LAUSR

OBJECTIVE To identify the rehabilitative effects of robot-assisted therapy on balance function among stroke patients. DESIGN A systematic review and meta-analysis of randomized controlled trials. DATA SOURCES Thirteen electronic databases were systematically searched from inception to March 2018: Web of Science, PubMed, EMBase, The Cochrane Library, Science Direct, CINAHL, MEDLINE, AMED, Physiotherapy Evidence Database, SPORTDiscus, WanFang Data, China National Knowledge Infrastructure and Chinese Scientific Journal Database. REVIEW METHODS Randomized controlled trials were retrieved for identifying the effects of robot-assisted therapy on balance function among stroke patients. Two authors independently searched databases, screened studies, extracted data, and evaluated the methodological quality and risk bias of each included study. A standardized protocol and data-collection form were used to extract information. Effect size was evaluated by mean difference with corresponding 95% confidence intervals. Methodological quality and risk bias evaluation for each included study followed the quality appraisal criteria for randomized controlled trials that were recommended by Cochrane Handbook. Meta-analysis was conducted by utilizing Review Manager 5.3, a Cochrane Collaboration tool. Data was synthesized with descriptive analysis instead of meta-analysis where comparisons were not possible to be conducted with a meta-analysis. RESULTS Thirty-one randomized controlled trials with a total of 1249 participants were included. The majority of the included studies contained some methodological flaws. The results of the meta-analysis indicated that robot-assisted therapy produced positive effects on balance function, as shown by an increase in the Berg balance scale score [random effects model, mean difference = 4.64, 95%CI = 3.22-6.06, P＜0.01], as well as Fugl-Meyer balance scale scores [fixed effects model, mean difference = 3.57, 95%CI = 2.81-4.34, P＜0.01]. After subgroup and sensitivity analyses, the positive effects were not influenced by different types of robotic devices, by whether robot-assisted therapy was combined with another intervention or not, or by differences in duration and intensity of intervention. CONCLUSION Evidence in the present systematic review indicates that robot-assisted therapy may produce significantly positive improvements on balance function among stroke patients compared with those not using this method. More multi-center, high-quality and large-scale randomized controlled trials following the guidelines of CONSORT are necessary to generate high-quality evidence in further research.