Evid Based Med 18:33-34 doi:10.1136/eb-2012-100723
  • Therapeutics
  • Randomised controlled trial

Acupuncture prophylaxis of migraine no better than sham acupuncture for decreasing frequency of headaches

  1. Hans-Christoph Diener
  1. Department of Neurology, University Hospital Essen, Essen, Germany
  1. Correspondence to: Hans-Christoph Diener
    Department of Neurology and Headache Center, University Hospital Essen, Hufelandstrasse 55, Essen 45147, Germany; h.diener{at}

Commentary on: [CrossRef]Google Scholar


Migraine is a frequent and often disabling condition, in particular when migraine attacks occur frequently. Preventive therapy is recommended for patients with frequent attacks and related disability. Migraine prophylaxis can be performed with medications such as beta-blockers, flunarizine, amitriptyline and anti-epileptics such as valproic acid or topiramate.1 Non-drug treatment includes education, behavioural therapy and exercise. Acupuncture is popular in many Western countries for pain conditions including migraine. Many acupuncture trials have been performed in the past for migraine prophylaxis with conflicting results.2 In general, large studies of good methodological quality find no, or only a small difference between real and sham acupuncture.3–5


Li and colleagues performed a well-designed, multicentre randomised trial of acupuncture for the prophylaxis of migraine. With 480 patients, the study was adequately powered, the design followed the recommendations of the International Headache Society6 and the methods were published beforehand. The authors randomised patients with frequent migraines into four groups: Shaoyang-specific acupuncture, Shaoyang-non-specific acupuncture, Yangming-specific acupuncture and sham acupuncture (control). All acupuncturists were highly experienced. The patients received 20 treatments over a period of 4 weeks. The primary endpoint was the number of migraine days during weeks 5–8 after randomisation. The population studied was typical for migraine studies with a majority of middle-aged women with a mean of 5.5–6.3 migraine days per month.


Migraine frequency decreased in all groups. However, there was no statistically significant difference between the four study groups for the primary endpoint. Most secondary endpoints showed superiority for the acupuncture groups compared with sham acupuncture, but the absolute differences were small and hardly clinically meaningful.

The authors concluded that acupuncture has clinically minor effects on migraine frequency compared with sham acupuncture.


The results of this well-designed study are in line with two other adequately powered trials, which also found no difference between real acupuncture and sham acupuncture.3 ,5 If one considers all the published trials of the efficacy of acupuncture for the prophylaxis of migraine there is an inverse relationship between the quality of the trial and positive outcomes, meaning that small and poorly designed trials were more likely to show superiority of acupuncture over sham acupuncture or drug treatment.

Do the results of this and other trials mean that acupuncture should no longer be used for the prophylaxis of migraines? Acupuncture per se appears not to have efficacy. However, studies with no sham control group suggest that acupuncture (some combination of the visits, the attention and needle insertion regardless of exactly where and how) may be associated with improvement. Some might wish to experience these benefits by undergoing acupuncture. We also need to consider the many patients who cannot use drug treatment for migraine prophylaxis because of either contraindications or adverse events. Adherence to and compliance with medications prescribed for migraine prophylaxis is very poor.7 In addition, many healthcare systems are unable to provide integrated headache care which combines drug treatment with counselling, behavioural therapy and exercise.

Are there scientific lessons to be learned from this study? The results indicate that the effects of acupuncture in migraine prevention are most likely because of a placebo effect. An ethical question arises whether placebo treatments can be applied. In my opinion, the clear answer is yes. The therapeutic gain for most drug treatments compared with placebo in terms of difference in efficacy is small.8


  • Competing interests None.