Exercise-based interventions for post-stroke social participation: A systematic review and network meta-analysis.

Background: Resuming participation in society is an important goal of post-stroke rehabilitation. Exercise-based interventions have been shown to be effective non-pharmacological methods for improving social participation in post-stroke survivors, however it is unclear what the most effective types of exercise interventions are.

Objective: To assess the comparative effects and ranks of all exercise-based interventions in improving social participations in patients after a stroke.

Methods: A random-effects network meta-analysis was performed to identify evidence from relevant randomized control trials. We searched MEDLINE, CINAHL, EMBASE, PsycINFO, CINHAL, Cochrane Library, AMED, SPORTDiscus, Web of Science and Clinical Trials.gov from their earliest records to January 2020. Included trials must include at least one types of exercise for patients with stroke. The primary e was social participation. Bias will be assessed according to the revised Cochrane risk of bias tool. Data were analysed using Stata v14.0. Registration number of this study is CRD42020152523.

Results: A total of 16 randomized control trials involving 1704 patients and 12 intervention arms were included in our study. We performed three subgroup analyses divided based on follow up time (1 to <6 months post-treatment, and ≥6 months post-treatment), and intervention adherence. Based on the ranking probabilities, motor relearning programme was ranked as the most effective among all exercise interventions (surface under cumulative ranking curve values [SUCRCV]: 95.6%, standardized mean difference [SMD]: 2.72, 95% confidence interval [CI]: 1.76 to 3.69) in overall and short-term treatment efficacy. In the long-term subgroup, home-based combined exercise ranked the best for the efficacy of social participation improvements among stroke survivors (SUCRCV: 71.8%, SMD: -0.23, 95% CI: -0.61 to 0.15). In the analysis of all interventions with adherence of >90%, cognitive-based exercise ranked the best (SUCRCV: 100%, SMD: 2.64, 95% CI: 1.62 to 3.66).

Conclusions: Interventions that emerged with the highest ranks in our analysis might be considered in practice when resources allow. More large, well-designed multicentre trials are needed to support the conclusion of this study.