The most recent Cochrane review of acupuncture for chronic low back pain, published in December 2020 by Mu and colleagues, pooled 33 randomised trials with 8,270 participants. Against no treatment, acupuncture gave pain relief large enough for the reviewers to call it clinically important. Against sham acupuncture, a procedure built to look and feel like treatment without being it, the difference fell below the threshold they had set for one that patients would notice. Every source on this page reports some version of that split between the two comparisons, and the two guidelines covered below reach opposite recommendations.
Chronic means pain lasting more than three months. Non-specific means pain without an identifiable cause, such as a fracture, an infection or a compressed nerve root. Sham acupuncture sets out why no sham control is inert, a point that bears on every comparison with sham reported here.
The 2020 Cochrane review
Mu and colleagues searched to August 2019 and compared acupuncture with three controls: sham, no treatment and usual care. They rated the certainty of each finding with GRADE, the standard scale that runs from high through moderate and low to very low, where low means the true effect may differ substantially from the estimate. They took 15 points on a 0 to 100 pain scale as the smallest difference that matters to a patient. A standardised mean difference (SMD) is the measure reviews use to pool trials that scored the same outcome on different questionnaires; a negative value favours acupuncture.
| Compared with | Outcome | Trials (participants) | Result | Certainty |
|---|---|---|---|---|
| Sham | Pain | 7 (1,403) | 9.2 points lower on 0 to 100, below the 15-point threshold | Low |
| Sham | Back function | 5 (1,481) | No clear difference (SMD −0.16, 95% CI −0.38 to 0.06) | Very low |
| Sham | Quality of life | 3 (1,068) | SMD 0.24, judged not clinically meaningful | Low |
| No treatment | Pain | 4 (366) | 20.3 points lower on 0 to 100, clinically important | Moderate |
| No treatment | Back function | 5 (2,960) | SMD −0.53, better function | Moderate |
| Usual care | Pain | 5 (1,054) | 10.3 points lower on 0 to 100, below the threshold | Low |
| Usual care | Back function | 5 (1,381) | SMD −0.47, better function | Low |
| Usual care | Physical quality of life | 1 (731) | 4.2 points higher | Moderate |
The review downgraded its evidence for risk of bias, inconsistency between trials and imprecision. Most trials were at high risk of performance bias because the acupuncturist always knows which treatment they are giving. Seven trials run in Germany, with 5,572 participants, supplied 67% of everyone in the review. Four trials with 465 participants found adverse events at a similar rate after acupuncture and after sham (low certainty); the events most often reported were pain where a needle went in, bruising, bleeding, worsening of the back pain and pain elsewhere. The authors concluded that whether to use acupuncture for chronic low back pain might depend on its availability, its cost and the patient’s preferences.
Before and after: the 2005 review and the 2025 overview
The 2020 review replaced a 2005 Cochrane review by Furlan and colleagues, which covered 35 trials: 20 published in English, seven in Japanese, five in Chinese and one each in Norwegian, Polish and German. For chronic pain it found acupuncture better than no treatment or sham immediately after the sessions and at short-term follow-up only, no better than other conventional and alternative treatments, and of small added benefit on top of conventional therapy. It found only three trials of acute low back pain, too small and too weak to support a conclusion. Mu and colleagues split that review into separate acute and chronic reviews; this page found no full Cochrane review of acupuncture for acute low back pain published since.
In March 2025 a Cochrane overview of reviews by Rizzo and colleagues gathered 31 Cochrane reviews of non-drug, non-surgical treatments for low back pain, covering 644 trials and 97,183 adults, and took comparison with placebo or sham as its primary question. For chronic low back pain it reported, all at moderate certainty, that acupuncture probably improves function slightly compared with sham (SMD −0.38; 3 trials, 957 participants), probably reduces pain compared with no treatment (10.1 points on 0 to 100; 3 trials, 144 participants) and probably improves function slightly compared with usual care (1 trial, 734 participants). For comparison, the same overview found that exercise therapies probably reduce chronic back pain by 15.2 points on 0 to 100 against no treatment or usual care, from 35 trials.
The trials behind the reviews
Three large trials show the comparisons the reviews pool, two of them with a sham group and one without.
| Trial | Participants | Groups | Main result |
|---|---|---|---|
| GERAC, Germany, Haake et al. 2007 | 1,162, treated in 340 practices | Ten sessions of acupuncture; sham (shallow needling away from acupuncture points); conventional therapy of drugs, physiotherapy and exercise | Responders at 6 months: 47.6%, 44.2% and 27.4%. Acupuncture and sham did not differ; both outperformed conventional therapy |
| Cherkin et al. 2009, US | 638 | Individualised acupuncture; standardised acupuncture; simulated acupuncture, a toothpick in a needle guide tube tapped against the skin at the same points; usual care. Ten treatments over 7 weeks | At 8 weeks, 60% of the three needling-type groups improved meaningfully on a disability scale against 39% with usual care. The three did not differ from each other |
| BackInAction, US, DeBar et al. 2025 | 800 adults aged 65 or over | 8 to 15 sessions over 12 weeks; the same plus 4 to 6 maintenance sessions; usual medical care alone. No sham group | Meaningful improvement in disability at 6 months: 39.1% and 43.8% against 29.4%, still present at 12 months |
Cherkin’s group wrote that tailoring needle sites to the patient and piercing the skin both appeared unimportant to the benefit, and that it remained unclear whether their simulated procedure was a physiological stimulus or a placebo. At one year, 59% to 65% of their three treated groups had improved meaningfully on disability against 50% with usual care; the difference in how much the symptoms bothered people was no longer significant. BackInAction was run to inform Medicare’s decisions on paying for acupuncture, and its serious adverse events were rare, similar across the three groups, and fewer than 1% judged possibly related to acupuncture.
What the guidelines say
NICE, which writes clinical guidance for the NHS in England, published its guideline on low back pain and sciatica, NG59, in November 2016. Recommendation 1.2.8 reads: do not offer acupuncture for managing low back pain with or without sciatica. The recommendation is marked [2016], meaning its evidence was last reviewed that year. NICE last updated the guideline in July 2026, when it withdrew the recommendations on psychological therapy, and left the acupuncture recommendation as it was. Sciatica and lumbar disc herniation reports the trials and reviews in leg pain from a disc.
The American College of Physicians issued a guideline in 2017, written by Qaseem and colleagues, that went the other way. For acute or subacute back pain it said clinicians and patients should choose a non-drug treatment first, from superficial heat (moderate-quality evidence), massage, acupuncture or spinal manipulation (low-quality evidence). For chronic back pain it listed acupuncture, on moderate-quality evidence, among the non-drug treatments to select first, alongside exercise, multidisciplinary rehabilitation and mindfulness-based stress reduction. Both were graded strong recommendations. The College now lists the 2017 guideline on its page of inactive guidelines; its guideline committee treats any guideline five or more years old as no longer active. Clinical guidelines sets these positions beside the other bodies’ condition by condition.
What the reviews could not settle
The size of the gap between acupuncture and sham depends on what the sham does. GERAC’s sham pierced the skin and Cherkin’s only pressed on it, and in both trials the sham group did about as well as the acupuncture group and better than conventional therapy or usual care. The reviews can measure that pattern and cannot say what produces it: needling of any kind, the ritual and attention of treatment, or patients’ expectations. A trial cannot blind the person holding the needle, which is the main reason the Cochrane review rated most trials at high risk of bias.
Two further limits narrow what the pooled figures describe. Two thirds of the participants in the Cochrane review came from seven German trials, so the pooled estimates rest heavily on acupuncture as practised in Germany. Most of its results were measured immediately after treatment or in the short term; BackInAction, at 12 months, is one of the few with longer follow-up, and it compared acupuncture with usual care rather than sham. Acute low back pain remains without a sizeable body of trials. The individual patient data analyses that pool back and neck pain with other chronic pain are covered in Chronic pain.
AcuiQ’s low back pain page lists the protocols individual studies prescribed, each with its citation, and the chronic low back pain, back pain and sciatica pages do the same for the terms those studies used.