Autism spectrum disorder (ASD) is a developmental condition marked by difficulties with social interaction and communication and by a lack of flexibility in thought and behaviour, as the 2011 Cochrane review of acupuncture for it describes. That review, by Daniel Cheuk, Virginia Wong and Wen Xiong Chen at the University of Hong Kong, found 10 trials in 390 children and concluded that the evidence did not support acupuncture for ASD; its two sham-controlled trials showed no difference in core autistic features. Reviews published since 2018 pool dozens of trials that mostly add acupuncture to rehabilitation and compare it with rehabilitation alone; they report lower autism rating scores with acupuncture, and those that grade the evidence rate its certainty low or very low. Cochrane found no trial in adults, and several later reviews admitted only children.
The Cochrane review
The review searched to September 2010 and admitted randomised and quasi-randomised trials that compared acupuncture with no treatment, placebo or sham; it excluded trials comparing acupuncture with another treatment. Its 10 trials, in Hong Kong, mainland China and Egypt, enrolled children aged 3 to 18 and treated them for four weeks to nine months. Published in September 2011 and since edited with no change to its conclusions, it is the current version: a PubMed search on 21 October 2026 found no update and no other Cochrane review of acupuncture for autism.
- Needle acupuncture against sham, 2 trials. Pooled, they found no difference in core autistic features on the Ritvo-Freeman Real Life Rating Scale (mean difference 0.09, 95% CI -0.03 to 0.21). Some secondary outcomes, among them communication, cognition and global functioning, pointed towards acupuncture.
- Acupuncture added to conventional treatment, against conventional treatment alone, 6 trials. The trials used different primary measures and most showed no improvement in core features; one reported more children improving on the Autism Behavior Checklist (risk ratio 1.53, 95% CI 1.09 to 2.16).
- Acupressure added to conventional treatment, 2 trials. Neither reported core features; single results suggested some benefit on communication, cognition and global functioning.
Four trials reported adverse effects, among them bleeding, crying from fear or pain, irritability, poor sleep and more hyperactivity, sometimes in the control group too. No trial measured quality of life. The authors judged six trials at high risk of bias, and listed the other problems: few trials, short treatment and follow-up, inconsistent and imprecise results, and so many analyses that some positive findings were likely to be false. Virginia Wong, one of the authors, led both of the sham-controlled trials, described below, and the review declares her and Chen's part in a funded acupuncture project for autistic children in Hong Kong.
Later systematic reviews
None of the reviews below is a Cochrane review, and none of their trials is restricted to a sham comparison; most compare acupuncture plus behavioural, educational or rehabilitation training with that training alone.
| Review | Methods | Trials and people | What the authors report | Certainty |
|---|---|---|---|---|
| Lee 2018, searched to December 2016 | Any acupuncture in children | 27 trials, 1,736 children | Added to behavioural and educational intervention, CARS lower by 8.1 points and ABC by 8.9; unclear which symptoms improved; not conclusive because treatment methods varied | Not stated in the abstract |
| Liu 2019, searched to September 2018 | Scalp acupuncture against behavioural and educational interventions | 14 trials, 968 children; 11 pooled | Lower CARS and ABC scores, better PEP-3 communication, physical and behaviour scores | Not stated in the abstract |
| Wang 2021 | Acupuncture, alone or with massage | 16 trials, 1,332 people | Lower CARS and ABC scores; expression age improved but comprehension age did not differ | Not stated in the abstract |
| Lun 2023, searched 2012 to September 2022 | Acupuncture against no acupuncture, in children | 38 trials, 2,862 children | Higher effective rate (risk ratio 1.33) and lower ABC, CARS and ATEC scores; 37 trials at high risk of bias | Very low for most outcomes |
| Yu 2023 | Acupuncture and behavioural therapies, each against its control, for social function | 30 acupuncture trials, 36 behavioural | Body acupuncture SMD 0.76; cognitive behavioural therapy SMD 0.42 | Low for acupuncture; high for cognitive behavioural therapy |
| Zhao 2025 | Scalp acupuncture added to rehabilitation training, against training alone | 33 trials, 2,701 people | ABC lower by 5.7 points and CARS by 4.0; better Gesell and PEP-3 scores; conclusion needs rigorous trials with long follow-up | Not stated in the abstract |
| Dong 2026, searched to November 2024 | Laser acupuncture in children | 2 trials, 76 children; not pooled | Benefit against no treatment; no adverse effects reported by parents | Low to very low |
| Ma 2026, network meta-analysis, searched to October 2024 | 21 non-drug and newer drug treatments for autism, compared through a network | 106 trials, 5,986 people | Against placebo, acupuncture and moxibustion SMD -0.87 on core symptoms, beside exercise (-1.05) and bumetanide (-0.57) | Not stated in the abstract |
CARS is the Childhood Autism Rating Scale, ABC the Autism Behavior Checklist, ATEC the Autism Treatment Evaluation Checklist and PEP-3 the third edition of the Psychoeducational Profile; on the first three a lower score means fewer autistic features. A standardised mean difference (SMD) expresses the gap between groups in units of the spread of scores. The Lee review, by Boram Lee and colleagues at Kyung Hee University in Seoul, also reported that acupuncture on its own lowered CARS scores; the journal published a corrigendum to it in 2023. Yu and colleagues, at Guangzhou University of Chinese Medicine, rated their evidence for acupuncture as low or very low and that for three behavioural therapies as moderate or high. Dong and colleagues, at the University of Alberta, looked at laser acupuncture because needles may cause children distress.
Two reviews have examined the reviews and trials themselves. A 2023 overview by Meng and colleagues at the China Academy of Chinese Medical Sciences appraised nine systematic reviews of acupuncture and moxibustion for childhood autism: their reporting quality was low, common methodological problems were the lack of a prespecified protocol and incomplete searches, and the overall quality of evidence under GRADE was low, mainly for risk of bias, then inconsistency, imprecision and publication bias. A 2025 literature review by Long and colleagues of 31 trials published from 2015 to 2024 found that they mostly used indirect controls rather than a sham, needled points on the head and treated for 12 weeks, and that they chose among 37 different outcome measures, CARS the most often. Its authors call for a core outcome set for acupuncture trials in ASD. Where the trials come from covers where acupuncture trials are run and how results differ by language, and Evidence maps covers six overviews of acupuncture's systematic reviews.
Trials with a sham or placebo arm
- Tongue acupuncture against sham, 2010. A single-blind trial by Virginia Wong and Sun at the University of Hong Kong randomised 50 children to 40 sessions of tongue acupuncture over 8 weeks or to sham needling at non-acupoints on the tongue. Both groups improved on every measure, more so with true acupuncture, but the only statistically significant differences were in the self-care and cognition domains of the Functional Independence Measure for children (WeeFIM).
- Electroacupuncture against sham, 2010. A double-blind trial by Wong and Chen randomised 55 children, 30 to electroacupuncture and 25 to sham electroacupuncture at sham points, for 12 sessions over 4 weeks. The electroacupuncture group did better on the language comprehension domain of WeeFIM, the self-care caregiver assistance domain of the Pediatric Evaluation of Disability Inventory and the Clinical Global Impression-Improvement scale, and parents reported better social initiation, receptive language and attention. More than 70% of children adapted easily to needling and 8% complied poorly; side effects were minor bleeding and irritability. Electroacupuncture describes the method.
- Laser acupuncture against placebo laser, 2020. A double-blind trial by Surapaty and colleagues at the University of Indonesia in Jakarta gave 46 children sensory-occupational integrative therapy with either real or placebo laser acupuncture. The real laser group scored better on WeeFIM speech perception, expression and social interaction; parental reports did not differ. Laser acupuncture describes the method.
The two Hong Kong trials are the sham-controlled trials the Cochrane review pooled. Trials since have mostly compared acupuncture with rehabilitation alone. A 2026 trial by Zhang and colleagues at Linyi People's Hospital in China, for example, enrolled 120 children and gave all of them daily behavioural and educational rehabilitation for 12 weeks, with scalp and tongue acupuncture added in one arm; children and guardians knew their group, and no sham was used. Acupuncture was followed by a larger fall in CARS scores and better PEP-3 scores, and no side effects were seen. A sham-controlled trial at Sangji University in South Korea, recruiting until December 2026, will compare autistic and non-autistic people after real or superficial sham needling, with measures of brain activity, not symptoms, as its main outcomes. Sham acupuncture sets out what a sham can and cannot rule out, and Acupuncture for children covers safety in children.
What guidelines say
NICE's guideline on autism spectrum disorder in under 19s: support and management (CG170), published in August 2013 and last updated on 14 June 2021, does not mention acupuncture. Its list of interventions not to use names secretin, chelation and hyperbaric oxygen therapy in any context, and neurofeedback and auditory integration training for speech and language problems. The adult guideline, autism spectrum disorder in adults: diagnosis and management (CG142), published in June 2012 and last updated on 14 June 2021, does not mention it either. Both remain current. Clinical guidelines lists the guidelines the evidence pages cite.
What the reviews could not settle
Whether acupuncture differs from sham in autistic children remains open. The Cochrane review pooled two sham-controlled trials of 105 children, from one research group, and found no difference in core features; the later reviews draw almost entirely on trials that compare acupuncture plus rehabilitation with rehabilitation alone, where no sham is used, so families know which group a child is in, and the outcomes are mostly rating scales such as CARS. The trials differ in method (body, scalp, tongue, electro- and laser acupuncture), in points and in length of treatment, which Lee 2018, Wang 2021 and Zhao 2025 each give as a reason their conclusions are uncertain. Follow-up is short, and Cochrane found no trial in adults. Cochrane has published no update of its review since 2011.
AcuiQ's autism spectrum disorder and autism pages list the protocols individual studies prescribed, each with its citation.