Pattern-Matched Powered Gait Orthosis Training in Patients with Neurological Gait Impairment: A Multicenter...

Background: Wearable powered gait orthoses offer a clinically flexible alternative to treadmill-based robotic systems, yet evidence on different device configurations matched to the site of neuromuscular impairment remains limited. Methods: In this multicenter prospective pilot study, 75 participants with neurological gait impairment were allocated to a hip orthosis (HO; n = 39) or a knee-ankle-foot orthosis (KAFO; n = 36) group based on clinical assessment of predominant gait pattern. Both groups completed six overground gait-training sessions over three weeks. Primary outcomes were the Six-Minute Walk Test (6MWT) and Ten-Meter Walk Test (10MWT), assessed without (WO) and with (WITH) the device. Secondary outcomes were the Berg Balance Scale (BBS), Timed Up and Go Test (TUG), and Dynamic Gait Index (DGI), all assessed without the device. Wilcoxon signed-rank tests were used for pre-to-post comparisons. Results: Both groups demonstrated significant improvements in primary walking outcomes, with consistent gains in unassisted (WO) 6MWT and 10MWT performance across groups and in device-assisted (WITH) 10MWT speed; the one exception was a small statistically significant but clinically negligible decrease in device-assisted 6MWT in the KAFO group (-4.1 m, below established MCID). In the KAFO group, BBS improved by a median of 5.5 points (43.5 to 49.0, p = 0.0005), TUG decreased by 5.1 s (p < 0.001), and DGI improved by 6.0 points (p = 0.002); all three changes exceeded published minimum detectable change thresholds. In the HO group, pre-to-post differences in BBS (+1.0), TUG (+0.8 s; an unfavorable direction), and DGI (-2.0; an unfavorable direction) were statistically detectable but small in absolute magnitude, fell at or below published thresholds for minimum detectable change, and should not be interpreted as clinically meaningful improvement. The WO-WITH performance gap in the KAFO group narrowed substantially after training, with 10MWT time no longer differing significantly between conditions at post-training (p = 0.116). Conclusions: Six sessions of gait pattern-matched powered gait orthosis training produced clinically meaningful within-group improvements in walking outcomes in both groups. In the KAFO group, balance and functional mobility outcomes also showed clinically meaningful improvements; in the HO group, balance and functional mobility outcomes showed only statistically detectable but clinically non-meaningful fluctuations around near-ceiling baseline scores. Walking benefits generalized to unassisted ambulation in both groups. These findings support the feasibility of an individualized orthosis prescription framework and provide a basis for future randomized controlled trials.