Skip navigation

Acupuncture for back pain?

Bandolier well remembers a television programme in which an acupuncturist blithely claimed that acupuncture could 'cure' 60% of all chronic back pains. A bit over-the-top, we thought at the time, but we all know someone who claims to have been 'cured' by acupuncture, or by medical interventions of which we are sceptical. A systematic review of acupuncture in back pain [1] is therefore a welcome relief from ignorance when it helps make sense of a difficult subject.


Very considerable efforts went into trying to find all the trials. These included MEDLINE, Cochrane Library, and a database specialising in complementary medicine. Authors publishing within the last five years or so were also contacted.

Included studies

Studies were selected on the basis that they were randomised, that dry needles were inserted into the skin, and which was described by the authors as acupuncture. Studies were assessed methodologically both in terms of methodological quality (randomisation, blinding, withdrawals) and on the quality of the acupuncture as judged independently (and blind) by six experienced medical acupuncturists.

Most of the included studies used sham acupuncture as control, though waiting list controls and lidocaine injections were also used. The outcomes were almost always short-term pain relief, as judged by the patient in some instances, and by the practitioner in others. There were few long-term outcomes.


Of the twelve included studies, only four were blinded. Adequacy of acupuncture was judged (on a 0-2 scale, where 2 is best) as 2 in one, 1 in eight and 0 in three. Not all these studies had extractable outcome data.

Four blinded studies showed no difference from control (Figure, Table).
Effects of acupuncture on short-term outcomes in back pain
Type of study Number of trials Improved with acupuncture (%) Improved with control (%) Relative benefit (95% CI) NNT (95%CI)
Blind 4 73/127 (57) 61/123 (50) 1.2 (0.9 to 1.5) 13 (5 to no benefit)
Non-blind 5 78/117 (67) 33/87 (38) 1.8 (1.3 to 2.4) 3.5 (2.4 to 6.5)
Fifty-seven percent of patients improved with acupuncture and 50% with control, a relative benefit of 1.2 (95% confidence interval 0.9 to 1.5). The number needed to treat with acupuncture for one patient with back pain to achieve a short-term improvement was 13 (95% confidence interval 5 to no benefit).

Five non-blinded studies did show a difference from control, with 67% improved with acupuncture and 38% with control. Here the relative benefit was significant at 1.8 (1.3 to 2.4) and the NNT was 3.5 (2.4 to 6.5).


Acupuncture trials are often difficult to assess. There are issues of whether the acupuncture has been done correctly, about the appropriateness of control interventions, whether trials can be truly blinded, and about the relevance of outcomes. Ernst & White have examined these in their review, and make thoughtful comments on all of them.

Perhaps the biggest problem is that these trials, as a group, have avoided the hard question of longer-term outcomes. Even if acupuncture provides short-term relief, its place in management of back pain remains unknown.

The question is whether this review provides evidence of lack of effect, or lack of evidence of effect. The inability of the four highest quality blinded trials to show a statistically significant short-term improvement must be worrying for those providing acupuncture services, and for the health services or individual who purchase acupuncture. A sceptical view seems to be most appropriate until trials of high quality prove that to be wrong.


  1. E Ernst, AR White. Acupuncture for back pain: A meta-analysis of randomised controlled trials. Archives of Internal Medicine 1998 158: 2235-2241.

previous or next story in this issue