Advanced Technology in a Real-World Rehabilitation Setting: Longitudinal Observational Study on Clinician Adoption and Implementation

Introduction

There is burgeoning interest in harnessing the power of digital interventions to enhance health care services [1]. Such interventions have potential to help address the growing need for rehabilitation worldwide [2,3]. Within rehabilitation, this may involve devices such as virtual reality, robotics, smartphone apps and activity trackers [4]. While evidence is emerging, research has shown digital interventions can improve patient outcomes (eg, in mobility and upper limb function) [5-9], improve patient engagement in rehabilitation [10-13], and increase therapy dosage [6,14,15].

As technologies advance and become more affordable [16-18], they are increasingly accessible in rehabilitation [10,19,20]. Despite this, low device uptake among clinicians and frequent reports of technology abandonment persist [19,21-24]. Addressing these issues is imperative to avoid wasting health care resources on unsuccessful technology implementation efforts [24,25]. With a paucity of research conducted within rehabilitation settings, key issues include a poor understanding in selecting devices for clinical settings, who to use devices with and how to effectively integrate technology with current rehabilitation practices [1,19,26-28].

A new technology therapy center was opened by an Australian-based rehabilitation provider in 2022. This study is based within the center, which incorporates a range of advanced technologies, defined by the provider as “new technological interventions used in therapy to empower physical, cognitive, social or communicative skill development.” Advanced features include robotic componentry to maneuver limbs (eg, powered and nonpowered exoskeleton and end-effector devices) [29,30], immersive and nonimmersive virtual reality (VR) [6,31], multichannel functional electrical stimulation (FES), built-in sensors which provide real-time feedback, or a combination of these features.

Current evidence for advanced technologies is in its infancy, demonstrating potential benefits for additional therapy dosage [6,8,15,32], multisensory feedback and reducing physical demands on therapists, especially when working with patients who have little to no active motor control in major muscle groups [4,13,22,30]. In the landscape of rapidly evolving technology, research cannot afford to wait for effectiveness to be definitively established before examining implementation within clinical settings. For complex interventions, evidence of effectiveness alone is not adequate to facilitate integration into practice due to the interplay of factors such as device novelty, substantial learning requirements, and extensive organizational adaptation [24,33]. Investigating the real-world implementation of a range of advanced technologies presents a unique opportunity to advance this research field, generating practice-based evidence to directly inform technology use within clinical rehabilitation and future large-scale trials [34].

The aim of this study was to investigate the implementation and adoption of advanced technology within a real-world clinical rehabilitation setting. Specific research questions regarding this rehabilitation setting were as follows:

1. Which advanced technologies did clinicians use?
2. Who used the advanced technology?
3. Why did clinicians use advanced technology?
4. What therapy dosage did patients achieve with advanced technology?

Methods

Study Design

This longitudinal observational study of usual care is reported according to the Reporting of Studies Conducted using Observational Routinely-collected health Data (RECORD) statement [35].

Setting and Participants

The study site is a combined private-public rehabilitation setting in Australia which serves adults with neurological and nonneurological conditions, specializing in spinal cord and brain injury. Rehabilitation services include 5 combined outpatient and community services, 3 inpatient units, and 2 community only services. The technology therapy center opened in July 2022 and all allied health clinicians in the organization (n=119) had access to devices.

Clinicians received training on devices before and after center opening. Training requirements for each device varied and are outlined in Multimedia Appendix 1; training timeframes ranged from 2 hours to more than 35 hours. Various models of training were used, including external trainers, web-based modules, internal train-the-trainer, and communities of practice. Clinicians had clinical support, ongoing training, and technical support through an internal advanced technology clinical lead and ad hoc support from technology suppliers.

Intervention: Advanced Technologies

Between September and October 2020, the advanced technology clinical lead, senior allied health clinicians and management collaborated to finalize a list of technologies for purchase, based on current Australian Therapeutic Goods Administration–approved rehabilitation devices and using a selection criteria tool [36]. Another element of device selection was data security. Devices were only selected if data could be stored locally or on secure cloud-based software in Australia. Devices were acquired between November 2020 and February 2022. At center opening, 25 advanced technologies were available, 21 of which were physical rehabilitation devices. Multimedia Appendix 1 contains details of these devices, categorized in Textbox 1.

Data Sources and Variables

Data were recorded by clinicians and extracted weekly through a Business Intelligence report, which searched all patient electronic medical records (Kyra Clinical, Telstra Health, Melbourne, Australia) organization-wide from July 1, 2022, to June 30, 2023. Progress note entries which included an “advanced technology therapy session” template were eligible for inclusion. This template includes device-specific subheadings to prompt clinicians to record device data (refer to Multimedia Appendix 2) and was introduced as usual practice before commencing data collection. Our only exclusion criterion was any progress note entry which detailed use of a device not outlined in Multimedia Appendix 1 (eg, single or dual-channel electrical stimulation devices), resulting in exclusion of 54 progress note entries. Data extracted included: session date, patient details (medical record number, age, gender, primary health condition, and clinical service), clinician discipline, device used, location, technology session goals, duration (total time, active time [eg, walk time]) and steps, repetitions, or meters walked. In addition to medical record data, nonidentifiable service data regarding number of staff trained per device were collected. Variables used for analysis are detailed in Textbox 2.

Statistical Analysis

Data analysis for this study was conducted in R (version 4.3.1; R Core Team) and consisted of summary statistics to describe device use by device category, patient characteristics, clinician discipline, and clinical service, each month and across the 12-month period. Missing data within included progress note entries were also recorded and summarized descriptively. Chi-square tests of independence were conducted to investigate associations between type of device used and patient primary health condition, as well as rehabilitation service. A significance level of .05 was used for chi-square tests.

Ethical Considerations

Ethics approval for this study was provided by Northern Sydney Local Health District Human Research Ethics Committee on May 19, 2022 (2022/ETH00364). A waiver of consent was granted to collect data from patients’ electronic medical records, including patients’ medical record numbers to allow analysis of the number of sessions and variety of different technologies used by each patient. However, no personal information regarding patients or clinicians was collected and no individual participants are identifiable in any findings presented in this study.

Results

Overall Advanced Technology Use and Staff Training

Across the 12 months, clinicians used advanced technologies 4208 times with 269 patients (14% of all patients). All rehabilitation services within the organization accessed devices for patient therapy sessions. In total, 80 allied health clinicians (80/119, 67%) were trained across devices (n=41 outpatient and community clinicians, n=29 inpatient clinicians and n=10 clinicians working across inpatient, outpatient, and community services); including 35 physiotherapists, 33 occupational therapists, and 4 allied health assistants.

Which Advanced Technologies Did Clinicians Use?

Our data captured clinicians using 20 of 21 (95%) physical rehabilitation devices. Monthly device use gradually increased, excluding holiday months (Figure 1).

Lower limb devices were most frequently used (2513/4208, 60%). The most used device was an augmented VR treadmill with optional BWS (580/4208, 14%). Devices for upper limb training were also frequently used (1270/4208, 30%); most commonly, a robotic VR device for distal upper limb joints (374/4208, 9%). Table 1 and Table 2 detail device use across patient populations and clinical services.

‎

Who Used the Advanced Technology?

Clinicians

Physiotherapists were the highest users of advanced technology (2716/4208, 65%), followed by occupational therapists (1396/4208, 33%), and allied health assistants (96/4208, 2%). Occupational therapists gradually increased device use each month, relative to physiotherapists (from 25% in the first month to 48% by the twelfth month, Figure 2).

‎

Patients

Over 12 months, 445 inpatients and 1502 outpatient and community patients received rehabilitation services. Out of the 269 patients with whom clinicians used advanced technology, the average age of patients was 52.3 years (SD 14.1, range 14-83 years), 95 were female (35%) and 255 had neurological diagnoses (95%); predominantly stroke, spinal cord injury, or brain injury (71%). Table 3 displays patient demographics.

The majority of patients (178/269, 66%) used more than 1 device, including 20% of whom used 5 or more devices. Approximately 45% of patients had 5 or less sessions with advanced technology (120/269), 23% had 6 to 10 sessions (62/269) and 32% had more than 10 sessions (87/269). Data on repeated device use (ie, the same patient using a given device more than once) demonstrated a median of 5 (IQR 1-7) repeated uses overall across devices. The Robotic-T-BWS-VR device had the highest repeated uses (median 7, IQR 5-10), where 89% (32/36) of patients who used this device did so more than once. This was followed closely by the Robotic-OG device (median 7, IQR 4-12), where 81% (52/64) of patients who used this device did so more than once. Details regarding repeated use across each device are summarized in Multimedia Appendix 3.

There is strong evidence that use of robotic versus nonrobotic lower limb devices is associated with patient primary health condition (χ²₅=16.8, P=.005). For patients with spinal cord injury, clinicians used robotic and nonrobotic lower limb devices equally. For all other conditions, clinicians typically used nonrobotic lower limb devices (between 69% and 76%). Figure 3 depicts these differences.

For upper limb devices, there is no evidence of an association between type of device used (robotic vs nonrobotic) and primary health condition (χ²₄=1.1, P=.78). Almost all patients with nonneurological conditions only used sensor and BWS devices, with only 1 having used a robotic upper limb device.

‎

Rehabilitation Service Type

Outpatient rehabilitation services were dominant users of advanced technology (3936/4208, 94%). Inpatient services used advanced technology 264 times (6%), while offsite community services used advanced technology 4 times (0.1%). Inpatient clinicians used advanced technology with 49 of 445 patients (11%), while onsite outpatient and community clinicians used advanced technology with 226 of 1502 patients (15%).

There is strong evidence of an association between type of lower limb device used (robotic vs nonrobotic) and rehabilitation service (χ²₁=6.6, P=.01). Outpatient clinicians used more nonrobotic lower limb devices (1372/2335, 59%), whereas inpatient clinicians used robotic (91/178, 51%) and nonrobotic (87/178, 49%) lower limb devices equally. There is also very strong evidence of an association between type of upper limb device used (robotic vs nonrobotic) and rehabilitation service (χ²₁=13.3, P<.001). Inpatient clinicians used robotic upper limb devices (47/55, 85%) considerably more than nonrobotic upper limb devices (8/55, 15%). Meanwhile, outpatient clinicians also used robotic upper limb devices (739/1211, 61%) more than nonrobotic upper limb devices (472/1211, 39%), however the difference was not as prominent. Figure 4 portrays these differences. The relative use of augmented or immersive VR and advanced FES devices was similar across all onsite services. Offsite community services only used portable upper limb sensor devices (4/4, 100%).

‎

Why Did Clinicians Use Advanced Technology?

Clinicians reported 144 different goals for using advanced technology in patient therapy sessions. On average, clinicians reported 1.8 goals per session (SD 1.2, range 1-5), with only 41 sessions (41/4208, 1%) missing goal documentation. Goals were predominantly at the impairment (5063/8499, 60%) or activity-limitation level (2600/8499, 31%). Although more participation-level goals were reported as the year progressed, these only formed 1% of goals across the 12 months (most commonly return to employment [16/8499, 0.2%]). Clinicians frequently cited goals related to gait training (1959/8499, 23%), lower limb strengthening (555/8499, 7%) and balance (537/8499, 6%). Upper limb goals, particularly fine motor control (942/8499, 11%), range of motion (827/8499, 10%), and strengthening (690/8499, 8%) were also frequently reported. Table 4 details advanced technology therapy session goals.

What Therapy Dosage Did Patients Achieve With Advanced Technology?

Therapy dosage data were collected from progress note entries, in which clinicians reported data from devices. Therapy dosage data from each device are summarized in Multimedia Appendix 4, displaying median active therapy time for 16 devices (80%), median total time on device for 18 devices (90%) and median number of repetitions for 5 devices (25%). However, due to inconsistent device measuring and reporting of therapy dosage data, there were large amounts of missing data and results should be interpreted with caution.

Overall, median total time on devices was 30 minutes (IQR 28-45 minutes), while median device active therapy time was 20 minutes (IQR 13-28 minutes). Active time for lower limb devices was highest for the robotic treadmill BWS device with VR (median 30 minutes, IQR 23-37 minutes), followed by the robotic device with integrated FES (median 27 minutes, IQR 21-33 minutes), and the VR treadmill with optional BWS (median 24 minutes, IQR 16-32 minutes). Drawing conclusions regarding active therapy time relative to total therapy time for lower limb devices was particularly challenging due to inadequate recording of either total time or active time. An OG robotic exoskeleton was the only lower limb device which consistently recorded both active therapy time and total therapy time. This showed approximately 50% of the time spent on the device was active, where median active time achieved in a session was 17 minutes (IQR 13-22 minutes) and total time on the device was 35 minutes (IQR 28-40 minutes).

Active time for upper limb devices was highest for a robotic device with VR used to train the proximal upper limb (median 24 minutes, IQR 16-32 minutes), followed by a sensor-based VR upper limb device (median 20 minutes, IQR 17-20 minutes). There were no marked differences in median active therapy time between robotic and sensor-based upper limb devices. Generally, active therapy time relative to total time on the device was roughly similar across upper limb devices, from between 53% and 67% of time spent active.

A more accurate measure of therapy dosage is number of repetitions or steps, and meters walked [38]. However, this was only reported by four lower limb devices and 1 advanced FES device. No upper limb devices consistently reported number of repetitions. Across the 4 lower limb devices, the range of median repetitions was 356 to 1292 repetitions per session. The highest repetitions and meters walked per session were achieved through the robotic treadmill BWS device with VR (median 1292 repetitions, IQR 994-1674 repetitions; median 756 meters, IQR 573-983 meters).

Discussion

Principal Results

This longitudinal observational study investigated the implementation of advanced technologies in a clinical rehabilitation setting. A total of 20 devices were used by clinicians 4208 times, with 269 patients in the first year of the center opening. Frequently used devices included gait training BWS (VR treadmill and OG), OG robotic exoskeletons, and upper limb robotic VR devices. Device use differed considerably between rehabilitation settings, with the majority (3940/4208, 94%) from outpatient services. Clinicians used advanced technologies with a wide range of patients, aged 14 to 83 years, the majority diagnosed with neurological conditions (255/269, 95%). Devices were most often used to target impairment and activity limitation–related goals. Conclusions regarding therapy dosage were limited due to inconsistent recording across devices. These insights derived within a rehabilitation setting can directly inform clinical practice and future research in the field.

Our study demonstrates that while clinicians can successfully incorporate advanced technology into practice, this differs between inpatient, outpatient, and community settings. Notably, inpatient services only accounted for 15% of device use and offsite community services accounted for less than 1%. Reasons for this are likely multifactorial, including marked differences in the available number of inpatients (n=445) compared with outpatient and community patients (n=1502). Coupled with the need to address other goals such as equipment prescription and discharge planning, inpatient clinicians used advanced technology less frequently. Community therapists based offsite showed little use of advanced technologies despite availability of portable devices (n=8). Logistical challenges to accessing and planning device use, and differing service models (eg, focus on hospital-to-home transitions) are likely reasons. These factors collectively influenced inpatient and community-based clinicians’ opportunities to use advanced technology and familiarity with devices. Organizational factors (eg, clinician training and support personnel) also likely contributed to variations between settings, which are being investigated in a separate qualitative study.

Advanced technology use also differed across patient populations and rehabilitation settings. Clinicians most commonly used advanced technologies for patients with stroke (88/269, 47%), and the greatest variety of devices (n=20) for patients with stroke and spinal cord injury. The relative use of lower limb robotic devices was greater for patients with spinal cord injury. Meanwhile, the relative use of both upper limb and lower limb robotic devices was greater in inpatient services. One possible explanation is the nature of impairments in spinal cord injury and acute or subacute neurological conditions in inpatient settings. Clinicians often seek to promote neuroplasticity and address impairments such as muscle weakness through high repetitions, which robotic devices can facilitate [29,30]. Furthermore, the manual handling benefits and opportunities for therapy with robotic devices are potentially greater in these settings and populations, where muscle weakness can be significant [22,39,40]. Although a key argument for using technology in rehabilitation is to increase therapy dosage [6,14,15], our study has highlighted that in practice, advanced technology prescription is highly variable and measuring actual therapy dosage is challenging. Digital health developers should address this as a priority through designing devices which report and standardize dosage output. Devices should include measures regarding the number of repetitions and time spent active, and ensure this information is visible and easily accessible to clinicians. This has important implications for facilitating rigorous research to guide rehabilitation technology prescription and recommended dosage. Overall, further research is required to provide guidance on device selection and prescription across rehabilitation settings and populations. Importantly, future research should seek to establish the effectiveness of using advanced technology alongside conventional interventions to improve patient outcomes in rehabilitation.

Despite the range of devices in the study site, not all devices were used equally. While devices targeting gait training and upper limb impairments were used frequently, advanced functional electrical stimulation and immersive VR devices were used least. In addition, 1 upper limb sensor-VR rehabilitation device was not used at all by clinicians. Reasons for this are likely related to clinician preference and device functionality, both of which warrant investigation in future research. Another possible explanation is the level and reach of clinician training on these devices, which was lower relative to other devices. There was no formal training process for using the upper limb sensor-VR device, and immersive VR and advanced FES were 2 of 5 devices with no increase in number of clinicians trained between baseline and 12 months (n=13 and 11, respectively). With a busy clinical workload and many devices to learn, there are fundamental complexities to providing clinician training within clinical practice which merits further research. Interestingly, there was no evidence in our study of clinicians preferring devices with easier or quicker training processes, reflected in the fact that among the highest used devices was the robotic OG exoskeleton which had the longest training requirements (30-35 hours). A scoping review conducted by our team found training and educating clinicians were the most common strategies used to implement digital interventions in rehabilitation [27]. With evidence to show this is an important element for technology implementation, future research should focus on how training should be conducted, the impact of training on clinician uptake of technology, and determining models of training that are efficient, cost-effective, and sustainable.

Comparison With Previous Work

Other studies have investigated clinical use of advanced technologies, mostly focusing on OG robotic exoskeleton devices [39,41,42]. These studies also found competing priorities in inpatient settings, where discharge planning often took precedence. In contrast to our study, Gillespie et al [41] found patients with brain injury used the robotic exoskeleton more than patients with stroke or spinal cord injury. However, it was also reported that patients with brain injury had the most variability in therapy dosage achieved with the robotic device [41]. Swank et al [42] also highlighted differences in OG robotic exoskeleton use between patients with spinal cord injury and stroke. Our study extends these findings across a broad spectrum of advanced technologies and neurological conditions. While our study found advanced technologies were primarily used to address impairments and activity limitations, Putrino and Krakauer [43] argue that neurological rehabilitation technologies are best used to address impairments, and conventional rehabilitation is better at targeting activity and participation-level goals. Therefore, it is important that use of technology is combined with conventional rehabilitation methods, rather than using technology in isolation [6,20,22]. Finally, our study found clinicians using advanced technologies with only 11% of inpatients and 15% of outpatient and community patients. This suggests that current advanced technologies may not be suitable for everyone and is consistent with other literature in the field demonstrating 15%-20% patient eligibility for studies evaluating rehabilitation technologies [5,6]. This emphasizes the need for a better understanding of who technology is appropriate and effective for, and how to design technologies to better meet the needs of both patients and clinicians.

Limitations

This study has a few limitations. First, a limitation was the inability to capture patient outcome measure data due to a lack of standardized recording and reporting of outcome measures clinically. As such, this study focuses on clinician use and uptake of advanced technology in a real-world rehabilitation setting, rather than the effectiveness of using advanced technology to improve patient outcomes. Inconsistent measuring and recording of outcome measures is a common issue in rehabilitation and should be prioritized for clinical practice and practice-embedded research [44-47]. Second, a multisite study was not feasible given the novelty of the technology center. We sought to address generalizability by collecting data on a wide range of patients, clinical disciplines, and rehabilitation services. However, the authors of this study acknowledge that the study site is a well-resourced, urban, private-public rehabilitation setting within a high-income country. Noteworthy factors which likely contributed to substantial device uptake are the construction of a purpose-built technology therapy center within the existing site, availability of a wide variety of advanced technologies, appointment of an advanced technology lead to support multiple aspects of implementation including staff training, and patient access to long-term rehabilitation services through public funding models. Therefore, lower resourced settings would likely face challenges in applying the findings of this research. Current evidence has shown low-cost devices such as commercially available VR, activity trackers and mobile phone apps are cost-effective in rehabilitation [48]. Future research should investigate the cost-effectiveness of high-cost advanced technologies. Research in the field should also continue to place equal emphasis on affordable and low-cost rehabilitation technologies. Third, data in this study were reliant on clinicians using note templates for documentation. As such, device use is likely underreported, particularly for disciplines such as speech pathology or recreational therapy who were not captured in our data. However, given the large number of advanced technology sessions captured, the data likely adequately reflect clinical practice for physiotherapists and occupational therapists in the setting, currently the main users of advanced technologies.

Conclusions

This is one of the largest longitudinal observational studies conducted within a real-world rehabilitation setting which successfully implemented numerous advanced robotic and virtual reality technologies for patients with a wide range of health conditions. This study provides valuable insights into the types of advanced technologies most likely to be used in rehabilitation, by which clinical disciplines and rehabilitation service types, for which kind of patients, and for what purpose. Findings are important to inform rehabilitation technology implementation, aiding clinicians in device selection, outlining the likely uptake of devices, and providing recommendations for future research in the field.