Acupuncture for functional recovery after stroke: a systematic review of sham-controlled randomized clinical trials

Data sources

We searched the following databases from their inception through October 2009: MEDLINE, AMED, British Nursing Index, CINAHL, EMBASE, PsycINFO, the Cochrane Library (2009; issue 4), and Asian journals and 12 major Korean traditional medicine journals (Appendix 1, available at www.cmaj.ca/cgi/content/full/cmaj.091113/DC1). We used the following search terms: (acupuncture OR acu*) AND (stroke OR apoplexy OR cva OR cerebrovascular attack OR cerebrovascular accident OR cerebrovascular* OR cerebral infarction OR cerebral hemorrhage OR cerebral*). (Details about our search strategy for each database are shown in Appendix 2, available at www.cmaj.ca/cgi/content/full/cmaj.091113/DC1). We also searched our own files manually for relevant articles. Hard copies of all articles were obtained and read in full. No language restrictions were imposed. We included dissertations and abstracts, provided they contained sufficient detail for critical evaluation.

Study selection

Two of us (J.C.K, B.C.S.) independently identified eligible articles. Disagreements were resolved through discussion with the other authors. All prospective clinical studies of needle acupuncture (with or without electrical stimulation) as an adjunct treatment for functional recovery after stroke were considered. Randomized clinical trials involving patients of any age or sex with ischemic or hemorrhagic stroke in the acute, subacute or chronic stage were eligible. Stroke had to have been diagnosed by means of computed tomography or magnetic resonance imaging, or clinically according to the World Health Organization definition. 7

Eligible trials were included regardless of time of treatment or the length of the treatment period. The control interventions were either sham acupuncture (acupuncture with or without needle penetration at the acupuncture points or non-acupuncture points) or subliminal acupuncture (electro-stimulation using electrodes attached to the skin). 8

Outcome measures

The outcome measures we considered were those related to neurologic deficit, activities of daily living (i.e., Barthel Index or Fugl–Meyer Assessment), motor recovery and quality of life. In addition, we recorded adverse events.

Data extraction, quality and validity assessment

Two of us (J.C.K, B.C.S.) extracted data according to predefined criteria. The Cochrane risk-of-bias tool 9 and the PEDro (Physiotherapy Evidence Database) scale 10 were used to assess methodologic quality of the trials.

The same two authors independently assessed the quality of the acupuncture technique as described previously. 11 Disagreements were resolved by discussion with the other authors.

Statistical analysis

In the absence of clinical heterogeneity, we synthesized the results in a meta-analysis. We compared the mean change in outcome measures with baseline values to assess differences between the intervention and control groups. Weighted mean differences or standardized mean differences and 95% confidence intervals (CIs) were calculated. Where necessary, we contacted the primary authors of the trials to acquire or verify data. Differences compared with sham control were considered relevant. The variance of the change was inferred using a correlation factor of 0.5. 12 Cochrane’s Q-test and I² were used to assess heterogeneity. Heterogeneity was assumed if the p value was less than 0.10 in the χ² test and the I² value was above 75%. 9 In case of heterogeneity, we attempted to identify and explain the heterogeneity using subgroup analysis. Subgroup analysis was performed for subsets of studies and stages of stroke. Where possible, we assessed publication bias using a funnel plot. Post hoc sensitivity analyses were per-formed to test the robustness of the overall effect.

Study description

We identified 664 potentially relevant articles (Appendix 3, available at www.cmaj.ca/cgi/content/full/cmaj.091113/DC1). Ten studies, involving a total of 711 participants, met our inclusion criteria (Figure 1). Key data are summarized in Table 1. 13^–22 Seven trials included patients in the acute or subacute stage of stroke, and three included patients in the chronic stage. Eight trials were from Western countries and published in English. 13^–15^,17^,18^,20^–22 Two trials were from China and published in Chinese. 16^,19 Details of the treatment regimens are summarized in Appendix 4 (available at www.cmaj.ca/cgi/content/full/cmaj.091113/DC1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1: Selection of sham-controlled randomized clinical trials of acupuncture for functional recovery after stroke. RCT = randomized controlled trial.

Two trials used sham electro-stimulation on acupoints. 14^,15 The other eight used sham acupuncture: 13^,16^–22 one trial used superficial acupuncture in the control procedure; 13 four 16^,18^,20^,21 used nonpenetrating acupuncture on nonacupuncture points or real acupuncture points; and three 17^,19^,22 used penetrating acupuncture on nonacupuncture points.

Study quality

The mean PEDro score was 6.6 (standard deviation [SD] 1.5), with a range of four to nine points (Table 2). The Cochrane risk of bias varied (Table 2). Only four studies reported appropriate sequence generation. 13^,15^,18^,20 Five trials used blinding of participants and assessors, 17^,18^,20^–22 three used blinding of participants only, 14^,16^,19 and two used blinding of assessors only. 13^,15 Six trials assessed the outcome of more than 85% of the allocated participants at the last follow-up. 13^,15^,17^–19^,21 Five trials mentioned that the outcomes were analyzed by intention-to-treat analysis. 13^–15^,18^,20 Incomplete outcome data were addressed adequately in six studies. 13^–15^,18^,20^,21 Three trials reported details about allocation concealment, and all of them used adequate methods. 15^,18^,20 Ethical approval was reported in four studies, 13^,15^,18^,22 and informed consent was described in seven. 13^–15^,17^,18^,20^,22 Success of blinding was assessed in five studies and found to be successful in all cases. 17^,18^,20^–22 The six studies with a high risk of bias 14^,16^,17^,19^,21^,22 showed a low quality score on the PEDro scale (mean score 5.7) compared with the other trials that had a low risk of bias (mean score 8.0). The degree of confidence in the quality of the acupuncture varied from 55% to 93% (mean 82.4%); there was good agreement between the two reviewers.

Outcome measures

Four trials tested the effects of acupuncture on neurologic deficits using the National Institutes of Health Stroke Scale, 16^,18 the European Stroke Scale 14 or the Scandinavian Stroke Scale. 13 One of the four studies showed positive therapeutic effects. 16 Seven trials tested the effects of acupuncture on activities of daily living according to the Barthel Index, 13^,14^,18^,20 the modified Barthel Index 15^,16 or the Sunaas Index of Activities of Daily Living. 13 Two of these studies (both from China) showed favourable effects on activities of daily living. 16^,19 Of the five trials that assessed the effects of acupuncture treatment on quality of life, none showed favourable effects. 13^,15^,18^,20^,22

Two of the seven trials that involved patients in the acute and subacute stages of stroke were excluded from the meta-analysis: one because it reported data as medians and inter-quartile ranges (IQRs) 18 and the other because it did not focus on functional recovery and did not report numerical data. 17 Our meta-analysis of the remaining five studies did not yield a significant difference in favour of acupuncture (n = 368; standard mean difference 0.47, 95% CI −0.02 to 0.96; I² = 81%) (Figure 2A). A post-hoc sensitivity analysis that excluded the two trials from China with a high risk of bias 16^,19 failed to show significant effects of acupuncture (n = 244; standard mean difference 0.07, 95% CI −0.18 to 0.32; I² = 0% (Figure 2B). In the trial excluded from the meta-analysis because it reported medians and IQRs, the median Barthel Index score for activities of daily living at baseline was 6 (IQR 3.8–9) in the acupuncture group and 6 (IQR 4–9.3) in the sham acupuncture group. After two weeks of treatment, the median score was 11 (IQR 5–17) in the acupuncture group (difference 4, IQR 0–8), and the median score was 11 (IQR 7.3–15.8) in the sham acupuncture group (difference 3, IRQ 0–7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2: Meta-analyses of sham-controlled randomized clinical trials of the effectiveness of acupuncture for functional recovery after stroke. Trials included patients in the acute and subacute stages of stroke. The subgroup analyses (panels B and E) excluded trials with a high risk of bias. Standard mean differences greater than zero favour acupuncture. CI = confidence interval, SD = standard deviation.

The same findings applied to activities of daily living after more than six months of follow-up measured with the Barthel Index (n = 244; standard mean difference 0.10, 95% CI −0.15 to 0.35; I² = 0%) (Figure 2C). For global neurologic deficits, none of the three trials included in the meta-analysis showed a significant difference in favour of acupuncture (n = 216; standard mean difference 0.20, 95% CI −0.23 to 0.62; I² = 59%) (Figure 2D). A subgroup analysis that excluded the trial from China with a high risk of bias 16 showed no significant effect of acupuncture (n = 152; standard mean difference −0.01, 95% CI −0.33 to 0.30; I² = 0%) (Figure 2E).

Of the three trials that involved patients in the chronic stage of stroke, 20^–22 one was excluded from the meta-analysis because it did not report numerical data. 21 A meta-analysis of the data from the other two studies did not show favourable effects of acupuncture on function according to the Modified Ashworth Scale (n = 58; mean difference 0.01, 95% CI −0.65 to 0.67; I² = 0%) (see Appendix 5, at www.cmaj.ca/cgi/content/full/cmaj.091113/DC1). 20^,22

The low number of trials prevented assessment of publication bias.

Adverse events

Four trials reported on the occurrence of adverse events: one reported a case of seizure in the acupuncture group, 18 another noted five cases of hematoma around the acupuncture points, 14 a third trial recorded two cases of dizziness (without group information), 19 and the fourth trial reported no adverse events. 15

Limitations

Our review has several limitations. Although great efforts were made to retrieve all trials on the subject, we cannot be certain that we succeeded. The paucity and often suboptimal quality of the primary data are further drawbacks.

Conclusion

Our meta-analyses of data from rigorous randomized sham-controlled trials did not show a positive effect of acupuncture as a treatment for functional recovery after stroke.