Real-Time Decision-Making on Whether to Incorporate Neurorehabilitation Technology: Informing Developers on a Day in the Life of a Therapist

Courtney Celian Shirley Ryan AbilityLab https://orcid.org/0000-0003-4471-020X Veronica Swanson University of California Irvine Henry Samueli School of Engineering Maahi Shah University of Illinois at Chicago Caitlin Newman Shirley Ryan AbilityLab Bridget Fowler-King Shirley Ryan AbilityLab Sarah Gallik Shirley Ryan AbilityLab Kaitlin Reilly Shirley Ryan AbilityLab David J. Reinkensmeyer University of California Irvine Henry Samueli School of Engineering James Patton University of Illinois at Chicago Miriam R. Rafferty (  mrafferty@sralab.org ) Shirley Ryan AbilityLab https://orcid.org/0000-0002-3182-0314

therapy or have cognitive de cits which inhibit their participation (4). Clinicians question the effectiveness strength and clinical necessity of the device (4). Within the clinical setting, devices sometimes are too large and bulky to adapt use within an organization (11). Clinician use is also in uenced by institution facilitation of use, organizational culture and intention of use (2).
Outside clinical setting barriers also exist when a device is unavailable to the patient post-discharge (10).
Clinicians are the gatekeepers to promote the implementation of new interventions (12). The process for adopting RT into the clinic must undergo intense scrutiny before uptake including the clinical applicability, cost-bene t analysis, and safety of the device (13).
Therefore, it is vital to determine the gaps between the theoretical bene ts and the practical application of such RT that would enable clinician uptake. Several previous studies have used survey methods (10,14) or focus groups (4) to identify these gaps, but such approaches may not fully capture the real-time, pragmatic decision making that therapists must engage in during treatment sessions. Our approach here combined implementation science methodology to help make research more generalizable. Our premise is that integrating implementation science with neurorehabilitation engineering can accelerate the future integration of novel RT.
Our purpose is to describe clinician decision-making around incorporating RT into treatment sessions to improve understanding of clinician uptake, the critical step to device implementation. To provide a window into clinician decision-making during a typical treatment session, we had OTs and PTs write vignettes describing a treatment session, along with their thought processes. Then we synthesized the vignette data using an implementation science framework, the Consolidated Framework for Implementation Research (CFIR), a common implementation framework used to classify the determinants (barriers and facilitators) of successful implementation (15). From our qualitative data analysis, we were able to pinpoint several constructs mentioned in the vignettes to highlight the hurdles encountered by therapists in treatment sessions. Presenting and synthesizing vignettes will help engineering audiences to understand the practical application of the devices they develop and how to improve the success of future RT.

Context
The Shirley Ryan AbilityLab agship research rehabilitation hospital includes interdisciplinary inpatient acute rehabilitation, outpatient rehabilitation, and research infrastructure. It differs from many interdisciplinary inpatient acute rehabilitation facilities in two key ways: (1) the facility was built with planned integration of state-of-the art RT in clinical areas and (2) research infrastructure is embedded in the clinical areas. The environment strives to be conducive to RT use through greater availability and integration of clinician researchers. Within this technology-encouraging context, we ask the questions of how RT is chosen for use, or if it's chosen at all.

Vignette Development
We asked three OTs and two PTs to write vignettes sharing (1) their decision-making process related to RT use, (2) describe their comfort level with RT and (3) provide 1-2 examples of a recent clinical treatment session, focusing on their clinical reasoning behind their decision to use, or not use, RT. We provided the following vignette template: I am a PT/OT … I am a (describe yourself --technophile, technology early-adopter, skeptic, techno-phobe, or other) and normally use technology when… The barriers to my access to technology are… A patient with this diagnosis… and characteristics... Their goals were…. I had a ____ min session, I opted to do these interventions (tools/technologies) …. because…. It took me this much time to set up… I provided these instructions… The patient responded in this way… I chose not to use tools because… It worked/did not work because...

Analyses
We used deductive qualitative analysis to identify codes in the provided vignettes related to barriers to RT use and knowledge translation identi ed in literature (10,14,16). We named these barriers using the CFIR framework, which explains 39 implementation constructs across 5 domains. These constructs can be barriers or facilitators, making implementation more or less di cult, respectively (15). The codebook (Table 1) contained 15 original CFIR constructs identi ed in prior research (10,14,16). Two constructs were added to distinguish between the attributes, knowledge and beliefs of clinicians compared to patients. Three reviewers coded each vignette in their entirety, but the vignettes are presented in a summarized form to follow the template more concisely and provide novel information. The full, unedited vignettes are available upon request. Summative content analysis included used the total number of codes presented, and the proportion of times each code was used across clinicians and vignettes (17). This qualitative analysis plan provided a systematic method to synthesize the vignette results.

Results
The constructs, their de nitions, and results of summative content analysis are presented in Table 1.
Nine vignettes provided by ve therapists detail experiences with patients with the following diagnoses: traumatic brain injury (n = 2), SCI (n = 1), stroke (n = 4), and multiple sclerosis (n = 1). Six vignettes were provided by OTs. Three vignettes were provided by PTs.
All therapists have at least 4 years of clinical experience and have assisted with research projects in the past. The 17 codes (listed in Table 1) were applied 174 times. Most statements were coded with one code (n = 91), but when all three coders agreed, other statements were either double (n = 31), triple (n = 6) or quadruple (n = 3) coded. Each code is presented with exemplar quotes in Table 1, providing examples of the barriers and facilitators to implementation.
Content analysis of the 17 codes resulted in all ve clinicians making statements coded as relative advantage, personal attributes of the patients, clinician knowledge and beliefs of the device/intervention, complexity of the devices, and overall implementation climate. Furthermore, all nine vignettes included statements coded as complexity, relative advantage, and personal attributes of the patient.
Occupational Therapy Vignette 1 I would describe myself as a skeptic when it comes to consistent use of technology to treat the arm post neurologic injury. As an OT, we are taught "function, function, function!", and this sentiment is reinforced by our professional organizations and therapy leaders. Patient goals and function inform treatment plans as paid for by Medicare/Medicaid, and treatment must improve functional Quality Reporting Measures (i.e. the ability to complete toilet transfers, dressing, grooming, etc.). It is challenging to connect devices with functional improvement so I sparingly choose technology.
The patient was a 56-year-old male with left-sided (dominant) weakness after stroke. He had little distal active movement, was unable to participate in electrical stimulation because of the presence of a cardiac event monitor. His goals were feeding, reaching to cabinets, and managing medications, and he wanted to return to living alone after discharge. He was a "good candidate" for technology because young, motivated, and willing to try. However, the goal areas and need to be independent upon discharge swayed my decisions to use a mirror therapy protocol. The patient quickly progressed within 2 weeks to participate in a high repetition training. At this point technology such as an exoskeleton or smart glove could have been selected, but patient goals pushed me towards functional repetitive practice. We used forks, spoons, cups, coins, and standing to reach in our therapy kitchen and bathroom, areas where technology does not t well. I hesitated to use technology because it could not have the patient working within 3 minutes of the start of the OT session, nor could it go home with him.
Occupational Therapy Vignette 2 The hospital acquired a high-tech upper extremity exoskeleton that provided proximal support while a patient plays games on a screen. The hospital provided a half-day course for training; we were instructed that although the initial use would be confusing, consecutive uses become e cient. My colleague and I both lacked con dence and did not want to use normal therapy time to test out the device, but we decided to treat a patient pro bono after clinical hours to setup and run through with the device to improve familiarity.
The patient was a 50-year-old male who recently suffered a stroke resulting in left hemiparesis. We asked this patient to help primarily because he was able to verbally consent to extra treatment and had some movement in his left arm, which is rare in acute stroke. During the treatment session, we noticed that when the patient fatigued his exor tone became worse, requiring removal of his arm from the exoskeleton for stretching. We did three rounds of 1-2 min of exercises, followed by a stretching break. This process repeated, resulting in a total of 9 minutes of treatment in a 40-minute session, including setup. Then, the patient informed us of the need for the restroom, resulting in us ending the session earlier than anticipated.
My colleague and I could not coordinate another time to practice on the new device, and we were hesitant to use it during a normal treatment time with patients since the device involved games when most of our patient's goals revolved around function (dressing, grooming, etc.). It is easier to setup a treatment using task-speci c training with functional, everyday objects than try to make an unfamiliar device work. The device lived on our oor for a while, a daily reminder of the guilt of never nding time to improve familiarity. However, I did not have many patients who would be appropriate for this equipment because they have physical de cits with extremely limited movement on their affected side or they have cognitive de cits that limit attention, initiation, or comprehension of such games. The device was eventually removed from the oor since it was unused by the therapy team.

Occupational Therapy Vignette 3
A 75-year-old male with a prior medical history of atrial brillation, aortic valve repair, hypertension, epilepsy, and a prolonged (4 month) stay in a long-term acute care facility. He then had a subdural hematoma near the right frontal lobe resulting in a traumatic brain injury (TBI). His hospitalization was complicated by pneumonia, seizures, acute ischemic infarcts, and a tracheostomy placement. Impairments included bilateral upper extremity weakness (Action Research Arm Test: 0 left, 10 right), intention tremors in the right arm, decreased trunk control, decreased cervical control, and general malaise. Patient was unable to stand, walk, and hold himself upright in his wheelchair. He identi ed goals of "walking" and becoming more independent with brushing his teeth and getting dressed.
This was a one-hour session with a focus on grooming (oral hygiene and shaving) to improve arm control with both his upper extremities, strength, and upright head control. Since the patient had a low ARAT score on his left arm compared to his right, I decided to use a mobile arm support (MAS) to assist the patient with unweighting his left arm to engage bilaterally in the task. Setup took about 15 minutes, which involved retrieving the MAS from the gym, setting it up in the patient's bathroom, wheeling the patient to the bathroom, and padding the device with towels to prevent his arm from slipping out. The patient had a lot of di culty with incorporating his left arm into the task since he did not have effective grip strength, and overall, he required > 50% assistance with brushing his teeth due to di culties reaching his face. The forward tilted wheelchair frequently led to his chin resting on his chest requiring the right arm to be propped on the sink. We then focused on task-oriented mass practice (retrieving toothbrush from the sink, moving it to the face, then placing it on a nearby table) assisted by the MAS, accomplishing only 15 repetitions because of frequent stops to readjust posture and the tension of the MAS system. This was di cult as I was simultaneously preventing falls out of the wheelchair, readjusting the MAS, cueing, and answering questions from the patient's wife. The wife cued the patient but also expressed disappointment, which overwhelmed the patient. This took 35 minutes after the setup, with another 15 minutes for cleaning, moving the patient back, educating, and answering questions. Running 5 minutes late I excused myself, offering to check in again at the end of the day.

Occupational Therapy Vignette 4
A young female in her twenties with no prior medical history presented 2 months after TBI resulting in multiple brain hemorrhages, diffuse axonal injury of the corpus callosum, splenium, and midbrain. She presented in a minimally conscious state (Rancho Los Amigos 3). She was non-verbal, restless, and moved her arm and right leg around non-purposefully with limited movement on her right side. She stared in the direction of sound cues most times and would follow some commands, but responses were inconsistent, delayed, and eeting.
For this treatment, there were three areas of focus: functional communication, functional object use, and command follow. The communication device used in the session was a large button the patient was asked to hit with either her hand or foot. The patient successfully hit the device with her left foot in 4/20 trials. She also fatigued quickly from the activity, only able to try about 30 seconds at a time. She was unable to hit the device using her arm, and she repeatedly rubs her face and hair when cued to hit the target requiring hand-over-hand assistance. This process took about 25 minutes. We next tried brushing teeth with the left hand with cuing. The patient repeatedly brought the toothbrush to her mouth and chewed it in 3 instances after 20 minutes of engagement in the activity. During this time, the attending physician and students entered to watch and speak to the mother and the patient becomes fatigued and falls asleep in her wheelchair. I leave the session 5 minutes early.
Occupational Therapy Vignette 5 As a oat occupational therapist who sees patients on all different units, I am often seeing new patients each day that I have never met. I am a techno-phobe in my treatments because I feel strongly that using familiar objects/tasks are more motivating and patients understand how the intervention will be helpful to reach their goals. Consistent use of technology can be di cult because chart reviews may not paint the picture of how the patient is going to look when you nally see them.
A recent patient who suffered a stroke had visual de cits, left hemiparesis, was motivated, and wanted to work on visual scanning. I considered an interactive light board designed to train visual scanning and reaction timing, but it had been a while since the twohour training session. When I am seeing patients more consistently and have built a rapport, I will know who will bene t from something new and I will brush up on my skills with such equipment. In this case, I chose to stick to familiar functional tasks and use time wisely, opting to use a visual scanning kit to organize a tackle box as shown in a picture. The patient had some di culty completing it and needed some cueing, however, ultimately was able to complete the task with some support.

Occupational Therapy Vignette 6
As an occupational therapist working with the Spinal Cord Injury population, I am always looking for ways to adapt every self-care task to allow them to be more independent. I worked with a young patient who was in a motor vehicle accident and sustained a C2 fracture, was on a ventilator, and had no movement in his upper or lower extremities other than the ability to shrug his shoulders. He was 19 years old and was in the US Army at the time of his accident. He did not have any surgical intervention, and when he was admitted he was classi ed as an ASIA B (sensory function preserved, motor function is not, below the level of injury).
Since he was active prior to injury, I thought he might enjoy the functional electrical stimulation (FES) bike as he became more stable. However, I was not as comfortable with setting up this piece of equipment at the time, so I contacted the vendor who is also an OT. The vendor attended, co-treated, and helped me with optimal setup. It was also helpful to have a second person since we added scapula and arm electrodes that took a long time. The patient loved this intervention because it was the closest to lifting weights, an activity that was important to him. Also, electrical stimulation below his injury level can see if he might regain strength in those muscles. When it was time for him to discharge, I provided him with information to get a cycle for home at a discounted military rate. This is one of my favorite interventions with SCI patients because of high repetitions and endurance, although it can take a while to set up.
Physical Therapy Vignette 1 I am a PT working almost exclusively with traumatic and acquired brain injuries in an acute inpatient setting, and I am neutral to technology. There have been times when I frivolously support it, times I am skeptic, and times, I believe some devices will gather dust. The determining factor for my technology use is its adaptability, and if it's quick to learn (~ 2 practice trials under supervision after formal training session). One of the pieces of technology I unequivocally support is Body Weight Supported Treadmill Training (BWSTMT). BWSTMT allows for earlier ambulation interventions, while providing assistance to patient needs and high-intensity when appropriate. Speed can be modi ed to focus on aspects of gait that need more ne-tune control, such as step length or terminal knee extension. Other adaptations can be made to encourage upright trunk through use of strapping, additional inclines and allowance for backwards and lateral stepping. If not for such technology, I would be unable to perform this treatment. This also frees my hands, so I can assist with a speci c target area (by positioning limbs in a speci c phase in gait, or adding bolsters for a patient to step over) or stand back and see the bigger picture of gait mechanics. The portable remote allows me to change parameters such as speed and incline in real time, making the treadmill e cient to operate. Additionally, there are built-in safety/backup measures available and can be adjusted by the treating clinician.
As a student, I was lucky to work in a hospital that taught me how to set up and frequently use BWSTMT. Now as a Clinical Instructor for PT students, performing BWSTMT is an integral part of my daily practice. Thus, most students learn this as a part of their clinical experiences. For those who are unfamiliar with the BWSTMT, it takes practice, with variable patient assist levels, before using it independently. Often times, the practice is needed to learn more challenging set ups of the harness system on someone who requires 100% assist to don/doff a treadmill harness from a seated position and without a second hand to make the set-up smoother. Any sort of learning needed for this machine is integrated as part of the new hiring system.
While I am in full support of using BWSTMT for ambulation interventions, I have identi ed some drawbacks both in design and in application. Limitations in documenting body weight supported objectively or tting the equipment to all patient sizes and shapes do not outweigh the positives. However, scheduling problems and patient cognition do limit my ability to use BWSTMT in a session.
It can take about 15 minutes to set up someone and about 5 minutes to remove/clean up (less if the patient is higher level and the clinician is experienced and/or there is an extra assistor like a tech/aide). This amounts 20 minutes of therapy time. If a patient is scheduled for a 30-minute session, or has an incontinent episode, a delay in medication, eating, or in bed at the beginning of therapy, it may be more effective and e cient to choose another treatment. Additionally, cognitive characteristics that accompany brain injury can decrease a patient's tolerance for BWSTMT. A lack of insight into de cits in conjunction with a low frustration tolerance with high verbal and physical outbursts can impact safety of clinicians and patients, can reduce buy-in, or lead to treatment resistance. These limitations can present ethical implications. As a clinician, understanding how to modify the treatment parameters to the individual is just as important as knowing how to operate the technology itself.
Physical Therapy Vignette 2 I am a PT with 9 years of experience in a large hospital, primarily inpatient. Because of my neurologic physical therapy residency, I am an early adopter, enthusiast, and proponent of novel technology, and have even participated in some e cacy and feasibility trials. My organization has many advanced devices available, but this does not equate to regular use unless it is easy to set up, does not cause patient downtime, or extensive personal time. Limits to technology use are: (1) is the device unready; (2) the patient is unready for treatment; (3) the patient goals do not match; (4) imbalance between organization or insurance requirements and ideal (evidence-based practice).
The patient was a 70-year-old female with right sided hemiparesis due to a left pare median pontine perforator infarct. Her goals were to improve walking speed and distance without assistance at home. I had a 60-minute session and recognized the need for intense task-speci c training. In previous sessions, she had di culty reaching target heart rate due to poor right foot clearance, but more recently demonstrated improvement in proximal leg strength with remaining di culty with knee control and foot clearance.
I chose to use BWSTMT to strengthen proximal leg muscles because we had equipment ready and I am extensively trained. To improve foot clearance, I had di culty deciding whether to use a traditional ankle foot orthosis (AFO) or an FES orthosis. Although trained in FES, the barriers were (1) the device was four oors away; (2) the device was locked in a manager's o ce because it is expensive; (3) I am required to send an email to the manager to check it out; (4) the device may not be charged; (5) I was unsure of where the electrodes were, and (6) if I delegate to a rehab aide, they might grab the incorrect one. Instead, I chose traditional AFO in the nearby cabinet, which worked okay but required more verbal cues more hands-on assistance for timing of step initiation and step lengths to maximize intensity.
The next day, I considered nding 20 minutes of personal time to locate and setup the FES device, but had an insurance progress note due, which meant I needed to assess the patient's mobility skills and had limited time for gait training. The progress notes also led to extra documentation requirements, limiting my available time during lunch to track down the device. Additionally, I have some concerns with using this device in inpatient rehabilitation, for fear that it may not be covered by the patient's insurance when they are discharged.

Physical Therapy Vignette 3
A 48-year-old female with secondary progressive MS had goals to stand and make stepping actions. She had severe lower extremity weakness, already wore custom AFOs, and wanted to get stronger. Her parents were aging and were dependently lifting her in and out of bed. At the time, she required maximum assistance to stand and we only had her for a short length of stay to get her home safely with less assistance since she already had good support at home. I considered the FES bike at one point but ultimately determined it would not have been bene cial time investment during inpatient rehabilitation as she was also seeing one hour of speech therapy, one hour of occupational therapy, and one hour of physical therapy 5 of 7 days a week. Using her one-hour of physical therapy and taking at least 20 minutes or more to set up 4-6 electrodes on each leg before testing and trialing for one 30min training session is not ideal for the time frame of her length of stay, and not supportive of her over all goals of transferring independently. Also, using the FES bike one time would not achieve the recommended frequency and duration to obtain the known bene ts of the device. I chose to focus on slide-board transfers to reduce burden of care on her aging parents. Ultimately, her transfers improved signi cantly and she was able to transfer herself using the slide-board.

Discussion
Applying therapeutic RT for individuals with neurological impairments requires successful progression through a long and fragile chain of events: development, testing, validation, clinician uptake, and patient acceptance. Although each step presents its own set of challenges, we focused on the often overlooked but critical step of clinician uptake (12). This study moved beyond surveys or focus groups to a vignette methodology that allowed us to understand clinicians' real-time decision-making process in the moments when they are with their patients. We found that the ve most common themes fell within the CFIR constructs of relative advantage, personal attributes of the patients, clinician knowledge/beliefs, device complexity (including time and setup), and organizational readiness to implement.

Relative Advantage
The most commonly discussed barrier to using RT was its perceived relative disadvantage due to lack of relevance to everyday functional activities. Priority in this setting was task-speci c practice of day-to-day functional activities. In contrast, many therapeutic RT enable repetitive practice, but usually in the form of games and simple strength and range of motion impairmentfocused activities. Therapy outcomes are evaluated and reimbursed based on functional outcomes (e.g. bed mobility, transfers, walking, and self-care skills such as dressing, toileting, eating, and bathing) (18) , rather than impairments or assessment scores (e.g. Action Research Arm Test, Fugl Meyer Assessment of Motor Recovery, Berg Balance Scale). Research has shown the biggest predictor of intention to use a RT is performance expectancy, or the degree to which an individual believes in the potential bene t of that RT (2,19). Thus, if therapists are unable to connect the impairment-focused task or game to a functional bene t they are less likely to use the device. RT endorsed in our vignettes increased repetitions of functional tasks, and enabled patients to complete an action they could not otherwise complete, such as BWSTM and MAS systems. To improve clinician uptake in inpatient rehabilitation, RT development should demonstrate the e cacy of RT in addressing functional outcomes. Furthermore, it may be of interest to developers to include functional metrics used by payer sources in the validation process and demonstration of e cacy of their devices.

Patient Attributes
Attributes of the patients and adaptability of the device determine clinician uptake and use of RT. Understanding the attributes of patients, as well as their needs and resources were common themes in the vignettes. In other studies, patient acceptance was a highly important factor for RT adoption (10). Patient diagnoses, goals, and physical and cognitive abilities play a large role in guiding treatment decisions. Many engineering development studies exclude patients with cognitive de cits and can have very tight inclusion criteria related to physical function and sensation. Although necessary in the development phase, these restrictions limit generalizability in inpatient rehabilitation. It has been proposed that developers should clearly identify the appropriate patient population for their devices (4). However, it would be more valuable to clinicians for developers to design quality devices that are adaptable to a variety of diagnoses, patient needs, and environments.

Clinician Knowledge and Beliefs
A third CFIR construct endorsed by all ve therapists was clinician knowledge and beliefs about the intervention. Clinicians' individual experiences and comfort with RT, as well as their readiness to change, greatly in uenced their decision to use RT in treatment. We found that clinicians appeared to value RT more when it is incorporated into their academic training, onboarding, or a part of their regular clinical practice. Providing only a single training session for complex RT may result in lower mastery or clinician self-e cacy, which are required for use with real patients. Interestingly, prior studies did not link RT uptake or barriers to therapists' employment status, age, discipline, educational level, experience or technology acceptance (2,10,14). This suggests that clinician experience and training with speci c RT may dictate RT uptake more than general clinician experience.

Device Complexity and Time
One of the most striking issues described narratively in the vignettes was that therapists have extraordinarily little time to use complex RT. Studies frequently mention the importance of simplicity, ease of set up, and convenient availability of RT (10,14). One limitation of the CFIR framework is that it does not include a speci c construct for time. Instead, comments related to time barriers fell under several other constructs, including complexity, relative advantage, clinician stage of change, clinician knowledge and beliefs, implementation climate, and external policy. RT developers cannot create more time for therapist, and it is di cult for them to in uence implementation climate and external policy. However, RT developers can create devices that are quick, intuitive, present a clear advantage over traditional interventions, and optimize clinical work ow. Additionally, training protocols should target moving the clinician through readiness to change into adoption of novel RT by addressing their knowledge and beliefs about the bene ts of RT to support the amount of time it takes to use RT, particularly if the RT is perceived as complex.
Organizational Readiness to Implement Although RT developers may have low in uence on organizational readiness, our vignettes support the importance of the organizational implementation climate for clinician uptake of RT. Implementation climate is de ned as the "absorptive capacity for change, shared receptivity of involved individuals to an intervention, and the extent to which use of that intervention will be rewarded, supported, and expected within their organization" (15). Further organizational investments in the implementation process are helpful to improve clinician uptake of devices. Examples of investments include time for practice and re ection, organizational incentives to increase motivation to use RT, as well as assistance with RT setup from rehabilitation technicians. Organizational implementation factors in the literature include support from the institution to facilitate use (2) and making RT use mandatory and seamless in clinical treatment (4). Despite a goal of RT to supplement and assist therapists with treatment, it can place an unintentional burden on therapists when it comes to uptake and use in the clinic. Adding expectations for therapists to be trained on different RT oftentimes is an investment from therapist's personal time, which in turn might contribute to burnout (12). RT developers should consider how their devices t into the overall organizational priorities and work ow in order to address the clinician uptake barrier related to organizational readiness to implement.

Other Constructs
Our interdisciplinary research team was surprised with infrequent mentions of evidence strength and cost compared to previously reported barriers to device implementation. Others have reported that clinicians rated cost as a very important acquisition factor (10,14) or indicate cost as a barrier due to limited cost-effectiveness compared to intensive therapy evidence to warrant use (4,20). Our vignettes mentioned cost in the context of inconvenient procedural controls put in place to protect an expensive device or costs relative to a patient's ability to acquire the device after discharge through insurance or other discounts. Other studies have shown that OTs and PTs rate evidence as the most important factor behind device use (14). These vignettes dichotomously presented evidence strength as both allowing therapists to administer the amount of repetitions recommended for signi cant clinical change; but also deterred another therapist due to inability to provide repeated use of RT to reach clinically signi cant gains due to short patient length of stay. The clinic already acquired and made the RT available to the clinicians in the vignettes, which suggests that the RT available to the therapists had enough evidence and were believed to be cost-effective to support the organization acquiring the device. These constructs suggest the importance of studying clinician uptake of RT after addressing the initial development and acquisition barriers.

Limitations and Directions for Future Research
Limitations are present in this analysis related to generalizability and the qualitative research methods. We acquired vignettes from one inpatient rehabilitation hospital with state-of-the-art RT, which may hold entrenched biases that are not re ective of the wider community of practitioners. A larger qualitative study exploring the barriers and facilitators in multiple organizations and therapists working in different rehabilitation setting would improve the generalizability. Our report here should only be understood as a single snapshot of beliefs that still may be important for developers to understand as they prepare for the design or development process in this eld. Related to the qualitative methods, our vignette prompt may have led to increased mentions of certain topics, such as the time taken to setup. A different vignette prompt could explore greater consideration of impairment-focused treatment and measurement RT. Additionally, we limited our coding to the 17 CFIR constructs that were already present in the literature, rather than inductive analysis to allow themes to emerge. However, using established implementation frameworks have demonstrated generalizability of ndings in the presence of limited number of settings (21).
Future research and development should attend to determinants of successful clinical uptake supported in these vignettes. New RT should address relative advantage of functional task practice over impairment-focused interventions, the adaptability required to address varied patient populations, and the complexity of the RT. Developers also need to consider the importance of clinicians' knowledge and beliefs about the intervention and support from the institution to facilitate a positive implementation climate.