Stroke. 2010 Apr;41(4):e171-9. Epub 2010 Feb 18.

Acupuncture in poststroke rehabilitation: a systematic review and meta-analysis of randomized trials

Wu P, Mills E, Moher D, Seely D.

Source

Department of Research & Clinical Epidemiology, the Canadian College of Naturopathic Medicine, Toronto, Ontario, Canada.

Abstract

BACKGROUND AND PURPOSE:

Acupuncture is a low-risk treatment with purported claims of effectiveness for poststroke rehabilitation. To comprehensively assess the efficacy of acupuncture in poststroke rehabilitation, we conducted a systematic review and meta-analysis of all randomized clinical trials of acupuncture for poststroke rehabilitation.

METHODS:

We searched 7 English and 2 Chinese databases from inception to September 2009. Eligible studies included randomized clinical trials that evaluated the clinical efficacy of acupuncture in adult patients with disability after stroke. We extracted data on trial quality, protocol, and outcomes assessed. A summary OR was calculated based on pooled dichotomous results. I(2) was used to infer heterogeneity and we conducted metaregression to determine if specific covariates explained heterogeneity.

RESULTS:

Thirty-five articles written in Chinese and 21 articles written in English were included. The overall quality of the studies was "fair" and most studies were small (median n=86; range, 16 to 241). The majority (80%) of the studies reported a significant benefit from acupuncture; however, there was some evidence of publication bias. In 38 trials, data were available for meta-analysis and metaregression, yielding an OR in favor of acupuncture compared with controls (OR=4.33, 95% CI: 3.09 to 6.08; I2=72.4%). Randomization, modes of delivery, method of control, study source country, and reporting of randomization may explain some of the heterogeneity observed between the studies.

CONCLUSIONS:

Randomized clinical trials demonstrate that acupuncture may be effective in the treatment of poststroke rehabilitation. Poor study quality and the possibility of publication bias hinder the strength of this recommendation and argue for a large, transparent, well-conducted randomized clinical trial to support this claim and implement changes to clinical practice.