Neck pain

The Cochrane review of acupuncture for neck pain is not currently available. Trinh and colleagues published an update in May 2016 that pooled 27 trials and found acupuncture relieved mechanical neck pain better than sham at the end of treatment and in the short term, and Cochrane withdrew it from the Library in November 2016 while comments on it were addressed. It remained withdrawn when this page was checked. The reviews published since agree with it on the short term and disagree on whether any advantage over sham lasts, and a Canadian guideline for recent neck pain advises against electroacupuncture.

Mechanical or non-specific neck pain means pain in the neck without an identified cause such as a fracture or a trapped nerve. Sham acupuncture explains the controls that mimic treatment, which several results below turn on.

The withdrawn Cochrane review

The 2016 update searched to August 2015 and replaced the 2006 version, which had covered 10 trials of chronic neck pain. Its 27 trials fell into six groups: whiplash-associated disorders (3 trials, 205 participants), chronic myofascial neck pain (5 trials, 186), chronic pain from arthritic changes (5 trials, 542), chronic non-specific neck pain (6 trials, 4,011), neck pain with signs of nerve root involvement (2 trials, 43) and subacute or chronic mechanical neck pain (6 trials, 5,111).

For mechanical neck pain the reviewers judged the evidence moderate in quality. Acupuncture relieved pain better than sham at the end of treatment and at short-term follow-up, and better than inactive treatment in the short term; people on a waiting list reported more pain and disability in the short term than people who had acupuncture. The short-term trials against sham were consistent enough to pool, giving a standardised mean difference (SMD) of −0.23 in favour of acupuncture, where SMD is the measure reviews use to combine trials that scored pain on different scales. The reviewers found that the effect did not seem to last into the long term, and none of their trials tested whether repeated courses would help. Reported adverse effects were minor: more pain, bruising, fainting, worse symptoms, local swelling and dizziness.

Reviews since 2016

Fang and colleagues searched English and Chinese literature to March 2024 and read their own result two ways: either acupuncture has no lasting effect on neck pain beyond that of sham, or the sham procedures themselves had a therapeutic effect. Xie and colleagues applied trial sequential analysis, a test of whether enough participants have accumulated for a pooled result to be trusted, and reported that their main results crossed its threshold. Because their “inert” group mixes sham with no treatment, their estimate cannot say how much of the difference is specific to needling.

The individual patient data meta-analyses by the Acupuncture Trialists’ Collaboration pooled neck pain with back pain as non-specific musculoskeletal pain. Their 2012 analysis named a neck pain trial by Vas and colleagues as an outlier, with an effect against sham about five times the pooled estimate. Their 2018 update reported a large effect against sham for the neck pain trials when they were analysed apart from back pain. Chronic pain covers both analyses.

The large trials

Two trials with several hundred or more participants each compared acupuncture with usual care, without a sham group.

• The ARC neck pain trial by Claudia Witt and colleagues in Germany, published in 2006, randomised 3,766 people with neck pain lasting more than six months: 1,880 to up to 15 acupuncture sessions over three months alongside routine care and 1,886 to routine care alone. A further 10,395 who declined randomisation had acupuncture and were followed separately. On the Neck Pain and Disability scale, scores fell by 16.2 points with acupuncture and 3.9 with routine care, and the improvement largely held at six months. A companion economic analysis put the cost at 12,469 euros for each quality-adjusted life year gained, which its authors judged cost-effective by international thresholds.
• The ATLAS trial by MacPherson and colleagues in UK primary care, published in 2015, randomised 517 people whose neck pain had lasted a median of six years to 12 acupuncture sessions, 20 Alexander Technique lessons, or usual care alone. At 12 months, scores on the Northwick Park questionnaire were 3.92 percentage points lower with acupuncture than with usual care and 3.79 lower with Alexander lessons; from baseline, scores fell by 32% and 31%. No serious adverse event was judged probably or definitely caused by either intervention.

What the guidelines say

The Ontario Protocol for Traffic Injury Management (OPTIMa) Collaboration published a guideline for neck pain and associated disorders in 2016, covering adults whose pain had lasted less than six months. For the two milder of its three grades, both under and over three months’ duration, it advises that clinicians should not offer electroacupuncture, among other treatments, “in view of evidence of no effectiveness”. Grade III, the grade with neurological signs, receives no recommendation on acupuncture. One of the collaboration’s own systematic reviews, by Wong and colleagues, covering studies from 2000 to 2014, found that for persistent neck pain, Western acupuncture gave similar outcomes to non-penetrating placebo electroacupuncture and needle acupuncture gave similar outcomes to a penetrating sham, and that for whiplash, electroacupuncture gave similar outcomes to a simulated version.

NICE has no guideline on neck pain. Its Clinical Knowledge Summary on non-specific neck pain, last revised in April 2023, does not mention acupuncture. NICE’s 2021 guideline on chronic primary pain, which covers pain without a condition that adequately accounts for it, recommends that a single course of acupuncture be considered, and its rationale cites two economic evaluations, one from the UK, that found acupuncture good value for chronic neck pain, while noting that the acupuncture costs they assumed seemed low. A 2025 review of 17 guidelines on acupuncture for chronic musculoskeletal pain, published between 2014 and 2024, found neck pain the condition they addressed least, with recommendations on it pointing in both directions. Clinical guidelines sets these positions beside other bodies’.

What the reviews could not settle

The sources agree that people with chronic neck pain who have acupuncture report less pain in the weeks after treatment than people who do not. They disagree on how much of that remains once sham is the comparison, and on how long any difference lasts: the withdrawn Cochrane review and Fang’s review both found the advantage over sham did not hold into the longer term, while ATLAS and the ARC trial, both against usual care, found differences at 12 and 6 months. The heterogeneity in Xie’s pooled estimate, and the outlier the 2012 individual patient data analysis identified, mean single trials can move the pooled figures. The Cochrane review’s withdrawal leaves the field without a current Cochrane assessment, and in that review neck pain with signs of nerve root involvement rested on two trials with 43 participants.

AcuiQ’s neck pain page lists the protocols individual studies prescribed, each with its citation, and the chronic neck pain, cervical spondylosis and cervical radiculopathy pages do the same for the terms those studies used.