Effects of robotic gait training after stroke: A meta-analysis

Abstract

Background: Robotic devices are often used in rehabilitation and might be efficient to improve walking capacity after stroke.

Objective: First to investigate the effects of robot-assisted gait training after stroke and second to explain the observed heterogeneity of results in previous meta-analyses.

Methods: All randomized controlled trials investigating exoskeletons or end-effector devices in adult patients with stroke were searched in databases (MEDLINE, EMBASE, CENTRAL, CINAHL, OPENGREY, OPENSIGLE, PEDRO, WEB OF SCIENCE, CLINICAL TRIALS, conference proceedings) from inception to November 2019, as were bibliographies of previous meta-analyses, independently by 2 reviewers. The following variables collected before and after the rehabilitation program were gait speed, gait endurance, Berg Balance Scale (BBS), Functional Ambulation Classification (FAC) and Timed Up and Go scores. We also extracted data on randomization method, blinding of outcome assessors, drop-outs, intention (or not) to treat, country, number of participants, disease duration, mean age, features of interventions, and date of outcomes assessment.

Results: We included 33 studies involving 1466 participants. On analysis by subgroups of intervention, as compared with physiotherapy alone, physiotherapy combined with body-weight support training and robot-assisted gait training conferred greater improvement in gait speed (+0.09m/s, 95% confidence interval [CI] 0.03 to 0.15; p=0.002), FAC scores (+0.51, 95% CI 0.07 to 0.95; p=0.022) and BBS scores (+4.16, 95% CI 2.60 to 5.71; p=0.000). A meta-regression analysis suggested that these results were underestimated by the attrition bias of studies.

Conclusions: Robot-assisted gait training combined with physiotherapy and body-weight support training seems an efficient intervention for gait recovery after stroke.

Keywords: End-effector; Exoskeleton; Gait; Rehabilitation; Robot-assisted gait training; Stroke; Walking.