LAUSR

OBJECTIVE This study aimed to evaluate the effects of motor imagery (MI) on walking function and balance in patients after stroke. METHODS Related randomized controlled trials (RCTs) were searched in 12 electronic databases (Cochrane Central Register of Controlled Trials, PubMed, Science Direct, Web of Science, Allied and Complementary Medicine, Embase, Cumulative Index to Nursing and Allied Health Literature, PsycINFO, China National Knowledge Infrastructure, Chinese Biomedical Literature Database, WanFang, and VIP) from inception to November 30, 2016, and Review Manager 5.3 was used for meta-analysis. References listed in included papers and other related systematic reviews on MI were also screened for further consideration. RESULTS A total of 17 studies were included. When compared with "routine methods of treatment or training", meta-analyses showed that MI was more effective in improving walking abilities (standardized mean difference [SMD] = 0.69, random effect model, 95% confidence interval [CI] = 0.38 to 1.00, P < 0.0001) and motor function in stroke patients (SMD = 0.84, random effect model, 95% CI = 0.45 to 1.22, P < 0.0001), but no statistical difference was noted in balance (SMD = 0.81, random effect model, 95% CI = -0.03 to 1.65, P = 0.06). Statistically significant improvement in walking abilities was noted at short-term (0 to < six weeks) (SMD = 0.83, fixed effect model, 95% CI = 0.24 to 1.42, P = 0.006) and long-term (≥six weeks) assessments (SMD = 0.45, fixed effect model, 95% CI = 0.25 to 0.64, P < 0.00001). Subgroup analyses suggested that MI had a positive effect on balance with short-term duration (0 to < six weeks) (SMD = 4.67, fixed effect model, 95% CI = 2.89 to 6.46, P < 0.00001), but failed to improve balance (SMD = 0.82, random effect model, 95% CI = -0.27 to 1.90, P = 0.14) with long-term (≥six weeks) duration. CONCLUSION MI appears to be a beneficial intervention for stroke rehabilitation. Nonetheless, existing evidence regarding the effects of MI in patients after stroke remains inconclusive because of significantly statistical heterogeneity and methodological flaws identified in the included studies. More large-scale and rigorously designed RCTs in future research with sufficient follow-up periods are needed to provide more reliable evidence on the effects of MI in post-stroke patients.