Exoskeleton for post-stroke recovery of ambulation (ExStRA): study protocol for a mixed-methods study investigating the efficacy and acceptance of an exoskeleton-based physical therapy program during stroke inpatient rehabilitation

The RCT will be conducted at three rehabilitation hospitals, GF Strong Rehabilitation Centre (Vancouver, Canada), Glenrose Rehabilitation Hospital (Edmonton, Canada), and Parkwood Institute (London, Canada). Participants will be recruited from each respective inpatient stroke rehabilitation units over a period of up to three years.

Consecutive patients with subacute stroke admitted at each rehabilitation hospital will be identified by their treating physiatrists and therapists to be screened for eligibility by a member of the research team. Individuals will be included if they: 1) are within 3 months of stroke onset (ischemic infarct or intracerebral hemorrhage); 2) have one-sided hemiparesis; 3) are 19 years of age or older; 4) are able to understand and follow directions in English; 5) are able to communicate (verbal or physical yes/no indication); 6) are cleared to participate in physical therapy; and 7) require significant assistance (maximal assistance from one or two people) to walk. Individuals will be excluded from the study if they have: 1) a significant musculoskeletal or other neurological condition; 2) cardiovascular contraindications to exercise; 3) co-morbidities that would preclude activity; 4) or pain which is intolerably worsened with exercise. Individuals will also be excluded if they have any contraindications to using the robotic exoskeleton (pregnant, height/weight restrictions). The site coordinator at each hospital with obtain informed consent from potential trial participants.

Participants will be randomized by the site coordinator after the baseline assessment at a one-to-one ratio to either the Exoskeleton group or Usual Care group using an online, third-party, permuted block randomization service (www.​randomize.​net, Interrand Inc., Ottawa, ON). As such, randomization will remain concealed until group allocation. Randomization will be stratified by site, to control for differences in standard of care (e.g., frequency and duration of physical therapy treatment, length of stay, rehabilitation and admission discharge timing, etc.). Participants will also be stratified by physical function, as baseline functioning is an independent predictor of outcomes such as community mobility and discharge destination [8, 26]. Specifically, participants will be stratified using their baseline Berg Balance Scale score at enrolment, as it has been shown to be correlated with improved walking ability after robot-assisted gait training [27]. A cut-off score of 12 will be used to stratify participants, based on a study [28] which identified a score of 12 at rehabilitation admission to be predictive of regaining unassisted walking after four weeks.

The EksoGT powered robotic exoskeleton (Ekso Bionics, Richmond, California, USA) will be used to provide the experimental intervention for this study. This exoskeleton has bilateral motor-actuated joints at the hip and knee, as well as a spring-loaded articulation at the ankle to support toe-off and foot clearance during gait via a footplate. The EksoGT is able to power the user’s lower limbs in a walking pattern autonomously (without any active participation by the user), as well as with varied assistance to accommodate any force contribution by the user. The device software allows the therapist to control the degree of assistance, the parameters of gait (step height, step length, swing speed, etc.) and the automaticity of walking (how each step is triggered). These settings can be programmed to tailor the gait training to the individual to ensure active participation that is appropriately challenging. Guidelines for programming the device software to progress gait training with respect to robotic assistance are available in Additional file 1. The device does not provide balance support, and so the user is responsible for maintaining balance and shifting their weight appropriately.

Participants in the Exoskeleton group will have 75% of their standard physical therapy sessions replaced with exoskeleton-based gait rehabilitation. For example, 3 out of 4 weekly physical therapy sessions, or 45 min out of every 60-min session, will be dedicated to the exoskeleton intervention, reserving 25% of therapy time to be dedicated towards other goals. Participants in the Exoskeleton group will wear a powered robotic exoskeleton for their physical therapy sessions beginning after the baseline assessment to allow for repetitious stepping and walking practice from early in their rehabilitation stay. Training will be safely progressed, as tolerated, to reduce the amount of assistance provided by the exoskeleton and to increase the duration of continuous walking bouts. Guidelines for training progression can be seen in Table 1, and specific device programming is available in Additional file 1.

An algorithm will guide clinicians in deciding when to discontinue daily exoskeleton training (Fig. 2), as it has been found that therapist-guided overground walking practice is equally or more effective than electromechanically-assisted gait for improving walking function once people with stroke are ambulatory [18, 19]. Once a participant reaches a functional threshold wherein they are able to walk for an extended period of time with only minimal assistance, therapists may begin to substitute daily exoskeleton training time to overground gait training. If a therapist chooses to fully discontinue use of the exoskeleton, they will still be required to focus on gait retraining for 75% of their weekly physical therapy time.

Participants randomized to the Usual Care group will receive standard physical therapy care during their rehabilitation stay. Standard of care differs between sites, but typically involves 30–60 min physical therapy sessions, 4–5 days a week. No specific instructions will be given to therapists providing therapy in the Usual Care group, except that they cannot use the robotic exoskeleton. Generally, physical therapy during stroke rehabilitation is provided with patient-specific goals in mind, and typically places a large focus on mobility and gait training. Participants in both the Exoskeleton and Usual Care group will be monitored twice a week using an activity tracker (activPAL3 micro, PAL Technologies, Glasgow, UK) to observe the amount of upright standing and walking performed in the physical therapy sessions per group.

All participants will be assessed at recruitment (baseline), at discharge or after 8 weeks of the intervention, and at 6 months by an assessor who is blind to group allocation. The exoskeleton intervention will be discontinued after 8 weeks, and standard physical therapy will be provided to all participants beyond 8 weeks for whom it is deemed appropriate by their care team.

The primary outcome will be walking ability, measured using the Functional Ambulation Category (FAC) [29]. This is a 6-item scale designed to classify the level of physical support required by subjects in order to walk safely over 10 ft, extending from 1 (unable to walk without the assistance of two people) to 6 (independent walking overground on uneven surfaces and on stairs). It has been shown to have good test-retest reliability and validity in the hemiparetic stroke population [30]. The FAC is also responsive to change within the first four weeks post-stroke and up to six months post-stroke [30]; unlike other walking measures of speed or distance, a value is assigned for the FAC even if the participant is not yet independent in walking.

The secondary outcome measures will assess stroke impairment, walking performance (speed, endurance, daily step count), balance, cognition, and quality of life. The secondary outcome measures and schedule of data collection are listed in Table 2.

All sites will report minor and serious adverse events that occur from baseline through to the 6-month follow-up. Two expert physiatrists will review reports outlining adverse events, if they arise, annually, to advise on trial continuation.

A total of 20 participants will be enrolled in each group (total same n = 40). This sample size was calculated using Stata Software (version 11, StataCorp, USA) and assumes a 2-point between-group difference in the Functional Ambulation Category at the end of the intervention [40], setting power at 80% and level of significance at 0.05 (2-sided). This calculation also assumes a standard deviation of 2.0 based on stroke inpatient FAC data from a study by Mehrholz et al. [30]. This between-group difference is realistic, given that participants are 2-person assist (score of 1 on the FAC) on enrolment, and it is expected that those in the Exoskeleton group will make greater improvements in walking ability (independent) compared to those receiving standard care (assistance or supervision required).

Descriptive statistics will be used to summarize data. An analysis of covariance (ANCOVA) will be performed to detect post-intervention differences between groups for the primary and secondary measures, using the respective baseline score as the covariate [41]. The significance level will be set at 0.05, and all statistical tests will be two-tailed. Participant data will be analyzed on an intention-to-treat basis, and any missing data will be assessed and analyzed as appropriate using multiple imputation [42]. For measures without a baseline score (i.e., 5-Metre Walk Test, 6-Minute Walk Test), an analysis of variance (ANOVA) will be employed. Confidence intervals (95% CIs) will be reported, where applicable.

This study will be based in a postpositivist paradigm [45], assuming that exoskeleton users will have individual yet relatively patterned experiences and perceptions of the device. According to this paradigm, conclusions that are drawn regarding the acceptability of using a powered robotic exoskeleton can be generalized to other stroke rehabilitation sites regardless of the social contexts of the researcher and participants of this study.

Participants randomized to the Exoskeleton group in the quantitative RCT will be recruited from all sites to participate in qualitative interviews. All participants who undergo at least five training sessions in the exoskeleton will be invited to participate, provided they are able to communicate fully (English proficiency, non-severe aphasia). It is expected that 10–15 participants with stroke will be thus be eligible and interviewed, during which time data sufficiency will be reached and identified themes would not need to be adjusted by further data collected [46, 47].

All physical therapists who have been fully trained to use the exoskeleton device and who provide the intervention for the RCT will be invited to participate in qualitative interviews. Five to 10 physical therapists are expected to be eligible and participate.

Individual semi-structured interviews will be conducted in a private office with participants with stroke as well as with physical therapists from the RCT. Interviews will last approximately 30 min and will be conducted by the same researcher, whether in-person or by telephone, to maintain consistency. Interviews will be audio-recorded.

The semi-structured interview guides were developed by the lead author and reviewed by two physical therapists and two physiatrists (Additional file 2). Development of the interview guides was informed by the Unified Theory of Acceptance and Use of Technology (UTAUT) [48]. Questions explore user perspective towards the usage (fitting, duration, frequency, etc.) and utility (efficacy, perceived benefits, drawbacks, etc.) of the device.

All interviews will be transcribed verbatim and analyzed using thematic analysis [49]. Transcripts will be read and re-read to develop ideas and interpretations about recurring, converging, and contradictory patterns. Once familiarized with the transcripts, raw data will be inductively coded by two investigators, then iteratively conceptualized into broad categories which will eventually be grouped into relevant themes to provide an understanding of how the exoskeleton is perceived by patients and therapists who use the device.

Drawing from the postpositivist criteria of credibility, transferability, dependability, and confirmability [50], various strategies will be employed to ensure the trustworthiness of this qualitative component of the study. Triangulation of multiple perspectives towards the exoskeleton by interviewing both individuals with stroke and physical therapists will promote the credibility of the qualitative findings. Furthermore, combining two research methods is another method of triangulation which will add depth and rigor to the study. The qualitative findings will provide context for the interpretation of the quantitative trial results.

Research reflexivity will support the transferability of the qualitative findings. By ensuring that the positioning of the authors and personal assumptions are accounted for in conducting the study and made known in reporting, readers will be able to determine the extent to which the findings can be generalized to their own context [50]. A reflexive journal will be kept in order to facilitate reflection on any assumptions, power differentials, and interpersonal dynamics that may arise during interviews that may influence data collection and analysis, which pertains to the dependability and confirmability of the qualitative methods [50].

Finally, negative case analysis and participant checking will enhance the credibility of the qualitative analysis. By exploring divergent perspectives during interviews and paying attention to opinions counter to the majority during analysis, we will develop a greater depth and understanding of the experience of using a powered exoskeleton for stroke rehabilitation. Bringing findings back to participants to ensure the analysis corresponds to their original account will help to ensure richness and accuracy of the findings. Synthesized analyzed data and resulting themes will be presented to participants in a document, written in non-scientific wording. Participants will be asked if the findings match their experience, and if they would like to change or add anything; any added data will be cross-referenced with existing codes and integrated into the analysis [51].