Comparative Effectiveness of Upper Limb Exercise Interventions in Individuals With Stroke: A Network Meta-Analysis

BACKGROUND: Although numerous effective exercise interventions can treat upper limb motor impairments after stroke, it remains unknown as to which are the most effective. The objective of the present study was to investigate the comparative effectiveness of various exercise interventions of the upper limb for individuals with an acute or subacute stroke. METHODS: For this systematic review with network meta-analysis, we searched PubMed/MEDLINE, Cochrane Library CENTRAL and Web of Science from database inception to September 2021 for randomized controlled trials examining individuals within 6 months of stroke onset, active upper limb exercise interventions, and any kind of control intervention. The primary outcome was upper limb motor function, secondary outcomes were activities of daily living and social participation, both assessed at post-intervention and follow-up. Nonspecific/multimodal active upper limb therapy was the standard comparator. Standardized mean differences, that is, Hedge’s g, were the effect size estimators. We calculated Frequentist-based network meta-analysis for the comparative effectiveness calculations using the R package netmeta. Main analyses were network plotting to display the geometry of the network and P-scores to summarize the intervention hierarchy. Results were derived from direct within-study and indirect between-study evidence comparisons. The Cochrane risk-of-bias tool II assessed all risk of bias domains. RESULTS: This review involved 145 randomized controlled trial on 6432 participants and 45 different treatment categories. The network meta-analysis analyzed 119 randomized controlled trials on 5553 participants and 41 different treatment categories. Electrical stimulation combined with task-specific training (standardized mean difference, 1.03 [95% CI, 0.51–1.55]; P <0.0001, P-score=0.11), high-volume constraint-induced movement therapy (0.86 [0.4–1.32]; P =0.0003, P-score=0.18), and strength training (0.65 [0.17–1.13]; P =0.01, P-score=0.28) were the most effective interventions (each k=107). CONCLUSIONS: Electrical stimulation combined with task-specific training (low evidence), high-volume constraint-induced movement therapy (moderate evidence), and strength training (low evidence) were the most effective interventions in improving upper limb motor function in individuals with a stroke. As the results were sensitive against a high risk of bias, likewise, these interventions should receive more attention in research and practice. Due to the heterogeneous use, electrical stimulation in combination with task-specific training should be further investigated in well-designed studies alongside other successful interventions (eg, constraint-induced movement therapy). REGISTRATION: URL: https://www.crd.york.ac.uk/prospero/ ; Unique identifier: CRD42021284064.