Stroke recovery

Acupuncture after stroke has been tested at three stages: in the first month, in rehabilitation over the following months, and for single problems such as difficulty swallowing or speaking. The Cochrane reviews of the first month and of rehabilitation find the same pattern. Trials that compared acupuncture with no acupuncture, where everyone knew who was being needled, report benefits; the smaller number that compared it with sham acupuncture mostly do not confirm them. The 2018 Cochrane review of acute stroke by Mangmang Xu and colleagues at West China Hospital concludes that the apparent improvements are confounded by the risk of bias that comes with open controls.

The first month

The 2018 review included 33 randomised trials with 3,946 people whose acupuncture started within 30 days of an ischaemic or haemorrhagic stroke. Every trial needled the skin. Its primary outcome was death or dependency, meaning needing help with daily life, at the end of follow-up. Odds ratios below 1 favour acupuncture; quality is the GRADE rating.

Outcome data came from only six sham-controlled trials, with 668 people. Death itself did not differ, either within two weeks or by the end of follow-up (odds ratio 1.08; 22 trials, 2,865 people). Adverse events such as pain, dizziness and fainting occurred in 6.2% of people given acupuncture (64 of 1,037), and 1.4% stopped treatment because of them.

A 2001 Swedish trial by Johansson and colleagues in 7 hospitals used electrical stimulation too weak to feel as its control. It randomised 150 people with moderate or severe impairment, 5 to 10 days after stroke, to acupuncture including electroacupuncture, to strong transcutaneous electrical nerve stimulation that made the muscles contract, or to the control, for 20 sessions over 10 weeks. At three months and one year the three groups did not differ in motor function, daily activities, walking, social activity or life satisfaction.

Rehabilitation

A 2016 Cochrane review by Ai Yang and colleagues covered acupuncture in the subacute and chronic stages, after the first weeks. It included 31 trials with 2,257 people and rated the evidence for every main outcome low or very low. Two trials compared real with sham acupuncture, both on top of standard treatment, and found no difference in motor function or quality of life. The other 29 compared acupuncture plus standard treatment with standard treatment alone, and reported:

• better independence in daily activities on the Barthel Index (mean difference 9.19 points; 9 trials, 616 people; very low quality);
• better global neurological function (odds ratio 3.89; 7 trials, 543 people; low quality);
• better motor function on the Fugl-Meyer Assessment (mean difference 6.16; 4 trials, 245 people; low quality);
• gains in cognition, depression scores, swallowing and pain, each from 2 to 6 trials at low or very low quality.

No trial reported death or the need for institutional care. The authors concluded that the evidence was inadequate to draw conclusions about routine use and called for large multi-centre trials.

A 2010 review by Kong, Ernst and colleagues took only sham-controlled trials and found 10. In the acute and subacute stages, the three trials at low risk of bias showed no effect on daily activities at the end of treatment (standardised mean difference 0.07; 244 people) or after follow-up (0.10). In the chronic stage, three trials of spasticity, measured on the Modified Ashworth Scale, all found no benefit.

Swallowing and speech

Difficulty swallowing after stroke, called dysphagia, raises the risk of death. The 2026 Cochrane review of swallowing therapies by Wilkinson, Bath and colleagues covered 181 trials with 11,500 people within six months of stroke, each therapy compared with usual care rather than sham. For acupuncture it found:

• swallowing impairment may be reduced, but the effect is uncertain (standardised mean difference −1.20; 37 trials, 3,335 people; very low certainty);
• dysphagia at the end of the trial is probably reduced (odds ratio 0.40; 36 trials, 3,647 people; moderate certainty);
• food or liquid entering the airway, scored on the penetration-aspiration scale, is probably reduced (mean difference −0.90; 8 trials, 496 people; moderate certainty);
• chest infection or pneumonia may be reduced (odds ratio 0.47; 4 trials, 426 people; low certainty).

The authors wrote that small trials at high or unclear risk of bias, publication bias and inconsistent results limit every conclusion in the review, across all the therapies it covers.

For speech, a 2024 sham-controlled trial by Li and colleagues in 3 Chinese hospitals randomised 252 people with motor aphasia, difficulty producing speech, to 6 weeks of manual or sham acupuncture, both alongside language training. The acupuncture group scored higher on the Western Aphasia Battery’s aphasia quotient (by 7.99 points at 6 weeks and 10.34 at 6 months) and on a Chinese functional communication scale. Scalp acupuncture describes the standard scalp lines and the schools that use them.

What guidelines say

A 2025 review of stroke guidelines by Zhu and colleagues found 19 rehabilitation guidelines published from 1990 to April 2024 that make recommendations on traditional and complementary medicine: 12 from China, 2 each from the United States, Brazil and Australia, and 1 from Canada. Rated with AGREE II, a tool for judging how well a guideline was developed, 9 were recommended for use. The guidelines commonly recommended acupuncture for spasticity and for swallowing problems after stroke, recommended against adding it to improve daily activities, and disagreed on upper-limb motor function. The review found the recommendations typically conditional. Clinical guidelines sets out other bodies’ positions.

What the reviews could not settle

Whether acupuncture changes whether a person lives independently after stroke is untested in trials large enough and blinded enough to answer it: the sham-controlled trials in the 2018 review numbered six, and their pooled estimates are compatible with benefit and with none. The swallowing estimates rated at moderate certainty compare acupuncture with usual care and cannot separate needling from extra attention. The 2018 review downgraded its evidence partly because the acupuncture given and the outcomes measured varied between trials, so the reviews cannot say which approach any benefit belongs to. Where the trials come from discusses where acupuncture trials are run and published. Trials in dementia and mild cognitive impairment, which include vascular dementia after stroke, are covered in Cognitive impairment and dementia. Trials after a brain injury from an outside force, such as a fall or a collision, are in Traumatic brain injury.

AcuiQ’s stroke, ischaemic stroke, stroke sequelae, post-stroke dysphagia and post-stroke aphasia pages list the protocols individual studies prescribed, each with its citation.