Participants with chronic stroke (>3 months since stroke diagnosis) were recruited from a single neuro-physiotherapy practice (Hobbs Rehabilitation, Winchester, UK). All participants were diagnosed with stroke by a specialist neurologist/stroke consultant from a UK National Health Service Trust and had completed rehabilitation activities (i.e., inpatient and outpatient) in accordance with recommended guidelines (19). Written informed consent was obtained from all participants prior to the commencement of the study.

Inclusion criteria included: Individuals between 3 months and 5 years post-stroke, who were living in the community, medically stable, and cognitively capable, able to stand and step with an aid or with assistance (defined as a Functional Ambulation Categories between 2 and 5) (20), and who were receiving physiotherapy or attending a community-based, stroke support group at the time of study enrolment. Exclusion criteria included: body mass index (BMI) ≥40 kg/m², major arrhythmias, unresolved deep vein thrombosis, recent fractures of the symptomatic limb, open wounds, severe osteoporosis, and/or individuals who were non-weight bearing.

Participants were tested between 07:00 and 10:00 am in the physiology laboratory at the University of Winchester. Participants refrained from intense physical activity for 24 h prior to testing, and could only consume water for the 12 h before testing. Following an initial Functional Ambulation Classification and Modified Rankin Scale assessment to provide an indication of the degree of disability, participants lay supine for 15 min. Thereafter, pulse wave analysis (PWA), and regional (cfPWV) and local (common carotid) measures of arterial stiffness were assessed. Participants were randomized using covariate adaptive randomization (21) to either a 10-week home-based O-RAGT program, which included weekly “usual care” physiotherapy, or to a 10-week “usual care” physiotherapy only program (CON). Randomization involved sequentially assigning participants to O-RAGT or control by taking into account their age (age?70 vs. <70 years), systolic blood pressure (SBP ≥ 160 vs. < 160 mmHg) and time since stroke (< 12 vs. ≥ 12 months). Identical assessments were completed at baseline, post intervention (PI) and 3-months post-intervention (3PI). Participants and researchers collecting outcome data were aware of the allocated treatment condition, however, data analysts were blinded to the allocation.

Participants wore an ActivPAL3™ device (PAL Technologies Ltd., Glasgow, Scotland) for seven consecutive days and nights at baseline, PI and 3PI. The ActivPAL3 device was wrapped in a protective Tegaderm™ (3M, St Paul, USA) and attached to the anterior aspect of the upper third of the thigh, on the asymptomatic side. The ActivPAL3 provided a daily measure of the: (1) percentage of time spent sitting or lying, (2) percentage of time spent standing, (3) percentage of time spent stepping, and (4) step counts.

The O-RAGT device (Alter-G, Bionic Leg orthosis, Fremont, CA, USA) is a battery-operated, externally-wearable, dynamic device worn by stroke patients during rehabilitation. The device provides sensory inputs (i.e., auditory and sensory feedback), mobility assistance for users with reduced lower-limb function, and is fitted and worn in a manner similar to an orthopedic knee brace. The orthosis shell functions as the user interface that transfers the assistive torque to the human body, while an actuation unit assists the movement of the limb.

Participants were familiarized with the O-RAGT device before commencing the 10-week home-based program. Participants were encouraged to undertake at least 30 min per day of continuous or non-continuous bouts of walking and sit-to-stand exercises, at a moderate ratings of perceived of exertion (RPE 12–13). There was no maximum daily wear-time. O-RAGT settings associated with a participant's weight, assistance, resistance, threshold and knee extension angle settings were individualized and re-assessed every 2 weeks. Participants reported their number of steps, duration of use, activities undertaken and RPE for each day of activity. During this time, participants also continued their “usual care.”

Participants in both the control group and O-RAGT program undertook one-to-one, “usual care” physiotherapy sessions for the duration of the study. This included stretching and muscle strengthening exercises, functional movement activities (e.g., walking, step-ups, sit-to-stand) and soft-tissue massage. There were also group therapy activities which were based on the same principles but with less therapist engagement. For the duration of the 10-week program participants were advised to engage in at least 30 min of physical activity each day, undertaking similar functional movement patterns as those reported above.

Demographical and clinical comparisons between Conditions (O-RAGT, CON) was undertaken at baseline with independent sample t-tests (e.g., age, time since stroke, Functional Ambulation Classification, Modified Rankin Scale, PWA, cfPWV and carotid arterial stiffness outcomes) and chi-square tests (e.g., sex, stroke diagnosis), as appropriate.

To assess the effect of the O-RAGT intervention on the aforementioned regional and local hemodynamic properties, mixed model, two-factor analysis of covariance (ANCOVA), Condition (O-RAGT, control) x Time (BL, PI, 3PI), adjusted for baseline measures and age, were used to assess all PWA, cfPWV, carotid artrial stiffness and accelerometry outcomes. For PWV analysis, Mean Arterial Pressure was also used as a covariate. Partial eta squared (ηp²) was used to demonstrate the strength of the effect of exercise on the various outcome measures with 0.0099, 0.0588 and 0.1379 representing a small, medium and large effect, respectively (27). Alpha was set at 0.05. Statistical analyses were performed using Statistical Package for Social Sciences version 26 (SPSS, Inc., Chicago, IL, USA). All data are reported as means (s.d.), unless otherwise specified.

Carotid-femoral PWV predicts mortality in patients with essential hypertension (28) and is a strong predictor of cardiovascular disease in a range of clinical populations (29). Past research has shown significant improvements in cfPWV following 12 weeks of supervised aerobic or resistance training in patient populations when compared to usual care (30). Our study demonstrated that cfPWV decreased by, on average, 0.91 m/s (~12%) in the O-RAGT group at PI, compared to a 0.12 m/s (~1%) in the control group. This is highly encouraging as a 1 m/s reduction in cfPWV is the minimal clinically important difference, and is strongly associated with decreased cardiovascular disease risk (29). Although O-RAGT did not quite elicit this minimal clinically important difference, the statistically significant interaction and the large effect size (ηp² = 0.135) indicates a robust and promising impact on regional arterial stiffness.

In the present study there were no statistical changes in central haemodynamic (PWA) or local carotid arterial stiffness parameters (Table 2; Supplementary Table A). Past research has demonstrated that aerobic training interventions typically elicit reductions in SBP (95% CI) of up to 5.0 mmHg (31), while during a large-scale analysis of randomized trials, a 5 mmHg reduction of SBP following pharmacological treatment reduced the risk of major cardiovascular events by ~10% (32). Accordingly, although not statistically significant, the ~6 mmHg reduction in SBP and cSBP for O-RAGT participants is comparable with prior literature and highly encouraging given the limited mobility of the population and the low-intensity O-RAGT intervention implemented. It is notable that unlike cfPWV, local carotid artery stiffness did not change in response to O-RAGT, but this is perhaps not surprising given that regional measures of arterial stiffness summate a larger portion of the arterial tree (e.g., cfPWV) and thefore may better detect the impact of cardiovascular disease risk factors (i.e., blood pressure and physical activity). Further, it is well-recognized that regional and local measures of arterial stiffness are not always closely associated (12). Although not significant, the average changes in local carotid artery stiffness, compliance and distensibility for O-RAGT particpants between BL and PI were −17, 10, and 16%, respectively. Past research has shown larger changes in common carotid arterial compliance (17%) and distensibility (22%) in people with stroke who engaged in a moderate to high intensity exercise program (33). Woolley et al. (33) also observed reductions in SBP and DBP of 6 and 12%, respectively, and stated that as these changes were concomitant with the reduction in carotid artery stiffness, it may be suggested that reduced blood pressures had greater influence on local (carotid) arterial stiffness than potential modifications to the elastic properties of the vessel.

Long-term outcomes are of utmost importance when evaluating the clinical importance of interventions. An important characteristic of successful behavior change is that individuals continue to engage in lifestyle modifications once the stimulus (i.e., use of the O-RAGT device) has been removed. A recent meta-analysis for stroke patients revealed that end-of-intervention benefits gained from regular physical fitness training do not persist following completion of an intervention (4). In non-stroke populations, some exercise studies have shown that following 1-month cessation of an exercise intervention, PWV values revert back to pre-intervention baseline levels (34, 35). However, in our study the improvement in cfPWV at PI was maintained at 3PI. This finding may be underpinned by the fact that the increase in physical activity (e.g., time spent stepping) observed between BL and PI was sustained between PI and 3PI (Supplementary Table B). For example, participants undertook an additional ~1,700 steps per day at the time of the PI assessment compared to BL (~39% improvement), which was generally maintained at the 3PI assessment. This positive change in habitual activity patterns may have important practical implications for the adoption of over-ground, lower-limb robotic technology in the rehabilitation of stroke patients. As we recruited a chronic stroke population, it will be of interest to see whether similar changes in cfPWV and habitual activity patterns occur when implementing O-RAGT interventions with acute stroke patients (≤ 3 months), and whether such devices are beneficial for individuals who do not receive ongoing rehabilitation.

The encouraging findings surrounding cfPWV is unique as the O-RAGT program focused on walking, a low-intensity activity, with RPEs of 11 to 13 typically recorded in the activity diaries (Supplementary Table B). Past research has often shown favorable changes in PWV when training interventions have prescribed moderate to vigorous volumes of physical activity (36, 37). However, low-intensity exercise may be more achievable and sustainable than higher intensity programs as feelings of enjoyment and wellbeing are strong motives for continued participation (38). Ekkekakis and colleagues' review into the pleasure and displeasure people feel whilst exercising reported that pleasure is reduced mainly above the ventilatory or lactate threshold, but that pleasant percpetions are often observed below such threshold intensities (39), which would likely have been the case in our study. Due to the encouraging findings of the present study and those associated with functional outcome measures (10), measures of enjoyment during and following robotic technology use at low-intensities of physical activity should be monitored in both the short- and longer-term (e.g., 12 months PI), as this type of technology and O-RAGT program could have a substantial impact in aiding the recovery of chronic stroke surivors.

In order to contextualize the present findings, specific limitations must be addressed. Firstly, the small sample size was determined based on a primary outcome measure which was not a focus in this study (6-min walk test) (10). However, an a priori sample size calculation based on the cfPWV reported between groups at PI demonstrated that a sufficienct sample size was recruited (n = 13 per group). Secondly, regional (cfPWV) and local (carotid) measures of arterial stiffness were only investigated on participants' left-side. As the stroke diagnosis (and hemisphere affected) varied between participants (Table 1), the assessment of regional and local stiffness measures on both the right and left-side may have been informative, particularly for those participants for whom the right carotid artery may have been symptomatic. Thirdly, participants were recruited from an independent neuro-physiotherapy practice which could be a determining factor to whether a home-based program is successful. The selected population were likely to be highly motivated to engage in rehabilitation due to the costs associated with engaging in physiotherapy with an independent provider. The total dosage of physical activity in the O-RAGT condition was likely higher than the control condition and could have also been a reason for the observed findings. Finally, findings should be interpreted with caution as multiple analyses inflate the risk of type I error, while researchers responsible for collecting outcome data were not blinded to group allocation. Strengths to the study included the use of gold-standard non-invasice measures of arterial stiffness, the inclusion of a 3-month PI assessment, and the implementation of a home-based exercise program which may have enabled participants to undertake a higher volume of walking as the participants could wear the O-RAGT device at any time or day during the program period. The observed increases in habitual physical activity could help prevent secondary complications associated with cardiovascular disease and future cardio- or cerebro-vascular events (i.e., reducing strokes) if such programs are implemented over the longer-term.

In conclusion, the present study has demonstrated that participation in a 10-week, home-based, O-RAGT program, in combination with weekly, usual care physiotherapy, can elicit greater improvements in regional (cfPWV) measures of arterial stiffness in people with stroke than “usual care” alone. Importantly, the changes reported in cfPWV were maintained at 3PI assessment suggesting this may be a sustainable and efficacious treatment option once access to the O-RAGT device has been removed. Individuals randomized to the O-RAGT program also demonstrated increases in physical activity which could have the potential to improve quality of life. However, larger randomized controlled trials are required to identify whether the use of O-RAGT is appropriate to recommend as a part of usual care, while further research is also needed to determine whether implementing “at home” O-RAGT programs should be a part of the stroke treatment pathway.