Users’ experiences of intensive robotic-assisted gait training post-stroke – “a push forward or feeling pushed around?”

Abstract Purpose Robotic-assisted gait training (RAGT) is suggested to improve walking ability after stroke. The purpose of this study was to describe experiences of robotic-assisted gait training as part of a gait training intervention among persons in the chronic phase after stroke. Materials and methods Semi-structured interviews were performed with 13 participants after a 6-week intervention including treadmill gait training with the Hybrid Assistive Limb® (HAL) exoskeleton. Data were analysed using qualitative content analysis. Results Four categories emerged: (1) A rare opportunity for potential improvements describes the mindset before the start of the intervention; (2) Being pushed to the limit represents the experience of engaging in intensive gait training; (3) Walking with both resistance and constraints reveals barriers and facilitators during HAL training; (4) Reaching the end and taking the next step alone illustrates feelings of confidence or concern as the intervention ended. Conclusions The gait training intervention including RAGT was considered demanding but appreciated. Support and concrete, individual feedback was crucial for motivation, whilst the lack of variation was a barrier. Results encourage further development of exoskeletons that are comfortable to wear and stimulate active participation by enabling smoothly synchronised movements performed during task-specific activities in different environments. IMPLICATIONS FOR REHABILITATION When provided in a suitable context, the mental and physical challenges of intensive robotic-assisted gait training can be both inspiring and motivating. Support and engagement along with informative feedback from therapists are suggested crucial for motivation. Intensive task-specific gait training may preferably be performed in an enriched environment and combined with other physiotherapy treatments to stimulate engagement.


Background
Stroke currently affects about 80 million people worldwide and thus the disorder is viewed as a global health problem [1].Due to impaired neuromuscular and sensory function, 20-30% of persons afflicted by stroke either remain limited in their ability or fail to regain the ability to walk independently in the long term after stroke [2].Additionally, an improved walking ability has been described as one of the most important rehabilitation goals among persons who have suffered a stroke [3][4][5].Gait training is thus prioritised in both clinical practice and research and efforts to develop efficient gait rehabilitation programmes are made globally.Since time is considered a crucial factor in motor recovery after stroke [6], rehabilitation is often initiated and most intense, during the first months after onset.Thereafter, a gradual decrease in intensity is commonly seen.However, growing evidence indicates that intensive and task-specific gait training improves outcome both in the early and chronic phase after stroke [6][7][8][9][10].
The development of robotic-assisted gait training (RAGT) systems is therefore of specific interest as these systems may potentially facilitate the provision of intensive and task-specific gait training in clinical practice [11][12][13].RAGT has been suggested to increase the chance of independent walking ability, specifically in non-ambulatory persons in the early phase after stroke onset [14].Previous reviews that have investigated clinical and biomechanical measures such as walking velocity and capacity (using a 6-min walking distance) following RAGT in different phases after stroke have however found contradicting results.Mehrholz et al. and Nedergård et al., did not find RAGT superior to conventional training with regard to walking capacity [14], velocity [14,15] or biomechanical measures of gait such as cadence or spatial symmetry [15].Bruni et al. conversely reported a superior effect of a specific type of RAGT (training with an end-effector) on walking velocity when compared to conventional gait training [16].The contradictory findings may be explained by heterogeneity across the stroke populations (e.g., severity level), differences between the electromechanical devices and/or contrasting intervention designs (e.g., differing number and intensity of training sessions).
While improved motor control is a product of personal learning, patients' experiences of RAGT are further likely to influence the results of the intervention but are seldom reported [13,17,18].Complementing the gait training with advanced technology generates new possibilities in stroke gait rehabilitation, but may also lead to additional challenges in the rehabilitation process.On the one hand, expectations of overcoming or even curing physical impairments may increase because of the introduced machinery.On the other hand, the presence of novel technology and the equipment associated with it may also give rise to scepticism or even fear [19][20][21].Information on how the wearer perceives gait training with robotic assistance would also enable bioengineers to optimise the design of the device to improve user adoption.Moreover, knowledge and understanding of experienced barriers and facilitators may guide therapists to successfully form and individually adapt gait rehabilitation programmes that include RAGT.
Thus, this study aimed to describe experiences of treadmillbased RAGT as part of an intensive gait training intervention among persons in the chronic phase after stroke.

Study context and participants
This qualitative study was undertaken within the context of a randomised controlled trial (RCT) investigating the additional value of training with a RAGT-system in the long term post-stroke (published protocol on clinicaltrials.gov;NCT02545088) [22].The trial was carried out at the University Department of Rehabilitation Medicine, Danderyd Hospital and Department of Clinical Sciences, Karolinska Institutet in Stockholm, Sweden, and was approved by The Swedish Ethical Review Authority (Dnr: 2015/1216-31).Participants were recruited from primary health care units and through advertising.The physiotherapist at the units reported eligible participants to the study coordinator over the telephone after receiving permission from the participants.Participants were presented with verbal and written information about the study by the study coordinator before giving their informed consent to participate in the study.
For the RCT, eligible persons were individuals who had received conventional rehabilitation because of hemiparesis of the lower extremities due to an ischemic or haemorrhagic first-ever stroke (verified by CT or MRI) 1-10 years earlier.They were still affected with regards to walking ability (Functional Ambulation Categories [FAC] score of 2-4, ranging from an ability to walk on even surfaces with assistance to an ability to walk independently on even surfaces but with reduced speed [<0.8 m/s]).After the baseline assessments, included participants were randomised into the following groups: (A) RAGT combined with conventional gait training, (B) conventional gait training without RAGT or (C) no intervention.For further details on the inclusion criteria and randomisation process, see Palmcrantz et al. [22].
The current study included only participants of group A (RAGT combined with conventional gait training).Participants had been scheduled for 1 training session/day, 3 days/week for 6 weeks.Each RAGT session was not to exceed 1 h and 30 min (including approximately 30 min of donning and doffing the exoskeleton and harness).In addition, each session included conventional gait and mobility training limited to 30 min of training time.The intensity of training was defined as high, following the suggested definition by Veerbeek et al. [8] and based on the number of hours spent in exercise therapy (time/sessions/weeks).Furthermore, the participants were encouraged to walk as fast and far as tolerated during both the conventional and robotic-assisted gait training.
RAGT was performed with a single-leg version of the Hybrid Assistive Limb system V R (HAL).The exoskeleton was attached to the participant's affected leg with straps around the waist, thigh and lower leg and a standardised shoe was mounted on the exoskeleton in a fixed position.HAL assists the stepping cycles through power units providing torque over the hip and knee joint, activated by the wearer's electromyography (EMG) signals (through surface electrodes) [23].The level of support is set and changed continuously by the therapist.HAL was used during treadmill walking in combination with a harness connected to a stationary bodyweight support system, to prevent falls and unburden the weight of the suit (9 kg).HAL settings were individualised for each participant and parameters such as gait speed, as well as time and distance of walking intervals were monitored.These parameters were used to provide feedback on the performance and progression of the patient.The training was led by physiotherapists experienced in stroke rehabilitation and educated in the HAL system and the study procedures.The additional 30 min/ session of conventional gait training focused on gait and mobility and was individually adapted according to current best evidencebased practice.Two participants were excluded halfway through the training period due to unforeseen logistical reasons related to the transportation of the HAL suits.Another two participants dropped out due to medical reasons, leaving 16 participants that completed the training period.
Interviews were performed with 13 of these 16 participants since three participants were excluded due to severe aphasia.Both men (n ¼ 8) and women (n ¼ 5) of different ages (range 37-70 years) and with a range of 13-49 months post stroke (mean 26 months) were represented, as well as persons with different levels of independence in walking (scoring FAC 2-4 according to the inclusion criteria) (see Table 1).

Data generation
In the final week (week 6) of the intervention period, when participants had completed between 10 and 18 training sessions, a semi-structured interview was performed alone with each participant.The interviews were carried out by one of two female physiotherapists, experienced in neurological rehabilitation and in performing research interviews.They were not otherwise involved in the study and had no experience with HAL-training.All interviews were performed after a training session, individually in a quiet and private room at the rehabilitation centre.An interview guide with open-ended questions was used to obtain information regarding participants' expectations, perceived challenges and/or

Data analysis
For the reader to judge the scientific rigour of the findings, aspects of credibility, dependability, transferability and confirmability were addressed during the study planning, analyses and reporting, to strengthen trustworthiness [24,25].The first author (HN) listened to all the interviews and read through the transcripts several times to obtain an overview.The transcriptions were exported to MAXQDA 2020, a qualitative analysis software.
Here the interviews were analysed using qualitative content analysis [26] with an inductive approach to find patterns of similarities and differences in the content, without imposing preconceived categories based on specific theories or hypotheses [25].Meaning units (words, sentences, or paragraphs containing aspects related to each other through content and context) were identified, condensed, and labelled with codes that were short descriptions closely mirroring the text.To address the confirmability, credibility and dependability of the results, the meaning units and formulation of codes were continuously discussed between the three authors that conducted the analysis (HN, SP and MS).
Based on the content, codes were organised into patterns or overarching concepts forming subcategories and, finally, descriptive categories [24].During the analysis, the subcategories, as well as final categories, were frequently checked to agree with transcripts and codes and were discussed between all four authors until a consensus was reached.All authors are physiotherapists representing a broad range of experience related to clinical neurological rehabilitation and rehabilitation research.The authors had no previous experience or bias related to robotic gait training, considered to have influenced the study.For credibility, examples of the abstraction and interpretation processes are provided in the presentation of the results.Quotations from the interviews are presented to illustrate and support the findings including information on the time passed since stroke onset (> or < of group mean) and level of independence in walking (not independent or independent in walking on level ground, presented as FAC score < or ¼ 4).Dependability refers to data stability over time and with regard to different circumstances [27] and establishes the study's findings as consistent and repeatable.The aspect of dependability was addressed by giving a clear and comprehensive description of the context, selection and characteristics of the participants.These factors will also reinforce the transferability of the results.

Results
The analysis resulted in four categories, each formed by three to five subcategories, presented in Table 2.

A rare opportunity for potential improvements
The first category, a rare opportunity for potential improvements, illustrates how the intervention was considered a unique chance.
Participants performed the training up to four years after stroke onset and described how their access to specified stroke rehabilitation had been limited during this time.In contrast, the study intervention offered training with professional support and advanced assistive technology.Physical training was recognized as a key factor for improvements among the participants and the most important component of their rehabilitation.Several participants indicated their continuous effort for improved functional ability and also emphasized the everyday struggles they faced when trying to reach this goal.The absence of improvements despite carrying out hard work during their earlier rehabilitation had for some led to an acceptance of reaching a plateau in the recovery of functioning.These participants still acknowledged the value of physical training but had revised their personal goals to not focus on improvements but to maintain functioning and independence.
It's the first thing you think of when you go to rehab.That you will recover.But in my case, it's about not getting worse.That's basically what it's all about, all the training I do.But it's not that easy, I do understand that.It would be terrible if one would get worse, then it's all over.(P 2, time post stroke > mean, FAC < 4) Even though the participants' experiences of previous rehabilitation and gains in functioning differed, all participants anticipated an opportunity for progress and entered the intervention with a certain degree of hope and optimism.They expressed how they felt lucky and grateful to find out that they had been randomised to the intervention group that combined conventional gait training with RAGT.
I was so happy … There was a drawing of lots on who would be allowed to participate and then she called me one morning and said I would.Oh God, I was so happy.(P 3, time post stroke > mean, FAC ¼ 4) The positive mindset when joining the intervention was however not necessarily without some doubts.Participants' ambivalence seemed mainly to be based on an uncertainness of the

Being pushed to the limit
The second category, being pushed to the limit, represents the facilitators and barriers reported by the participants concerning training on the verge of their abilities.For participants to keep engaged and focused, they needed to cope with the experienced physical and mental challenges induced by the exercises.The physical load was considered very high, both with regard to the aerobic capacity strain and because of the muscle strength required to accomplish a large number of repetitive steps.The intensity of the training, both with and without the HAL suit, was reported as a new experience compared to their former stroke rehabilitation.While the robotic device was mainly expected to facilitate walking performance, the physical and cognitive requirements of RAGT were surprising to the participants.In the beginning, the demands were met with concerns and some participants questioned if they had the abilities required to complete the intervention programme.However, when the participants overcame these challenges, they felt inspired and motivated to continue and exert themselves even more.They expressed an urge to walk faster and longer distances both with and without the robotic device.
The therapists that led and supervised the training made individual adjustments in cooperation with each participant and provided supportive feedback during training.The importance of receiving support from a skilled physiotherapist whom they could trust was emphasized by the participants.To be the focus of attention during the rehabilitation intervention, being acknowledged as an individual and recognised as a person with specific needs, resources and goals, were described as crucial.
… then you play it by ear until you find the speed that you notice is right for me.Then, the speed is increased as you improve, become more stable and can walk faster.They are with me all the time.(P 6, time post stroke > mean, FAC ¼ 4) Participants expressed the need for the therapist to tell them what to do and to help them follow through with the planned training.To be verbally encouraged and urged on during the training sessions was considered important.Several participants illustrated how their physiotherapists used feedback consisting of reported walking time, speed and distance as motivators.New goals were continuously formed according to different parameters, and these stimulated increased physical efforts.
But there is always one of them who keeps pushing me.If I've walked for four minutes, I should walk five.If I've walked five, damn it, I have to walk six too.(P 7, time post stroke < mean, FAC < 4) A perceived barrier during training was the lack of variety.While being task-specific and repetitive, no time or effort was dedicated to other training than gait-specific exercises.Over time, this training felt predictable and less stimulating for the participants.The training was performed in a gym with access to other training equipment and participants questioned why other types of exercises were not incorporated into the training.
… I felt that my inner mental resistance to come here grew.It's so damn boring … one more time … I knew exactly what was going to happen: we walk down the corridor first.Then, we walked three times with this, then we walked three times without this.It was the same, exactly the same schedule.And that's how it was with everything.And � Im the kind of person who gets bored easily if everything is just repeated.(P 1, time post stroke < mean, FAC ¼ 4)

Walking with both assistance and constraints
Walking with both assistance and constraint describes the contradicting positive and negative experiences of training with the HAL exoskeleton.The training required facility-bound rehabilitation 3 days a week.For some participants, getting away from home was perceived as a positive and stimulating break from everyday routines, while others were required to make journeys that resulted in additional fatigue.
With strenuous efforts from the wearer, the robotic device enabled the participants to walk longer distances and at faster speeds under circumstances that paradoxically made them feel constricted in their movements.Preparations before the electromechanical gait training, including donning and doffing of clothes and the HAL suit, were found to be time-consuming.When wearing the equipment, participants expressed feeling restrained by the harness and HAL suit, which was considered heavy and clumsy.The function of the harness was to unload the weight of the exoskeleton.It was however reported as uncomfortable and the participants felt constrained by the straps of both the harness and parts of the exoskeleton.In addition, the specific sensorequipped pair of shoe adherent to the robotic device was commonly considered large and heavy.
Heavy.Clumsy like crazy.Heavy and clumsy.And even if there is bodyweight support, it doesn't matter.It's heavy and clumsy.(P 5, time post stroke < mean, FAC < 4) There are a lot of straps that they tighten around the belly and the chest.You felt a bit trapped and like a parcel.(P 3, time post stroke > mean, FAC ¼ 4) The participants did not consider walking with the HAL system to be the same as walking without it.It was perceived as difficult to walk with a natural flow when cooperating with the robotic device.The device was described as taking control of their gait and as having a "life of its own," which sometimes gave rise to frustration and fear of falling.
Put on two lead shoes, and then someone is pushing you at the back of your knee every other second, while yo� ure supposed to keep stepping and then ther� es a treadmill.(P 4, time post stroke > mean, FAC < 4) However, the synchronisation with the robotic device was perceived to improve over time when the wearer got used to the motions of the device and could adjust their walking to follow the movement pattern facilitated by the exoskeleton.
The RAGT was described as assistive but strenuous.As the level of assistance was adjustable and the progression was individually adapted, the RAGT was not necessarily perceived as less physically demanding when the participant improved.
At first I didn't really get how it worked, but then you became stronger, and you had to do more by yourself: lift the leg forward, the left leg in my case.So now I have to lift everything forward by myself, so to speak.(P 2, time post stroke > mean, FAC < 4) An identified barrier for RAGT was the pain or discomfort expressed by some of the participants.Experienced discomfort was related to difficulties in keeping up with the pace of the robotic device, which resulted in a perceived loss of balance and uncomfortable falls secured by the harness.Reported pain during the intervention was described to originate from, e.g., a skin lesion or a sore and swollen foot.The skin lesion was handled with a bandage and the participant could continue the training.However, the sore foot made it impossible for one participant to complete the training.

Reaching the end and taking the next step alone
The last category, reaching the end and taking the next step alone, describes feelings of reaching the end of the intervention period and having no other choice than to continue the training independently.Perceived improvements included increased strength, balance, and independence in everyday mobility.These positive effects were described as moving on with greater trust in your physical abilities.The experienced improvement in confidence reduced the fear of falling, a factor often described by the participants as a major barrier in daily life.
But I am much more independent than before, thanks to the training now.I have control, I have balance and so I have courage too.That's been lacking for a long time.I've been terrified of falling and have done so a few times as well.(P 1, time post stroke < mean, FAC ¼4) Participants who described improved self-confidence were also motivated to continue training on their own.Transferring some of the intervention's exercises into a daily routine was realistic for those who believed they could manage their training without assistance.
I have learned a lot from walking outdoors here.So I will try to train outdoors at home as well and walk by myself.There are some slopes, and on the other side, there are stairs.But it … I will have to try.Because I fight.I don't give up.(P 3, time post stroke > mean, FAC ¼ 4) Even in the absence of improvements, some participants described how they had appreciated the training as good exercise.
I simply, see it as training.But nothing has happened in the leg.No, but for those who can't move at all, then it's a good start … No, nothing has happened.I've improved my endurance, but that doesn't help much.(P 2, time post stroke > mean, FAC < 4) Importantly, some participants increased their endurance and/ or strength but were disappointed that these improvements did not positively affect their daily life.When the hard work during training did not "pay off," motivation lessened and the participants expressed feelings of resignation and being lost.In these cases, functional improvements were no longer regarded as achievable through physical exercise.Moreover, the participants who did not improve were still dependent on professional help and support to manage their continued training and felt abandoned when the intervention ended.
Yes, I would have liked to continue … Yes, I would.But I guess it's the same as everything within the Swedish healthcare system in the current situation, you just get pushed back out in the cold again.(P 8, time post stroke < mean, FAC ¼ 4)

Discussion
This is one of few studies exploring experiences of RAGT among persons affected by stroke.The RCT in which this qualitative study was included, did not find a significant improvement on a group level, according to the 6-Minute Walk Test after 6 weeks of RAGT combined with conventional gait training in the chronic phase after stroke [22].However, the interviews revealed that the gait training intervention including RAGT was considered by the participants to offer a unique chance for progress.The training was appreciated for providing challenges that pushed the participants to their physical and mental limits and for strengthening them through encouragement, individual adjustments, support and feedback.Lack of variation, pain and feeling restricted in movements were reported barriers to engagement.Difficulties were further experienced in finding a smooth and rhythmical gait when using the robotic device.Even so, the ability to walk faster and for longer distances with the help of the robotic device was acknowledged.Feelings were mixed at the end of the intervention.The participants who had improved in functioning continued their training with greater confidence and determination, whereas those who did not improve felt disappointed and abandoned by the healthcare system.

The opportunity to be challenged and confirmed
A paradoxical reality was reported by our participants where their strive for independence made them dependent on professional guidance and encouragement, a phenomenon which has also been reported in a previous study [28].Our participants acknowledged a need for daily physical training to improve, or at least maintain functioning.However, they had also experienced difficulties in satisfying this need in the absence of external support.Being included in the study with access to training support, as well as to novel technology, generated new hope for improvements.
Fatigue has been identified as a potential barrier to physical activity following stroke [29][30][31], and was also reported as a response to the great mental and physical efforts required by the intervention.For several participants, however, the motivators seemed to counterbalance the impact of fatigue as a barrier to engagement.In fact, the demanding gait training (both with and without the robotic device), was not only accepted but also appreciated by the participants.Being pushed to the limit, both mentally and physically, was perceived as inspiring.This appreciation of highly intensive training is in line with previous results of several trials and reviews [29,32,33], and confirms that a challenging exercise programme can be implemented and valued in a stroke population when offered in a well-considered context.
Rehabilitation programmes, in general, should stimulate engagement and be person-centred [29].Physical as well as psychological support has been suggested as crucial [34], along with effective communication [29] and individualised goal setting [32].Good communication and a positive relationship between the participant undertaking therapy and the therapist have been suggested to foster hope, motivation and greater satisfaction during rehabilitation [29,32,35].All of these aspects were reported by our participants and the relationship between the participant and therapist was even described as being more imperative for a positive experience than the actual content of the training.
A lack of variation in the training was a criticism highlighted by our participants.Boredom due to a great number of movement repetitions and training consisting of similar exercises seemed to reduce motivation to engage in training.To stimulate motivation, task-specific training has been recommended to be variable and performed in an enriched environment [36].Nevertheless, the repetitive characteristics of this type of training enabled monitoring of well-defined gait parameters, such as gait speed, walking intervals and distances.Participants described how both feedback and goals were commonly defined in terms of these parameters, which made it easy for them to follow their progress and identify even small improvements.Identification of gains has been suggested as a powerful tool to enhance engagement in a rehabilitation intervention [33,37], and reporting of concrete and sensitive reference points is thus an important factor that strengthens the value of this type of training intervention.
This study, therefore, identified core aspects to consider when providing rehabilitation interventions long-term after stroke.
Intensive training to improve limitations in walking may be appreciated and well tolerated when led in alliance with a physiotherapist who individualises a variable training intervention.Frequent and timely feedback using easily available parameters can also elucidate progression towards achievable goals in a way that is comprehensive to the participant.

The impact of a technological device
A central issue reflected across the categories was technology.As described above, the use of novel technology generated hope while uncertainty related to not knowing what to expect, generated anxiety.Walking with the HAL system was perceived as both assistive and constraining at the same time.Donning and doffing of the robotic device was considered time-consuming, which has been reported earlier [38].In combination with the high physical and mental demands related to the training, this was a factor that increased the risk of fatigue.In line with the previous reporting [39,40], some discomfort was described due to the tight straps and the perceived heavy weight of the HAL device.The equipment further elicited feelings of being physically restricted, sometimes even causing pain among the participants.
Active patient participation during training has been advocated of certain value for potential recovery after stroke [11,41].To optimize this, technologies for movement intention detection are commonly used in exoskeletons.The concept of HAL involves active wearer engagement through the facilitation of joint movements triggered by the wearer's voluntary muscle activity, as recorded by bio-electrical signals (EMG) [42].Some limitations however remain to be overcome concerning the vulnerability of these systems [19].Participants in this study described difficulties in terms of finding a rhythmical and balanced gait.The device was described as an exterior skeleton taking over the gait and making the wearer lose control.This conflict has also been revealed in other studies exploring the experiences of RAGT [43,44].While anxiety and perception of lost control are suggested to reduce active training engagement [45], the perception of feeling safe and in control while performing RAGT is of specific importance and should be prioritised.The gait movement pattern and flow during walking with the gait-assisting robotic device are dependent on the interaction between the wearer and the robotic device [46].In our study, participants found it gradually easier to walk with the HAL suit.They experienced a higher degree of cooperation and control resulting in a smoother, more rhythmical gait.A challenge for future development includes streamlining/ tailoring the device and associated equipment to be more easily applicable, feel safer, be more supportive but less restrictive, and enable a high level of active participation from the wearer.

To continue alone
When the intervention had reached its end, the response from participants varied.They felt proud to have completed the intervention but were either relieved that it was over or disappointed for losing their professional support and once again, the role of the physiotherapist was emphasized.
Physical capacity has been suggested to reinforce the perception of self-efficacy, which in turn seems to increase perceived physical competence and improve the level of self-esteem [33,47] and maintenance of physical activity [48].This was confirmed in our study, in which increased confidence and a positive attitude towards the future were expressed by the participants who had experienced improvements in functioning.
Our study also involved participants who confirmed improvements during training (e.g., greater balance or strength), although did not consider these gains relevant if they were not transferable to their daily life.This might indicate a discrepancy between the participants' expectations of recovery from intervention and the perceived goals of therapy [33], something which has also been seen in a similar study exploring the experience of RAGT among persons with spinal cord injury [20].The feeling of not improving towards your personal goals despite hard effort and commitment, was by our participants associated with decreased self-confidence, resignation and loss of hope.Participants remained dependent on the support from healthcare professionals and expressed feelings of being abandoned and let down by the healthcare system when the intervention period was over.These expressed feelings highlight the perceived need for continuous support from healthcare professionals to improve impaired motor function that limits activity performance long-term after stroke.

Methodological considerations
Several participants had affected communication skills following their stroke.In combination with the point in time of the interview (directly after a training session), this might have increased the risk of fatigue and affected the comprehensiveness of the answers.
Dependability refers to data stability over time and with regard to different circumstances [27] and establishes the study's findings as consistent and repeatable.The aspect of dependability is addressed by giving a clear and comprehensive description of the context, selection and characteristics of the participants.These factors will also reinforce the transferability of the results.Participants were both men and women of different ages, with some variations in functional abilities and time since stroke.All participants were however able to walk shorter distances independently or with minor assistance.The results might not be transferable to other stroke populations, e.g., those that are not able to walk at all.Furthermore, the participants were in a longterm phase after stroke, which possibly implies other facilitators and barriers than the ones experienced in a subacute phase.Participant recruitment was achieved either through advertisement or contact with primary care due to their functional impairments.These circumstances imply a positive and active standpoint, which also might affect the results of this study.In addition, this study was performed in one rehabilitation centre with the use of one specific exoskeleton, namely the HAL system used on a treadmill.Several gait-assisting devices are however available and the experience of walking with electromechanical assistance may differ depending on the device used for training.

Conclusion
Intensive gait training, both with and without robotic assistance, was considered by our participants to be demanding but appreciated.The support and engagement of the physiotherapists along with direct and informative feedback on performance were crucial for motivation.A barrier to engagement was the lack of variation.This barrier may potentially be overcome by using a robotic suit and safety device that allow safe overground walking in an enriched environment.Future development of gait-assisting robotics should also include tailoring the device and associated equipment to facilitate preparation and improve comfort for the user.The device should stimulate a high level of active participation, smoothly synchronise with the intentions of the wearer and enable training in different environments without renouncing necessary physical support and safety.

I
'm going to get well.No doubt about it.� Ive been training two hours a day since it happened two years ago.(Participant (P) 1, time post stroke < mean (26 months), FAC ¼ 4)

Table 2 .
Categories and subcategories.operation of, and technology behind, the robotic device.Because the HAL system was new to them, they also expressed difficulties in foreseeing the possible effects of training with its assistance.I don't know what I expected.Because I've never seen anything like it.No, so I had no expectations.Well, yes, I thought: oh, what if it can help me walk?(P 4, time post stroke > mean, FAC < 4) So it's really demanding … I have trained a lot during my life but this is at maximum really, it drains the body of all energy.(P 1, time post stroke < mean, FAC ¼ 4) Furthermore, the training was also perceived as mentally demanding as concentration was necessary when coordinating movements and maintaining balance during different gait training tasks.I walk with this leg, I need to get my foot up, by putting down the heel first and so on.And this is probably what's making it so tough.And then I have to think about what � Im doing.You know … if I stand and let my mind wander it won't work.I have to focus, all the time.Because otherwise, I'll stumble and so.(P 5, time post stroke < mean, FAC < 4)