GRADE Working Group grades of evidence. High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.a Of the 33 included RCTs, the acupuncture treatment type and period were heterogeneous. The acupuncture treatment period ranged from one to three months. The acupoints varied across trials. The needling sensation could be provoked by manual stimulation or electrical stimulation. b Downgraded one level for serious inconsistency: variation in the definition of dependency and acupuncture treatment type and duration. c Downgraded two levels for very serious risk of bias: Among the 11 included trials, eight had risk of performance bias and seven had risk of detection bias; the result was not consistent with the sensitivity analysis using only sham controls. d Downgraded two levels for very serious risk of bias: Among the eight included trials, six had risk of performance bias and four had risk of detection bias; the result was not consistent with the sensitivity analysis using only sham controls. e Downgraded two levels for very serious risk of bias: Among the five included trials, four had risk of performance bias; the result was not consistent with the sensitivity analysis using only sham controls. f Downgraded two levels for very serious inconsistency: considerable statistical heterogeneity (I 2 > 50%) and variation in acupuncture treatment type and duration. g Downgraded two levels for very serious risk of bias: Among the 13 included trials, at least eight trials had risk of allocation bias, performance bias, or detection bias; the result was not consistent with the sensitivity analysis using only sham controls. h Downgraded two levels for very serious risk of bias: Among the 11 included trials, at least six had risk of allocation bias, performance bias, or detection bias; the result was not consistent with the sensitivity analysis using only sham controls. i Downgraded two levels for very serious risk of bias: Among the 18 included trials, at least 10 trials had risk of allocation bias, performance bias, or detection bias; the result was not consistent with the sensitivity analysis using only sham controls. j Downgraded two levels for very serious risk of bias: Among the 22 included trials, at least 11 trials had risk of allocation bias, performance bias, or detection bias; the result was not consistent with the sensitivity analysis using only sham controls. k Downgraded two levels for very serious inconsistency: variation between trials in reporting of adverse events and in acupuncture treatment type and duration. Summary of findings 2. Acupuncture compared with sham control for patients with acute stroke Acupuncture com...