Effect of Virtual Reality-Based Training on Upper Limb Dysfunction during Post-Stroke Rehabilitation: A Meta-Analysis Combined with Meta-Regression

Background: Recently, there has been a surge in virtual reality (VR)-based training for upper limb (UL) rehabilitation, which has yielded mixed results. Therefore, we aimed to explore the effects of conventional therapy combined with VR-based training on UL dysfunction during post-stroke rehabilitation. Methods: Studies published in English before May 2023 were retrieved from PubMed, Embase, and the Cochrane Library. We also included randomized controlled trials that compared the use of conventional therapy and VR-based training with conventional therapy alone in post-stroke rehabilitation. The meta-analysis was performed using Review Manager Software (version 5.3; The Nordic Cochrane Centre, The Cochrane Collaboration; Copenhagen, Denmark) and Stata/MP 17.0 (StataCorp, LLC, College Station, TX, USA). Univariate and multivariate meta-regression analyses were performed to investigate the effects of stroke duration, VR characteristics, and type of conventional therapy on VR-based training. Results: In total, 27 randomized controlled trials were included, which enrolled 1354 patients. Our results showed that conventional therapy plus VR-based training is better than conventional therapy alone in UL motor impairment recovery measured using Fugl-Meyer Upper Extremity (standardized mean difference [SMD] = 0.32, 95% confidence interval [CI]: 0.07-0.57, Z = 2.52, p = 0.01). Meta-regression showed that stroke duration had independent effects on Fugl-Meyer Upper Extremity scores of VR-based training in rehabilitation (p = 0.041). Furthermore, in subgroup analysis based on stroke duration, stroke duration >6 months was statistically significant (SMD = 0.20, 95% CI: 0.01-0.39, Z = 2.06, p = 0.04). No relevant publication bias (p = 0.1303), and no significant difference in activity limitation assessed using the Box-Block Test (mean difference [MD] = 2.79, 95% CI: -0.63-6.20, Z = 1.60, p = 0.11) was observed. Regarding the functional independence measured using the Functional Independence Measure scale, studies presented no significant difference between the experimental and control groups (MD = 1.15, 95% CI: -1.84-4.14, Z = 0.76, p = 0.45). Conclusions: Conventional therapy plus VR-based training is superior to conventional therapy alone in promoting the recovery of UL motor function after stroke. Therefore, VR-based training may be a potential option for improving UL motor function. The study was registered on PROSPERO (https://www.crd.york.ac.uk/prospero/), registration number: CRD42023472709.