Telehealth for rehabilitation and recovery after stroke: State of the evidence and future directions

What is telehealth? How can we move toward a common language?

The term telehealth covers the provision of healthcare contacts that are not provided in person. This can range from real-time phone and/or video conferencing, to asynchronous contact via digital apps, and remote monitoring using sensors. Other common terms that sit under this umbrella include telemedicine, which implies physician-led contact, and telerehabilitation, which may involve one or more rehabilitation disciplines. Over the past few years, the term “virtual care” has emerged. ³ However, in the field of rehabilitation, this can cause confusion with the term “virtual reality.” Virtual reality refers to the use of systems or devices that allow users to interact with computer-generated scenarios, objects, and events. Telehealth remains the most commonly used and readily understood term.

Clear definitions and consistent terminology are required to ensure we understand and advance the field. We therefore propose a taxonomy of terminology (Table 1) and recommend clinicians and researchers adopt this framework to improve clarity and consistency. In this taxonomy we suggest that any telehealth encounter is described by documenting (1) purpose of the encounter (assessment or intervention), (2) timing of delivery (synchronous, asynchronous, or continuous), and (3) type of technology used (audio/telephone, video conferencing, digital app platforms, wearable sensors). Beyond using appropriate terminology, the more important issue is to differentiate where telehealth is used as a medium to provide an intervention unchanged from in-person delivery or includes active intervention components. Clarity in terminology will enable us to more readily identify this difference.

OPEN IN VIEWER

Table 1.

Taxonomy of terminology.

Telehealth
Purpose of the encounter		Timing of delivery			Technology used
Assessment(To assess what?)	Intervention (To improve what?)	Synchronous (real time, supervised)	Asynchronous	Continuous (remote monitoring)	Audio only	Video link	Digital apps
Examples
Assessment of blood pressure		using asynchronous data collection			via (specify digital platform)
Assessment of rehabilitation needs		via real time			telephone consultations
Assessment of physical activity		via continuous monitoring			using (specify sensor type) or (specify digital platform)
Improving arm motor function		via real time, supervised			video conferencing
Exercise training to reduce secondary stroke risk		using a combination of real time, supervised			sessions via video conferencing
		AND asynchronous			support via (specify digital platform)

What is the current evidence base?

Asking “is telehealth effective?” is enormously challenging. A Cochrane review published in 2020 identified 23 randomized trials of some form of telehealth, to provide rehabilitation to people after stroke. ⁴ In order to identify new research published since the Cochrane review, we searched Medline using key terms related to stroke rehabilitation and telehealth or virtual care with limits on publication from 2019 to 2020. Table 2 (supplementary) summarizes trials included in the Cochrane review and subsequently published trials. The most common interventions tested were case management or advice (seven trials), motor retraining for arm, and/or hand function (seven trials). Other interventions included motor retraining for leg and/or balance and walking training (four trials), speech and language therapy (one trial), a package of rehabilitation (two trials), or interventions to address low mood (one trial). Overall, there is moderate-quality evidence that providing discharge support via telehealth is no different to usual care for measures of depression and quality of life; low-quality evidence of no difference between telehealth-delivered upper limb therapy; rehabilitation of activities of daily living or balance training compared with in-person therapy; and insufficient evidence about the effectiveness of mobility training delivered via telehealth. ⁴ Of note, only two trials monitored adverse events. Most trials were not adequately powered, meaning that the risk of Type II error, or false acceptance of no between group difference, could not be ruled out. One exception is the trial by Cramer et al. ⁵ comparing gamified, virtual reality arm motor therapy supported by synchronous teleconferencing and asynchronous feedback, with in-person therapy of matched dose. The telehealth-delivered arm therapy led to non-inferior improvements in arm motor function. This adequately powered trial is the only one to date to show that the telehealth mode of therapy delivery produced equivalent outcomes to in-person therapy.

In the absence of evidence specific to stroke rehabilitation, are there lessons to be learnt from trials in related fields? Two umbrella reviews of systematic reviews of interventions provided by telehealth^6,7 identified 13 reviews relevant to neurological rehabilitation broadly. The most common interventions tested were phone or SMS based, or used websites with education and/or games. Both reviews concluded that there was some evidence that rehabilitation provided by telehealth may be equivalent, or even superior to, in-person therapy. However, as with the Cochrane review, trials were generally small and inadequately powered for non-inferiority.

More inadequately powered trials will not advance the field. Equally, more reviews of small heterogeneous trials will be unhelpful as they inflate the, possibly spurious results, of no between-group differences. New trials are emerging (Supplementary Table 3). Searches of Clinicaltrials.gov and Australian and New Zealand Clinical Trials Registry identified 22 records of trials in progress or recently completed but not published (see Supplementary Table 3). Most of these are also small, underpowered non-inferiority studies, with the notable exception of four trials. Two of these are testing whether telehealth-delivered therapy to improve aphasia after stroke is superior to sham telehealth intervention (focussed on cognitive training; NCT0322826) or non-inferior to in-person therapy (NCT04682223). The third is investigating whether text messaging support for self-management after stroke will reduce hospital readmissions, compared with usual care (ACTRN12618001468213), while the fourth is specifically examining the impact of at home telehealth rehabilitation on arm motor function (NCT04620707).

How can we move the field forward?

While randomized, non-inferiority trials are the gold standard, they are problematic to conduct where the need for telehealth services is greatest. One way forward is to take interventions with evidence of effectiveness, when provided in person, and evaluate their implementation via telehealth; not as head-to-head trials to show non-inferiority with in-person delivery, but instead using robust implementation frameworks. This is beginning to occur. ⁸ This small, but robust study evaluated the telehealth implementation of an evidence-based ⁹ self-management program to improve arm motor ability. There are also several ongoing, relatively large implementation trials. For example, one trial in the United States of America is testing the implementation of the telehealth-delivered arm training intervention previously found to be non-inferior to in-person rehabilitation (NCT04657770; target n = 100). Another (NCT04440215; target n = 220) is investigating the implementation of a comprehensive package of stroke rehabilitation across Canada. The study by Yang et al. ⁸ , while small, is a good exemplar of telehealth implementation research. It used the Reach, Effectiveness, Adoption, Implementation, Maintenance (RE-AIM) ¹⁰ framework to evaluate implementation Template for Intervention Description and Replication (TIDieR). Other implementation science frameworks may be equally or more appropriate for future trials, depending on context. ¹¹ Other considerations for reporting trials include providing a detailed description of participant selection, particularly around factors likely to impact accessibility and compliance including age, disability severity, cognitive impairment, availability of caregiver support, and familiarity with technology. The intervention itself must also be clearly described using the TIDieR checklist, ¹² with particular focus on describing how and by whom interventions were supervised or monitored. We recommend the use of our taxonomy of terms to ensure all aspects of the intervention are succinctly described.

Telehealth and digital health care technologies also create opportunities to increase the recruitment to and efficiencies of randomized trials. There are few examples as yet of trials conducted entirely via telehealth, ¹³ but more are likely in the future. One of the limiting factors to telehealth-conducted trials is valid assessment of patient outcomes. The core outcome data set recommended by the Stroke Recovery and Rehabilitation Roundtable ¹⁴ consists of the National Institute of Health Stroke Scale (NIHSS), Fugl-Meyer motor arm and leg, Action Research Arm Test, timed 10 m walk test (10MWT), Modified Rankin Scale (mRS), and the EQ-5D. While the questionnaire-based assessments of outcome (Modified Rankin Scale, EQ-5D) can be easily administered virtually, the other performance-based measures can be problematic. Scoring of the Fugl-Meyer arm motor scale and the Action Research Arm Test based on video recordings is as reliable as in-person assessment, ¹⁵ but assessment via telehealth platforms has not been evaluated. There is some early research on adapting the Fugl-Meyer leg motor score for remote assessment which seems promising, ¹⁶ but may require specialist equipment. Measurement of the timed 10 m walk test is particularly problematic. While self-assessment following standard instructions and using a stop-watch may be possible, ¹⁷ such methods are prone to bias and may be difficult, and potentially risky, for many people after stroke.

It is clear that further research on establishing valid and reliable outcome assessment tools for telehealth is required. In doing so, it is important to first consider where agreement with in-person tests is required, or whether a new tool, specifically designed for remote assessment, may be more useful. There are promising early results from a small study (n = 13) of a new scale. ¹⁸ And another study is developing virtual reality–based arm assessments that can be conducted remotely (NCT04694833). Remote monitoring using wearable sensors and/or mobile apps has potential as both intervention and assessment tools.^19,20 However, what data are collected and how these are fed back to feedback to participants should be considered. Research in this area will be further boosted by the integration of artificial intelligence techniques for data processing, classification, and adaptive control.

Should we embrace telehealth as the new usual care?

Clinical guidelines updated prior to 2020 do not include specific recommendations for use of telehealth for rehabilitation. The only exception being the 2016 American Heart and American Stroke Association guidelines which recommend telehealth as a reasonable alternative to in-person assessment of communication impairment. ²¹ Updates to the Australian Living Guidelines include a new recommendation for telehealth to be considered an alternative approach to in-person care for rehabilitation services, particularly for those who cannot access in-person services. The Australian Guideline recommendation further specifies that only those interventions that have evidence of benefit when delivered in person should be considered for delivery via telehealth. The Canadian Heart and Stroke Foundation have published a Virtual Care Toolkit and associated evidence tables. ²² This Toolkit provides a comprehensive framework for decision-making about when to deliver virtual care, and how to evaluate its real-world effectiveness.

The long-term implications of the COVID-19 pandemic and the likelihood of future pandemics mean we do not have the luxury of waiting for more research evidence before implementing telehealth-based models of rehabilitation after stroke. However, implementation of telehealth should be undertaken thoughtfully, with caution and in an evidence-based manner. Evidence-based practice, by definition, involves the consideration of research evidence, patient values and preferences, and clinician expertise in decision-making. Resource implications and health economics are also essential considerations. Clinical Guidelines are the most accessible source of research evidence for clinicians. The Australian Living Guidelines are continuously updated in response to new research and published online, (https://informme.org.au/Guidelines/Clinical-Guidelines-for-Stroke-Management) making access to the most up-to-date recommendations for telehealth rehabilitation easily accessible.

The rights and preference of those receiving rehabilitation should be paramount. In an ideal situation, people with stroke would be provided with options to receive rehabilitation either in person or by telehealth. However, there will be circumstances in which telehealth is the only available option and is therefore likely to be preferable to no service. There is little published information about the preferences or attitudes to telehealth of people with stroke. One published report that drew together findings from a number of telehealth trials in Canada ²³ found that most people with stroke were satisfied with telehealth-delivered rehabilitation, as long as they were trained and supported appropriately. A qualitative study of the experience of people with stroke who completed a telehealth memory training program similarly reported high levels of satisfaction. ²⁴ Based on the expert opinion of our co-authors with lived experience of stroke, we present key factors that people with stroke feel should guide decision-making about telehealth service delivery (Supplementary Table 4).

Clinical expertise is the third pillar of evidence-based practice. If telehealth is to become a common mode of service delivery, health professionals need training in order to achieve its successful implementation. ²⁵ Without this training, it is not surprising that therapists prefer in-person models of delivery to telehealth. ²³ Such training needs to go beyond technical expertise in operating telehealth platforms and navigating privacy, legislative requirements, and billing systems. It is vital to understand how we can effectively translate the key therapeutic skills inherent in rehabilitation delivery—namely, observation of performance, provision of meaningful feedback, and development of patient rapport and the therapeutic relationship.

Some tools and resources have been developed to assist clinicians in decision-making about whether or not to use telehealth, and to support effective telehealth delivery (Supplementary Box 1). We propose the following simplified framework for decision-making. First, if there is no direct research evidence for telehealth delivery, the evidence base for the intervention as delivered in person should be considered. This should include the core components of the intervention, and how these can be effectively delivered via telehealth. Intensive motor practice is a core component of effective rehabilitation of motor function. Providing support for people with stroke to achieve this, and monitoring the amount of practice completed, is therefore crucial. Findings from qualitative research suggest the essential components for maximizing adherence to home-based exercises are education, patient–therapist rapport, and tailoring of interventions.^26,27

Second, we need to consider what aspects of rehabilitation care may be lost or compromised in using telehealth. This could include development of the therapeutic relationship, accurate measurement of outcomes, access to equipment, achievement of sufficient amounts of practice, and monitoring of adherence to practice. Third, safety is paramount. In the example of arm motor training tasks, safety is less of a concern. However, when implementing other evidence-based interventions, such training to improve standing balance, walking, or aerobic fitness training, the potential risk of falls must be managed. Indeed, concerns for patient safety may prevent the implementation of exercises with known efficacy for improving balance. Other safety considerations include patient privacy, the security of the telehealth platform, and the security of the data collected. Clear legislative guidance to address this is lacking in many countries.

If telehealth is ever considered a replacement to in-person services, it is important to ensure this does not disadvantage some groups of people with stroke: those with greater levels of disability, significant cognitive impairment, and/or limited social support. A retrospective review of the use of telehealth during the height of the COVID-19 pandemic in the United States highlights this issue. The review examined over 220,000 episodes between the second and third quarters of 2020. ²⁸ The use of telehealth rose from 6% to 10% of all consultations, but those who received telehealth were more likely to be young and live in larger metropolitan areas. Similar results have been found by others, with older adults, men, and Black Americans less likely to receive health care interventions via telehealth. ²⁹ Therefore, evaluation and measurement of reach and uptake is vital in any rollout of telehealth-based services. Tailored support and provision of Internet hotspots may be needed to ensure all eligible people can and are accessing the service.

What is the role of telehealth for rehabilitation in low- to middle-income countries?

Telehealth for rehabilitation after stroke holds particular promise in low-resource settings where there is immense stroke burden and limited rehabilitation services.^30,31 In a survey of 100 people with stroke living in Ghana, Africa, in 2017, 50% were under the age of 60 years and only 30% reported they had access to ongoing rehabilitation services. ³⁰ The growing access to mobile phone technology in recent decades has opened up opportunities for telehealth-based care in low-resource settings. For example, in India, a government initiative has helped digitize the economy and drive down data costs by more than 95% since 2013. As a result, monthly mobile data consumption per user is growing at 152% annually. ³² An estimated 251 million Internet users are in rural India (35% growth over the past few years). ³² This may not be the case for all low- and middle-income countries. For example, among people with stroke in Ghana, although 85% reported owning a mobile phone, only 35% owned smartphones ³⁰ and issues of bandwidth range and Internet speed present barriers to telehealth delivery in Brazil. ³¹

A handful of small feasibility studies on the use of telehealth in low- and middle-income countries show potential. Within Africa, recent studies conducted in Uganda ³³ and Ghana ³⁴ reported the feasibility and effectiveness (in terms of improved adherence to rehabilitation protocol and better patient outcomes) of family-centered and mobile phone–supported telerehabilitation programs after stroke. Similarly, small feasibility trials conducted in India have demonstrated the feasibility of various models of telehealth services for stroke and other neurological rehabilitation.^26,35 However, there have been no adequately powered randomized controlled trials (RCTs) to test effectiveness, or robust implementation studies of telehealth for stroke rehabilitation conducted in low- and middle-income countries.

Insufficient rehabilitation workforce, ³⁶ unsuccessful caregiver-led programs without continued health care professional supervision in India ³⁷ and rural China ³⁸ have clearly directed the focus of practitioners and researchers toward alternative modes, like telehealth, in low- and middle-income countries. The COVID-19 pandemic has further increased the need for telehealth-delivered rehabilitation services.^35,39 A survey of Indian health care practitioners found 50% were providing neurorehabilitation services via telehealth during the height of the pandemic. ³⁹ A consensus-based clinical practice recommendation has also been developed for telehealth for rehabilitation in low- and middle-income countries. ⁴⁰ However, as with implementation in high-resource settings, caution must be taken to ensure the best possible rehabilitation is provided. This includes ensuring adequate ongoing health care professional support, as well as personalized strategies to support people with stroke to actively engage in rehabilitation programs. In a randomized trial involving 58 people with stroke in India, personalized support strategies led to significantly better adherence to mobile phone–supported, home-based exercise programs, compared with no support. ²⁶

Developing effective telehealth models that have the ability to be rolled out at scale in low-resource settings will be challenging. Barriers including level of education, computer literacy, legal liability, and ethical issues such as patient confidentiality and data security will need to be overcome.