The 2013 guideline of the American Academy of Otolaryngology–Head and Neck Surgery Foundation (AAO-HNS) describes Bell’s palsy as a rapid weakness or paralysis of one side of the face, of unknown cause and typically self-limited, and makes a strong recommendation that people aged 16 and over be given oral steroids within 72 hours of the first symptoms. The acupuncture trials the reviews found compare acupuncture with drugs, other therapies, waiting or another way of needling; the Cochrane review found none against sham. Each review describes the trials as poor in quality, and the question they leave open is whether acupuncture changes how many people recover fully.
The Cochrane review
The 2010 Cochrane review by Chen and colleagues in Chengdu, the latest version of a review first published in 2004, searched English and Chinese sources to May 2010 and found 49 potentially relevant articles. Six randomised trials with 537 people met its criteria. Four compared acupuncture with drugs and two compared it with manipulation or physical therapy; none used sham acupuncture, so neither patients nor practitioners were blinded.
The review’s primary outcome was the number of people left with incomplete recovery, meaning lasting facial weakness that is cosmetically disabling, six months after onset. None of the six trials reported it, or any other outcome the review had specified. No trial reported harmful side effects. The authors concluded that the trials’ quality, from unclear randomisation, allocation concealment and blinding, and the clinical differences between them, were inadequate to allow any conclusion about efficacy.
Later meta-analyses
| Review | Trials | Finding | Authors’ caution |
|---|---|---|---|
| Li and colleagues 2015 | 14 trials, 1,541 people, acupuncture against other therapies | Higher “effective response rate” with acupuncture (risk ratio 1.14, 95% CI 1.04 to 1.25) | High risk of bias, very high heterogeneity (I² 87%), no usable data on complications |
| Zhang and colleagues 2019 | 11 trials, 1,258 people, acupuncture against drugs | Higher “cure rate” (risk ratio 1.77, 95% CI 1.41 to 2.21) and “total effective rate” (1.18) with acupuncture | Poor quality, substantial heterogeneity (I² 67% and 90%), not enough evidence on safety |
Heterogeneity, measured as I², is the share of the variation between trial results that chance does not explain. “Cure rate” and “effective rate” are categories each trial defines for itself, not a standard scale of facial function.
Two trials
A 2013 multicentre trial in CMAJ by Xu and colleagues in Wuhan asked a different question: whether needle stimulation strong enough to produce de qi, the aching, heavy or spreading sensation acupuncturists seek, matters. It randomised 338 people with Bell’s palsy, all of whom took prednisone. Both groups were needled at the same points: GB14 (Yangbai), ST04 (Dicang), ST06 (Jiache), ST07 (Xiaguan) and TB17 (Yifeng) on the affected side, and LI04 (Hegu) on the other. In one group (167 people) the needles were manipulated until de qi was reached; in the other (171) they were inserted and left. At six months the de qi group had better facial nerve function (adjusted odds ratio 4.16, 95% CI 2.23 to 7.78), less disability and better quality of life. The trial had no sham or no-acupuncture group, so it compares two ways of needling rather than acupuncture with none. An accompanying CMAJ editorial by John Fletcher noted that because everyone received real acupuncture and the outcome was assessed blind, expectations are less likely to explain the difference. Needling and de qi describes the sensation.
A smaller 2015 trial by Kwon and colleagues at Kyung Hee University in Seoul studied people left with lasting weakness, the sequelae of Bell’s palsy. It randomised 39 people, 26 to 8 weeks of acupuncture and 13 to a waiting list. After 8 weeks the acupuncture group scored better on the Facial Disability Index social and physical subscales, on the Sunnybrook facial grading system and on a stiffness scale. No severe adverse event occurred. With a waiting list as control, the trial cannot separate the needling from the attention that came with it.
What guidelines say
The AAO-HNS 2013 guideline states that no recommendation can be made on acupuncture for Bell’s palsy, based on poor-quality trials and an unknown balance of benefit and harm. Its profile of the statement rates the aggregate evidence as grade B and the confidence in it as low because of significant methodological flaws. It lists the possible benefit as a small improvement in facial nerve function and pain, and the harms as cost, time, side effects and delay in starting steroids. The panel recorded a major difference of opinion: it was divided over whether to recommend against acupuncture or to make no recommendation, and settled on no recommendation, with a large role for patient preference.
The Japan Society of Facial Nerve Research’s 2023 guideline update, summarised in English in 2024, used the GRADE approach on nine clinical questions, acupuncture among them. Its abstract states that it strongly recommends standard-dose corticosteroids and that the other treatments are weakly recommended because the evidence is insufficient. This page reports the abstract, which does not state the direction of the acupuncture recommendation separately. Clinical guidelines sets out other bodies’ positions.
What the reviews could not settle
The Cochrane review found no trial that measured incomplete recovery at six months, the outcome it set, and none with a sham control. Because the AAO-HNS guideline describes Bell’s palsy as typically self-limited, a trial needs a comparison group to show that acupuncture changed recovery, and the trials so far compared it with drugs, with other needling or with waiting. The later meta-analyses pooled trials whose results disagree widely. The Cochrane review has not been updated since 2010.
AcuiQ’s Bell’s palsy, facial paralysis, peripheral facial paralysis and Bell’s palsy sequelae pages list the protocols individual studies prescribed, each with its citation.