---
title: "Migraine"
url: "https://acuiq.com/wiki/evidence/migraine"
updated: 2026-10-09
last_verified: 2026-10-09
license: CC-BY-4.0
license_url: "https://creativecommons.org/licenses/by/4.0/"
section: "evidence"
basis: "Modern research"
audience: "Anyone"
published: "2026-10-09"
reading_minutes: 9
sources: 10
---

# Migraine

_What the Cochrane review, the large randomised trials and NICE report on acupuncture to prevent migraine, comparison by comparison._

Most acupuncture trials for migraine test it as prevention: a course of sessions given between attacks, with migraine days or attacks counted in a diary before and after. NICE, the body that writes clinical guidance for England, [defines episodic migraine](https://www.nice.org.uk/guidance/cg150/chapter/Recommendations) as migraine on fewer than 15 days a month, and chronic migraine as headache on 15 or more days a month for more than three months, at least 8 of them with features of migraine. The evidence on episodic migraine was last pooled by Cochrane in a [2016 review](https://doi.org/10.1002/14651858.CD001218.pub3) by Klaus Linde and colleagues at the Technical University of Munich, and what it found depends on what acupuncture was compared with: no preventive treatment, sham acupuncture, or preventive drugs. Sham acupuncture is a control procedure built to look and feel like acupuncture without being it, such as needles that retract into their handles or shallow needling away from the points; [Sham acupuncture](/wiki/research/sham-acupuncture) sets out the kinds in use and why none is inert. This page keeps the three comparisons apart throughout.

## The 2016 Cochrane review

The review included 22 randomised trials with 4,985 adults who had had episodic migraine for at least a year, each trial running at least eight weeks. It removed five trials that had been in its 2009 version, because their participants had had migraine for less than 12 months, and added five new ones. Fifteen trials had a sham control, five a no-acupuncture control (treatment of attacks only, or routine care) and five a preventive drug; some trials had more than one control group. The authors rated the evidence with GRADE, the scale Cochrane uses to say how much confidence a result supports, from high to very low.

The main outcome was the share of people whose headache frequency at least halved. The number needed to treat, in the fourth column, is how many people would need acupuncture for one more of them to reach that threshold than in the comparison group.

| Compared with | Trials (people) | Acupuncture vs control | Number needed to treat | GRADE |
| --- | --- | --- | --- | --- |
| No acupuncture, after treatment | 4 (2,519) | 41% vs 17% | 4 | Moderate |
| Sham, after treatment | 14 (1,825) | 50% vs 41% | 11 | Moderate |
| Sham, at follow-up | 11 (1,683) | 53% vs 42% | 10 | Moderate |
| Preventive drugs, at 3 months | 3 (739) | 57% vs 46% | not given | Moderate |
| Preventive drugs, at 6 months | 3 (744) | 59% vs 54%, not statistically significant | not given | Moderate |

Measured as headache frequency rather than as responders, the difference from no acupuncture was a standardised mean difference of 0.56, which the authors call moderate; a standardised mean difference expresses the gap between groups in units of the spread of the results, so that trials using different diaries can be pooled. The difference from sham was 0.18 after treatment and 0.19 at follow-up, which they call small but statistically significant. Their 2009 version had not found an effect over sham. Against drugs, acupuncture reduced frequency more after treatment (0.25), but the difference was no longer significant at follow-up. Only one trial followed people after treatment against a no-acupuncture group; it found a small benefit 12 months after randomisation, in 377 people, rated low quality.

On harms, the review found no significant difference between acupuncture and sham in the number of people who dropped out because of adverse effects or who reported any. Against drugs, people having acupuncture were less likely to drop out because of adverse effects (4 trials, 451 people) and less likely to report one (5 trials, 931 people). [After a treatment](/wiki/safety/after-a-treatment) covers what acupuncture’s adverse events are and how often they occur.

## Pooling the raw data

The Acupuncture Trialists’ Collaboration took a different route: instead of pooling published averages, it collected each participant’s data from the trial authors. Its [2018 update](https://doi.org/10.1016/j.jpain.2017.11.005), led by Andrew Vickers at Memorial Sloan Kettering, covered 39 trials and 20,827 patients across four chronic pain conditions, chronic headache among them, and admitted only trials whose allocation was concealed. Across the conditions it found differences of about 0.5 standard deviations against no acupuncture and about 0.2 against sham, and the effect fell by about 15% over a year. The effect was smaller in trials whose sham used a penetrating needle, and the analysis found no clear link between results and the style of acupuncture given. [Chronic pain](/wiki/evidence/chronic-pain) reports the collaboration’s work in full.

## The large trials

Four randomised trials, two from Germany in the mid-2000s and two from China since, compared real acupuncture with sham and with a third group.

| Trial | People | Treatment | Outcome against sham | Outcome against the third group |
| --- | --- | --- | --- | --- |
| [Linde 2005](https://doi.org/10.1001/jama.293.17.2118), 18 centres in Germany | 302 | 12 sessions over 8 weeks | Days of moderate or severe headache fell by 2.2 in both groups | Waiting list: fell by 0.8; difference from acupuncture 1.4 days |
| [Diener 2006 (GERAC)](https://doi.org/10.1016/S1474-4422(06)70382-9), Germany | 960 randomised, 794 analysed | 10 sessions over 6 weeks | Migraine days at 6 months fell by 2.3 vs 1.5; not significant across the three groups | Standard preventive drugs: fell by 2.1 |
| [Zhao 2017](https://doi.org/10.1001/jamainternmed.2016.9378), 3 sites in China | 249 | 20 sessions over 4 weeks | Attacks at week 16 fell by 3.2 vs 2.1; difference 1.1 attacks | Waiting list: fell by 1.4; sham did not differ significantly from it |
| [Xu 2020](https://doi.org/10.1136/bmj.m697), 7 hospitals in China | 150 | 20 sessions over 8 weeks | Migraine days at weeks 17 to 20 fell by 3.9 vs 2.2; adjusted difference 2.1 days | Usual care alone: smaller reduction than acupuncture in every period from week 1 to week 20 |

In the 2005 trial the shares of people whose headache days at least halved were 51% with acupuncture, 53% with sham and 15% on the waiting list, and the authors concluded acupuncture was no more effective than sham. The GERAC trial lost 125 people straight after randomisation, 106 of them from the drug group, and its authors concluded that outcomes did not differ between the three treatments.

The 2017 and 2020 trials enrolled only people with migraine without aura and gave 20 sessions rather than 10 or 12. The 2020 trial used a sham needle that touched the skin without piercing it, and checked whether the blinding held: 79% of the acupuncture group and 75% of the sham group believed the needle had gone in. Its acupuncturists needled the same five points on both sides in everyone: [`LI04`](/points/LI04) (Hegu), [`LV03`](/points/LV03) (Taichong), [`XH05`](/points/XH05) (Taiyang), [`GB20`](/points/GB20) (Fengchi) and [`GB08`](/points/GB08) (Shuaigu). They added [`ST08`](/points/ST08) (Touwei), [`BL10`](/points/BL10) (Tianzhu) or [`GV20`](/points/GV20) (Baihui) according to which channel the headache was attributed to.

Why the German and Chinese trials disagree about sham is not settled by any of them. They differ in the number of sessions, the participants enrolled and the country, and [Where the trials come from](/wiki/research/where-the-trials-come-from) sets out what the share of positive results by country shows. [Reading an acupuncture trial](/wiki/research/reading-an-acupuncture-trial) explains how control choice and blinding shape results like these.

## Chronic migraine and single attacks

The 2016 review covered episodic migraine only. For chronic migraine, a [2024 trial](https://doi.org/10.1177/03331024241261080) at the Beijing Hospital of Traditional Chinese Medicine randomised 60 people to acupuncture three times a week plus dummy topiramate tablets, or topiramate plus sham acupuncture, for 12 weeks. Monthly migraine days fell by 2.8 more in the acupuncture group over the 12 weeks and by 3.3 more over the following 12. Topiramate is one of the preventive drugs NICE names first.

Some trials have tested acupuncture during an attack instead. A [2023 review](https://doi.org/10.1136/bmjebm-2022-112135) from Capital Medical University, Beijing, pooled 21 comparisons from 15 studies with 1,926 adults, searching English and Chinese databases. Against sham, more people were free of headache two hours after acupuncture (2 studies, 180 people), at low certainty. Against drugs it found little or no difference in headache freedom at two hours (4 studies, 294 people), also at low certainty.

## What guidelines say

NICE’s headache guideline, [CG150](https://www.nice.org.uk/guidance/cg150/chapter/Recommendations), was published in September 2012 and last updated in June 2025. For preventing migraine it names propranolol, topiramate and amitriptyline first. Recommendation 1.3.21 says that if all three have not worked, are not tolerated or are unsafe for the person, a course of up to 10 sessions of acupuncture over 5 to 8 weeks can be considered, taking account of the person’s preference, other conditions and risk of adverse events; the wording dates from 2012 and was amended in 2025. [Clinical guidelines](/wiki/evidence/clinical-guidelines) compares what other guideline bodies say.

## What the reviews could not settle

The 2016 review found no trials longer than a year, and notes that the same is true of other migraine treatments. Its comparisons with no acupuncture and with drugs could not be blinded, so people knew which treatment they had, which raises the risk of bias in those results. The margin over sham is small, and the collaboration’s data show it shrinks further when the sham pierces the skin, so how much of the effect is specific to needling particular points remains open. The chronic migraine evidence above is a single-centre trial of 60 people.

Cochrane has started a new review. Its [protocol](https://doi.org/10.1002/14651858.CD015528), published in February 2025 from the Centre for Evidence-Based Chinese Medicine at Beijing University of Chinese Medicine, with Linde and Vickers among its authors, gives three reasons: trials published since 2016, Chinese databases the earlier review did not search, and changes to Cochrane’s methods. Until it reports, the 2016 figures above are Cochrane’s current estimate.

AcuiQ’s [migraine page](/symptoms/headache-migraine), with the related [migraine without aura](/symptoms/migraine-without-aura) and [chronic migraine](/symptoms/headache-migraine-chronic) pages, lists the protocols individual studies prescribed, each with its citation. [Tension-type headache](/wiki/evidence/tension-type-headache) covers the other common primary headache.

### Sources

1. Linde K, Allais G, Brinkhaus B, Fei Y, Mehring M, Vertosick EA, Vickers A, White AR, Acupuncture for the prevention of episodic migraine, Cochrane Database of Systematic Reviews 2016 — https://doi.org/10.1002/14651858.CD001218.pub3
2. Xia R, Linde K, Freilinger T, Vickers A, et al., Acupuncture for the prevention of episodic migraine (protocol), Cochrane Database of Systematic Reviews 2025 — https://doi.org/10.1002/14651858.CD015528
3. Vickers AJ, Vertosick EA, Lewith G, MacPherson H, et al., Acupuncture for chronic pain: update of an individual patient data meta-analysis, Journal of Pain 2018 — https://doi.org/10.1016/j.jpain.2017.11.005
4. Linde K, Streng A, Jürgens S, Hoppe A, et al., Acupuncture for patients with migraine: a randomized controlled trial, JAMA 2005 — https://doi.org/10.1001/jama.293.17.2118
5. Diener HC, Kronfeld K, Boewing G, et al., Efficacy of acupuncture for the prophylaxis of migraine: a multicentre randomised controlled clinical trial, Lancet Neurology 2006 — https://doi.org/10.1016/S1474-4422(06)70382-9
6. Zhao L, Chen J, Li Y, et al., The long-term effect of acupuncture for migraine prophylaxis: a randomized clinical trial, JAMA Internal Medicine 2017 — https://doi.org/10.1001/jamainternmed.2016.9378
7. Xu S, Yu L, Luo X, et al., Manual acupuncture versus sham acupuncture and usual care for prophylaxis of episodic migraine without aura: multicentre, randomised clinical trial, BMJ 2020 — https://doi.org/10.1136/bmj.m697
8. Liu L, Chen Q, Zhao L, et al., Acupuncture plus topiramate placebo versus topiramate plus sham acupuncture for the preventive treatment of chronic migraine: a single-blind, double-dummy, randomized controlled trial, Cephalalgia 2024 — https://doi.org/10.1177/03331024241261080
9. Wang Y, Du R, Cui H, Zhang L, Yuan H, Zheng S, Acupuncture for acute migraine attacks in adults: a systematic review and meta-analysis, BMJ Evidence-Based Medicine 2023 — https://doi.org/10.1136/bmjebm-2022-112135
10. National Institute for Health and Care Excellence, Headaches in over 12s: diagnosis and management (CG150), published 2012, last updated June 2025 — https://www.nice.org.uk/guidance/cg150/chapter/Recommendations