues searched to July 2019 and included 10 trials with 728 people, most at high risk of bias for blinding. It reported symptom severity scores lower with manual acupuncture than with ibuprofen or prednisolone, and lower with electroacupuncture plus a splint than with a splint alone, by 0.20 points with a confidence interval that just excluded no difference. The Cochrane review reported the splint comparison as no difference.
Two reviews published in 2022 and 2023 drew more heavily on trials from China. Li and colleagues’ 2022 review of electroacupuncture searched to May 2022 and included 26 trials with 1,698 people, 19 of them published in Chinese. Against what it calls routine treatment, which in its trial table includes neurotrophic drugs, acupuncture without current and herbal washes, it reported lower pain scores (0.79 points on a visual analogue scale) and better nerve conduction measures. It did not grade certainty, and asked for multicentre, double-blind trials. Electroacupuncture describes the method.
Dong and colleagues’ 2023 review searched English and Chinese databases to October 2022 and included 16 trials with 1,025 people. It rated one trial at low risk of bias, 14 with some concerns and one at high risk, and found only two that compared acupuncture with sham. Against night splints, acupuncture alone eased pain more but made no difference to symptom severity or function; against drugs, acupuncture alone showed no advantage. Added to another treatment, acupuncture improved symptoms, function, pain and nerve conduction. Every result was low or very low certainty. All 16 trials needled PC07 (Daling), and 12 needled PC06 (Neiguan).
A 2024 overview by Liu and colleagues assessed nine systematic reviews, searched to April 2024. On AMSTAR 2, a checklist for the conduct of a review, it rated eight critically low and one low. Of the outcomes it graded, 6 were low certainty and 47 very low. Read together, the reviews consistently reported lower pain intensity with acupuncture and were inconsistent on responder rate, symptom severity, function and nerve conduction; none found a serious adverse event.
The brain-imaging trial
The 2017 trial by Maeda, Napadow and colleagues at Massachusetts General Hospital enrolled 80 people, 65 of them women, and randomised 79 into three electroacupuncture arms: 28 local, 28 distal and 23 sham, each for 16 sessions over eight weeks. The local arm had needles at TB05 (Waiguan) and PC07 on the more affected forearm, plus three points chosen from HT3, PC3, SI4, LI5, LI10 and LU5. The distal arm had electroacupuncture at SP06 (Sanyinjiao) and LV04 (Zhongfeng) near the opposite ankle, with manual needling at GB34, KD3 and SP5. Both ran at 2 Hz for 20 minutes. The sham used non-penetrating Streitberger placebo needles at non-points on the forearm, with electrodes attached but not plugged into an active output. The paper writes the triple burner points with the prefix TW, which AcuiQ writes as TB; Reading a point code explains the difference.
Symptom severity on the Boston Carpal Tunnel Questionnaire fell in all three arms, and immediately after treatment the real and sham groups did not differ. At three months, completed by 56 people, symptom severity had fallen 25% with real electroacupuncture and 11% with sham, a significant difference. Median nerve sensory latency, the time a signal takes to cross the wrist, shortened by 0.16 ms with real electroacupuncture and did not change with sham. On functional MRI, the distance between the brain’s maps of the second and third fingers increased more after real treatment than after sham, and the people whose maps moved apart most kept more of their symptom improvement at three months. The authors propose that local and distal needling change median nerve function through different regions of the somatosensory cortex.
What guidelines say
The American Academy of Orthopaedic Surgeons (AAOS) adopted its clinical practice guideline on the management of carpal tunnel syndrome on 18 May 2024. Its announcement says it replaces the 2016 edition and concentrates on long-term outcomes. No recommendation names acupuncture, although several acupuncture trials are among its included literature. It names acupressure among non-operative treatments that did not show superiority over control or placebo, and states that non-operative treatments show no significant difference from one another in patient-reported outcomes, an option it rates limited after downgrading for heterogeneity and the lack of long-term follow-up. With strong evidence, it says corticosteroid injection does not provide long-term improvement.
In England, the NICE Clinical Knowledge Summary for primary care, last revised in August 2026, advises offering people with mild or moderate symptoms a six-week trial of a night wrist splint in a neutral position or a single corticosteroid injection, and referral when symptoms persist, progress or the diagnosis is unclear. Its management advice does not mention acupuncture. Clinical guidelines compares how guideline bodies treat acupuncture across conditions.
What the reviews could not settle
The Cochrane review found one usable sham-controlled needle trial, of 41 people, and Dong and colleagues found two sham comparisons among 16 trials, so the reviews could not separate an effect of needling from expectation. The comparisons with steroids, drugs and splints were open-label and mostly single trials, and the largest pooled review of electroacupuncture did not grade its certainty. The Maeda trial measured symptoms in fewer than 60 people at three months and found its difference from sham there, not at the end of treatment. Its nerve conduction and brain-map findings come from that one trial.
AcuiQ’s carpal tunnel syndrome page lists the protocols individual studies prescribed, each with its citation.