Upper Limb Robots for Recovery of Motor Arm Function in Patients With Stroke: A Systematic Review and Meta-Analysis

Robot technology to support upper limb (UL) rehabilitation poststroke has rapidly developed over the past 3 decades. We aimed to assess the effects of UL-robots (UL-RTs) on recovery of UL motor functioning and capacity poststroke when compared with any non-UL-RT and to identify variables that are associated with the found effect sizes (ESs).

Randomized controlled trials (RCTs) comparing UL-RTs with any other intervention on patients with UL limitations poststroke were identified in electronic searches from PubMed, Wiley/Cochrane Libraries, Embase, Cumulative Index of Nursing and Allied Health Literature, Web of Science, SportDISCUS, Physiotherapy Evidence Database (PEDro), and Google Scholar from inception until August 1, 2022. Two reviewers independently extracted relevant data using a Microsoft Excel spreadsheet. Meta-analyses were performed for measures of UL-muscle synergism, muscle power, muscle tone, capacity, self-reported motor performance, and basic activities of daily living (ADLs). Subgroup, sensitivity, and meta-regression analyses were applied to identify factors potentially associated with found ESs. Analyses were performed using Review Manager version 5.4 or IBM SPSS statistics version 27.

Ninety RCTs (N = 4,311) were included (median PEDro score 6 [6-7]). Meta-analyses of 86 trials (N = 4,240) showed small significant improvements in UL-muscle synergism (Fugl-Meyer Assessment of the UL [FM-UL]) (mean difference 2.23 [1.11-3.35]), muscle power (standardized mean difference [SMD] 0.39 [0.16-0.61]), motor performance (SMD 0.11 [0.00-0.21]), and basic ADLs (SMD 0.28 [0.10-0.45]). No overall effects were found for muscle tone (SMD -0.1 [-0.26 to 0.07]) or UL-capacity (SMD 0.04 [-0.10 to 0.18]), except with exoskeletons (SMD 0.27 [0.10-0.43]). Meta-regressions showed a significant positive association between baseline mean FM-UL and ESs for UL-capacity (r = 0.339; p = 0.03), in particular in the acute and early-subacute phases poststroke (r = 0.65; p = 0.01). No further significant subgroup differences or associations were found in our analyses.

The small significant effects found at the level of motor impairment do not show generalization to clinically meaningful effects at the level of UL-capacity. Meta-regressions suggest that selected participants with some potential of UL-recovery may benefit most from UL-RT, especially earlier poststroke. The robustness and consistency of our findings suggest that the development of the next generation of UL-RT needs to be guided by a better mechanistic understanding about assumed underlying interaction effects between motor learning and motor recovery poststroke.

A prospectively registered study protocol is available in the PROSPERO database under ID CRD42020197450.