Details of the methods, including participant eligibility, rehabilitation equipment, and example programs, are available in our previous publication [15]. Here, we describe the key elements of the methods according to the CONSORT (Consolidated Standards of Reporting Trials) guideline extension for reporting pilot and feasibility studies [16]; the checklist is provided (omitting the randomization protocol) in Multimedia Appendix 1.

This study was approved by the University of Strathclyde ethics board (UEC20/08).

This is a feasibility study of a novel, technology-based, rehabilitation intervention in a group of chronic stroke survivors. Feasibility was assessed through recruitment, retention, attendance, and adherence to the program, safety (incidence and nature of adverse and serious adverse events [AEs]), and participant acceptability using a mixed methods approach including semistructured interviews, attendance, activity, and safety records.

People living with stroke affecting their mobility or communication but otherwise well enough for light or moderate exercise were invited to participate. Recruitment was through a network run by a medical charity for stroke in Scotland. Individuals expressing an interest in participating registered with the charity and were invited, in the order they registered, to attend an initial meeting where eligibility was assessed and baseline measures of function recorded.

The intervention was developed from our previous work [15,17-19] and feedback from a pilot with chronic stroke survivors (n=5; mean age 51.6, SD 12.1 years; mean 19.6, SD 9.32 months post stroke; 2 females; 2 with aphasia). The small pilot sample size and limited attendance (twice weekly) were related to COVID-19 restrictions in place at the time. Feedback from these participants, through independent interviews, allowed us to implement changes to the intervention, most importantly this included an increase in the number of weekly available sessions from 2 to 5.

The resulting 8-week long rehabilitation intervention was delivered entirely through technology, including virtual reality (immersed and nonimmersed), treadmills, weight suspension and movement resistance, and assistance equipment located in a gym-like space on a university campus (Glasgow, UK). Individual programs were designed, supervised, and reviewed by a physiotherapist using principles of intensity, feedback, cognitive engagement, and aerobic activity [20] to address the goals identified by the participant and scores from outcome measures at baseline. An example program is detailed in a previous publication [15]. Participants were encouraged to use the exercise equipment on their own, wherever possible, while being supervised and to make alterations to the program, with support from the therapist.

Feasibility was assessed by rates of recruitment, adherence, AEs, and participants’ perceptions of acceptability from semistructured interviews [21] (see Multimedia Appendix 2 for interview schedule). To reflect the multidomain nature of the intervention, a range of outcome measures were included: 10-meter walk test (10mWT), five times sit to stand test, action research arm test, functional ambulatory category, Rivermead Mobility Index, and the Stroke Impact Scale-16 (SIS-16) [22-26].

Participant interviews were analyzed using the 6-stage thematic approach described by Braun and Clarke [27]. Initially, an independent researcher generated codes and candidate themes. These were then reviewed by 2 members of the research team. Through an iterative process of discussing and revising, a consensus was reached. Descriptive statistics were used to assess feasibility (recruitment, retention, adherence, and safety) and outcome data.

Between August 2021 and August 2022, 70 individuals registered their interest in participating in this study. The first 50 were invited to attend an initial meeting to assess eligibility, 39 attended, and 31 met the criteria and consented. In total, 8 individuals were not eligible due to conflict with ongoing rehabilitation (n=2), currently unwell or in pain (n=2), unable to attend at least twice a week due to lack of transport (n=2) or other reasons (n=1), and other (n=1). The first 5 recruited participants participated in a pilot of the intervention with the next 26 participating in the feasibility study. A participant flowchart is available in Multimedia Appendix 3. Since the program continues to be supported through charitable funding, the 20 individuals still on the register will be invited to participate in future cohorts. Full details of the sample, separated into 3-phased participating cohorts, recruited to this feasibility study are provided in Table 1.

All participants set individual goals in partnership with a physiotherapist, including the number of weekly sessions. A total of 493 total sessions were attended representing 986 hours of therapy. In total, 5 individuals achieved, or exceeded, their target number of sessions, and there was, overall, an average adherence rate of 82% (number of attended sessions or number of sessions planned). In total, 21 participants missed a total of 91 (18% of total) planned sessions for the following reasons: illness (n=13), hospital appointment (n=4), weather (n=15), work (n=5), vaccination (n=5), holidays (n=8), personal (n=24), child care (n=7), and transport (n=12).

No serious AEs were reported during this study. There were, however, a number of AEs reported (n=19) considered to be related to the study: joint or muscle soreness (n=6), viral illness (including COVID-19; n=5), cardiovascular (dizziness; n=3), fatigue (n=3), and skin irritation (n=2). These all resolved within 1 week without intervention.

All participants were able to complete the measurements taken before and after the program. Consistent with previous studies [28], there was considerable variation in group characteristics. In general, there was a positive effect on scores of physical ability with all outcome measures showing a mean improvement (Table 2). Differences were, however, not tested for statistical probability since this study was not set up for this reason; instead, they are reported here as 95% CI and effect size to allow sample size estimations for future studies.

This study assessed the feasibility of a novel model of stroke rehabilitation designed to deliver evidence-based stroke rehabilitation through the scalable model of a TERG and a self-management, supervised, approach.

The participant’s variability in age (SD 13.3 years), cognition (Montreal Cognitive Assessment: SD 23.1), communication (n=10, 38% aphasic), severity of motor impairment (action research arm test: SD 23.1 and 10mWT: SD 34.1), and overall impact on their lives (SIS-16: SD 9.6) reflect both the heterogeneity of this population [29] and the broad inclusion criteria. The findings can therefore be applied to the general stroke population with some confidence, albeit with the limitation that participants needed to be medically well, a criterion that excluded 2 potential individuals.

The intervention can be considered feasible within this highly variable population. Program adherence was generally good at 82%, with an average attendance of 2.4 sessions per week, and only 1 participant dropping out completely for transport reasons. Adherence to rehabilitation programs, in general, is low, ranging from 40% to 71% [30] but may be higher among stroke populations when offered in a structured manner, for example, through exercise facilities such as gyms (eg, Reynolds et al [31] report 81% adherence) or when technology is included, Valenzuela et al [32] reported 91% adherence to a technology-based exercise program in older adults. The number of AEs could be considered high (11 participants reporting 19 AEs) but should be seen in the context of the 493 total sessions attended (986 hours), 1 AE every 51 hours, and the minor nature of the AEs, many of which related to joint and muscle discomfort that could be explained by an increase in exercise and the viral illnesses, which should be seen in the context of the contemporaneous COVID-19 pandemic. AEs are relatively common in the poststroke population, Ostwald et al [33] reported 50% of 159 patients tracked after stroke experienced at least 1 AE in the first year poststroke.

Recruitment for this study was managed by a partner organization and considered broadly successful without the need to specifically advertise or promote the center. In trying to achieve 10 people for each group, 11 individuals were targeted for the assessment sessions (allowing for 10% attrition); however, across all the groups, 8 potential recruits were deemed ineligible due to an inability to attend frequently enough (transport and other reasons) or current pain or illness that prohibited use of the equipment. This finding suggests 2 improvements for future studies: clearer information at the start of the process (when registering interest) and an increase in the number of people invited to the baseline assessments to ensure that 10 participants start the program.

These positive findings of feasibility are balanced against continued reports of barriers to access. Although daily attendance was possible (40 sessions available in total) and encouraged, no participant achieved this; 33 was the highest number of sessions attended by a single participant. While unmodifiable barriers (illness, weather, national holidays, and personal) account for at least some of the issues around fully accessing the program, the lack of transport was mentioned frequently and was also reported in the interviews. This is consistent with previous reports of environmental barriers (including transport) to physical activity in stroke populations [34]. The TERG was situated in a city center campus which, for some, meant relatively long and costly travel arrangements that likely limited participation. Our plans to establish the TERG model in community locations could resolve some of these difficulties and have been strongly recommended by the World Health Organization [35].

Motor impairments have previously been reported as barriers to using equipment for exercise or physical activity participation [36]. Reassuringly, in this study, this was only mentioned in relation to grip, suggesting the existing adaptations to the TERG equipment enabled broad participation.

While this study was not designed to test efficacy, positive change to all outcome measures is worth noting, in particular, the changes to gait (mean reduction of 10.6 s in 10mWT) and sit to stand ability (mean reduction of 8 s in the five times sit to stand test) and the moderate effect sizes (0.74 and 0.60) estimated for Rivermead Mobility Index and SIS-16, respectively. The high variability in the change data (eg, the SD for change in 10mWT was 27.5 s) further demonstrates the variability of response to rehabilitation in this population that merits further investigation to understand explanatory factors. Despite this variability, improvements compare well to rehabilitation interventions in stroke [37] and suggest that greater improvements may be possible during the subacute phase and that time since stroke should not be seen as an exclusion factor to this kind of rehabilitation program.

A number of limitations should be considered when interpreting these findings. In particular, the lack of a comparator group means that any change recorded in physical ability may relate to natural recovery or a Hawthorne effect [37] and not the intervention. The chronic nature of the participants, however, suggests natural recovery is likely to be a small part of the positive response.

A greater issue, for interpreting feasibility and effect size, is the recruitment process, which is likely to be biased toward individuals with a pre-existing motivation and interest in rehabilitation. While this cannot be avoided in the context of research or ethics governance, it may mean that metrics like recruitment, retention, and adherence may not be as positive in the real world.

No cost analysis was performed on the intervention. This is recommended for future studies but should include health and societal benefits, including a return to economic and social activity, where appropriate.

Based on the findings of this feasibility study, a number of recommendations are suggested for further study and development: (1) establishing community versions of the TERG to resolve access barriers, (2) statistically powered randomized controlled trial of efficacy, (3) health economics analysis of the intervention, and (4) adaptable gripping systems for exercise equipment.

A novel approach to stroke rehabilitation using a TERG with professional supervision is feasible, with 82% attendance across almost 1000 hours of delivery and with only minor AEs reported. Reassuringly, the intervention was overwhelmingly well received by this diverse group of chronic stroke survivors. This approach has the potential to meet the overwhelming need for greater access to effective rehabilitation but requires an experimental approach, with a statistically powered sample, to confirm the early promising findings.