Abstract
This study aimed to evaluate the effects of treadmill-based virtual-reality (VR) biofeedback training on gait function in a cohort of children and adolescents with cerebral palsy (CP) or acquired brain injury (ABI). Secondary aims examined the influence of the amount and frequency of treatment sessions, treatment goals, and diagnosis. Data were collected from 135 participants (CP: n = 61; ABI: n = 74) aged 12.4 ± 3.5 years, at two rehabilitation centres with varying training protocols, (ranging between 5 and 24 training sessions). Participants completed a series of VR treadmill-based biofeedback sessions targeting gait function, gait quality, or balance. Gait function was assessed using the Six-Minute Walk Test (6MWT), self-selected walking speed (SSWS), and the Gillette Functional Assessment Questionnaire Walking Scale (FAQ-WS). This was a multicentre observational pre–post study using retrospectively collected clinical data from one center and previously gathered prospective data from a second center. Significant pre-post improvements were observed in the 6MWT (+ 27.2 m, 95% CI 20–41 m, p < 0.001), SSWS (+ 0.12 m/s, 95% CI 0.09–0.14 m/s, p < 0.001) and FAQ-WS score (median = 0, IQR= -1 to 2), p < 0.001). The amount and frequency of treatment sessions showed weak associations with the changes in SSWS (r(s)=0.27, p = 0.002 and r(s)=-0.25, p = 0.015, respectively) but did not influence 6MWT or FAQ-WS changes. Participants with a primary goal of improving gait function (n = 62) presented greater 6MWT improvements than those without this goal (mean change 45 m, 95% CI 29–62 m vs. 16 m, 95% CI 4–28 m; F1,120 = 8.280, p = 0.005). The ABI cohort showed larger 6MWT improvements than the CP cohort. In this multicentre observational cohort, gait-function measures were higher after the VR treadmill-based biofeedback training series than before. Pre–post improvements were larger in participants with gait-function-related goals and in those with ABI, whereas associations with the amount and frequency of sessions were weak. Given the observational design and concurrent care, the results highlight the potential clinical utility of GRAIL as an enjoyable, motivating biofeedback environment in which gait-related outcomes can improve over a training series.