Effectiveness of Physical Rehabilitation Interventions on Walking Capacity and Wearable Sensor-Derived Performance...

Improvements in walking capacity after stroke do not always translate into increased real-world performance, highlighting a critical gap in promoting long-term health after rehabilitation. This systematic review and meta-analysis evaluated the effects of physical rehabilitation interventions, including exercise, behaviour change techniques (BCTs), and combined approaches, on wearable sensor-derived movement behaviours and walking capacity in people with stroke. Four databases (MEDLINE, Embase, CINAHL, and Scopus) were searched through May 2026 for randomized controlled trials comparing these interventions with usual care or other interventions. Outcomes included steps per day, gait speed, and walking endurance (6-min walk test; 6MWT). Random-effects meta-analyses were conducted, reporting standardized mean differences (SMDs) and 95% confidence intervals (CIs). Thirty-one studies met inclusion criteria, and 26 distinct studies were included in the meta-analyses. For real-world performance, BCT-only interventions demonstrated the largest pooled effect estimate on steps per day (SMD = 0.45; 95% CI: 0.22-0.69), followed by combined BCT + Exercise interventions (SMD = 0.38; 95% CI: 0.10-0.66), while Exercise-only interventions showed a smaller but statistically significant effect (SMD = 0.23; 95% CI: 0.02-0.44). Conversely, Exercise-only interventions were the only category associated with statistically significant improvements in walking capacity, improving both self-selected gait speed (SMD = 0.36; 95% CI: 0.13-0.58) and walking endurance (6MWT; SMD = 0.41; 95% CI: 0.25-0.57). In contrast, neither combined BCT + Exercise nor BCT-only interventions demonstrated statistically significant effects across gait speed or endurance outcomes. These findings suggest that exercise improves walking capacity, whereas interventions incorporating BCTs may have greater effects on real-world walking performance after stroke.