Adding robot-assisted gait training (RAGT) to conventional rehabilitation was associated with a higher rate of improvement in functional walking three months after a first ischemic stroke, a small randomized-trial analysis found. Sixteen of 17 patients in the RAGT group, or 94.1%, improved their Functional Ambulation Category (FAC) score by at least two points, compared with 12 of 19 patients receiving conventional rehabilitation alone, or 63.2%. The difference was 30.9 percentage points, with a 95% confidence interval from 5.3 to 54.8 percentage points, and the reported p-value was 0.044.
The clearest apparent difference was in a prespecified group with the lowest starting FAC scores, 0 or 1. There, 9 of 10 patients in RAGT, or 90.0%, reached the two-point improvement threshold, compared with 5 of 11, or 45.5%, in the conventional-rehabilitation group. The risk difference was 44.5 percentage points, with a 95% confidence interval from 2.4 to 74.2 points. The comparison within that subgroup had a p-value of 0.039, while the test of whether the treatment difference varied by starting FAC had an interaction p-value of 0.038. The authors describe this as an exploratory directional signal because the subgroup was critically underpowered.
Balance and mobility measures moved the same way
Other measures pointed in the same direction, although the evidence was less clear in some subgroups. On the BBS balance score, median improvement was 28 points with RAGT versus 15 with conventional rehabilitation overall. In the FAC 0 to 1 group, the figures were 27 and 12 points; in the FAC 2 to 3 group, they were 30 and 20. The corresponding p-values were 0.049, 0.031 and 0.198. The larger low-FAC contrast may partly reflect lower starting scores and greater room for recovery.
The TUG mobility test, another measure of movement, also favored RAGT. It could be assessed only in patients whose baseline FAC was at least 3. Median improvement was 11.2 seconds versus 6.4 seconds overall, and 10.9 versus 6.1 seconds among patients with anterior-circulation stroke. The reported p-values were 0.034 and 0.041. The posterior-circulation figures, 12.0 versus 7.2 seconds, were descriptive only because those groups contained two and four patients.
A tightly defined inpatient test
This was a pre-specified secondary analysis of a single-center randomized controlled trial. The parent trial involved hospitalized patients with first-ever MRI-confirmed ischemic stroke, early mRS disability scores of 2 to 4, and baseline FAC scores of 0 to 3. The study had ethics approval and written informed consent.
Patients assigned to the intervention received 15 sessions of Lokomat Pro FreeD RAGT over three weeks, added to protocol-defined conventional multidisciplinary rehabilitation. The control group received the conventional program alone. Assessments took place at baseline, after three weeks and at three months. Of 42 people randomized in the parent trial, 36 had complete three-month follow-up and formed this analysis: 17 in the RAGT group and 19 in the control group.
Subgroup comparisons were prespecified by stroke localization, baseline FAC of 0 to 1 versus 2 to 3, age below versus at least the sample median of 67, and recanalization treatment versus conservative management. The primary outcome was the share of patients whose FAC score improved by at least two points at three months. Secondary outcomes were absolute changes in BBS and TUG, and blinded evaluators assessed all three outcomes.
To analyze the results, the researchers used exact tests for the binary FAC comparisons, risk differences with 95% confidence intervals, and regression models to test interactions. The continuous measures were assessed with rank-based methods. A p-value below 0.05 was treated as nominally significant, and no correction was made for multiple comparisons. That combination makes chance findings a real concern, especially in the smallest subgroups.
Why the result needs a larger test
The main reason for caution is the study's size. The parent trial anticipated enrolling 80 participants but randomized 42, and 76.6% of screened patients were excluded. In the 36-person analysis, subgroup cells ranged from five to 14 patients per arm. The conventional group also had more hypertension, 100% versus 76.5% in the RAGT group, and more diabetes, 73.7% versus 29.4%. No formal multivariable adjustment for these comorbidities was applied.
Those imbalances mean the full difference in recovery cannot be reliably attributed to RAGT alone. Across the other clinically relevant subgroups, outcomes consistently favored RAGT and no subgroup showed a reversal of treatment direction. But the small subgroup cells limit confidence in conclusions about treatment-effect differences, leaving the low-FAC result as an exploratory signal rather than a definitive answer.
Paper data and sources
Original title: Robot-assisted gait training yields superior functional gains in non-ambulatory subacute stroke patients
Authors: Najsrova A, Fiedorova I, Szegedi I et al.
Journal/Repository: Frontiers in rehabilitation sciences
Status: Peer-reviewed
First online: 2026-10-08
DOI: Not available
Original paper · Full text