An exploratory sequential mixed methods evaluation was conducted between April 2023 and March 2024. A mixed methods evaluation was used due to the known complexity of pragmatic health services evaluation, enabling deep insights unavailable to the evaluation team from either quantitative or qualitative data alone [12]. The current study is reported with attention to the GRAMMS (Good Reporting of a Mixed Methods Study) reporting guidelines for mixed methods health service research [13] (Multimedia Appendix 1).

To address study aims, the evaluation comprised two phases: regarding phase one, discussion groups with clinicians working in community stroke services, the voluntary sector, and members of an established stroke-specific patient and public involvement (PPI) group; and regarding phase two, conceptual framework development and an online survey for clinicians and people with lived experience of stroke.

Using the sequential exploratory design, findings from phase one directly informed and were connected to phase two by shaping the conceptual framework and survey content. Quantitative and qualitative data were subsequently integrated through triangulation during analysis, interpretation, and reporting stages.

Eligibility criteria were necessarily broad to reflect the range of perspectives across community stroke services. Clinicians working in stroke services in the East of England and PPI members with experience of stroke rehabilitation in the region were eligible to take part. Participation in phase one did not preclude participation in phase two. All were invited to take part in both phases of the evaluation, with the option to take part in phase two regardless of participation in phase one.

For phase one, purposeful sampling was used to recruit clinicians with experience of delivering telerehabilitation, to inform the development of the conceptual framework and survey.

Responses from participants working only in acute services were not included in the analysis.

Recruitment for phase one was conducted via clinical and stroke-specific PPI meetings. At these meetings, the evaluation team presented this study and its aims and invited attendees to participate. Individuals interested in taking part were asked to contact the evaluation team by email to request a participant information sheet with full details of the evaluation and consent procedure (phase one). All participants had the opportunity to ask questions, and written informed consent was obtained.

To maximize accessibility for PPI members, this study was presented at two meetings held at different times of the day.

Fifteen community stroke teams were invited to the clinical meeting; services represented were located within two eastern integrated stroke delivery networks in England.

Phase two differed from phase one in that participation involved the completion of an anonymous survey online rather than enrollment in discussion groups. The survey was accessible via an online link or a QR code. Participation was voluntary, and an introductory page outlined the purpose of this study, data handling practices, and participants’ right to withdraw at any time before submission. Consent was implied through survey completion and submission. Participation was advertised at clinical and PPI meetings, as in phase one, but at a later date. To extend the reach across the region, the survey was also promoted at the East of England Stroke Forum and shared via two stroke charities.

Data from the discussion groups were analyzed using reflexive thematic analysis within a contextualist approach, enabling flexibility of combining multiple sources of data such as people’s experiences and practices [14,15]. This approach highlights the researcher’s active role in knowledge production. Codes are understood to represent the researcher’s interpretations of patterns of meaning across the dataset. Understanding evaluators’ backgrounds in relation to the phenomenon at hand is an important aspect of conducting reflexive qualitative evaluation. Here, the lead author (EC) is a chartered physiotherapist and has worked in NHS community and stroke rehabilitation services. EC conducted all the discussion groups. The principal evaluator (NJH) is a chartered physiotherapist, researcher, and innovation lead working in Higher Education, with extensive experience in NHS stroke rehabilitation service delivery, evaluation, and research. These professional perspectives informed the interpretation of the data, and ongoing reflexive discussion was used to consider how our assumptions and experiences influenced the theme development. NJH was present for 2 of 4 discussion groups. Participants were aware of the evaluators’ background when taking part in the discussion groups.

The reflexive thematic analysis was conducted by working iteratively through a six-phase process of data familiarization, generation of initial codes, generating themes, and reviewing, refining, and naming the themes [15]. According to Braun and Clarke [15], coding was conducted inductively, with codes generated from the data rather than a predefined framework.

Qualitative data were analyzed using an inductive coding approach, with codes generated directly from the data. Following coding, relevant published literature and clinical guidelines (including those from related areas of physiotherapy practice) were reviewed to check for potential gaps. This process did not lead to the identification of additional codes.

EC was responsible for all coding and generating themes. EC and NJH then worked collaboratively to reexamine and redefine themes, achieving richer interpretations of the meaning. Differences in interpretation were explored through discussion to deepen understanding and refine themes, consistent with a reflexive approach.

Quote identifiers in the results section include the discussion group number and the participant’s role (eg, DG1, OT for an occupational therapist in discussion group 1). Roles include: OT (occupational therapist), physiotherapist (physiotherapist), SLT (speech and language therapist), RA (rehabilitation assistant), SU (service-user with lived experience of stroke), and UC (unpaid caregiver).

Survey data were exported from Microsoft Forms into Microsoft Excel (Microsoft Corp) for analysis. Responses were screened for eligibility, and ineligible submissions (where rehabilitation or service experience did not take place in the East of England) were excluded. Missing data were reviewed; if a respondent left a nonessential question unanswered, their other responses were retained. No imputation methods were used to estimate missing values.

Where contradictory responses were identified, these were reviewed for internal consistency. In one case, a participant reported discontinuing telerehabilitation entirely but also provided detailed examples of its use. Their response was amended to reflect internally consistent use, in line with the examples given.

Descriptive statistics, including frequencies and percentages, were used to summarize demographic characteristics and closed-ended responses. Likert-scale items were summarized using the proportion of respondents selecting each response option.

For questions that allowed respondents to select multiple answers, percentages of respondents who selected each option as part of their responses were calculated.

The ranking data were analyzed by calculating weighted scores for each option based on frequency and rank position. For each response, items ranked in first position were assigned a score of 10, second position a score of 9, and so on down to a score of 1 for the lowest rank. Scores were then summed across responses to produce an overall score for each factor, with higher scores indicating greater perceived importance.

Integration took place at both the design (building) and analysis (merging) stages. Findings from the phase one discussion groups were used to develop a conceptual framework, alongside relevant published literature, and this informed the design of the phase two questionnaire.

After data collection, findings from both phases were considered together. This allowed comparison of perspectives across the two datasets and helped identify areas of agreement and difference.

Ethical approval for this evaluation was granted by the University of East Anglia Faculty of Medicine and Health Sciences Research Ethics Service (ETH2223-1521).

All participants in the phase one discussion groups received a participant information sheet, had the opportunity to ask questions, and provided full informed written consent.

For phase two, the online survey was anonymous. An introductory section outlined the purpose of the evaluation, ethical approval, and data handling. Completion and submission of the survey were taken as implied consent.

Discussion group participants who traveled to the venue were entitled to reimbursement of their travel costs, but participants were not offered financial incentives to take part.

All data were securely stored on encrypted servers in compliance with regulatory requirements.

This large-scale service evaluation successfully mapped current models and experiences of telerehabilitation delivery in community stroke services by engaging relevant stakeholders, including frontline clinical staff and people with lived experience of stroke, using a sequential mixed methods approach (evaluation aim 1). It also identified practical lessons and enabling factors to support the sustained integration of telerehabilitation into routine stroke rehabilitation pathways, in line with current policy-drivers (evaluation aim 2). As a result of in-depth interrogation of data from multiple sources using recognized methods [15], we developed a conceptual model to underpin and guide the successful sustained use of telerehabilitation in community stroke services (Figure 1). The model coalesced around three themes: risks and benefits; individualized care, and support for staff.

In considering the possible risks and benefits of using telerehabilitation in community stroke services, a complex interaction between the need to reduce burden on services while still building meaningful relationships and understanding personal environments was exposed. While it was clear that immediate practical needs might be met, for clinicians, developing a sense of the person in their social environment and being effective and safe was generally viewed as more challenging using telerehabilitation than face-to-face approaches. Similarly, people with lived experience of stroke generally disagreed that telerehabilitation was comparable to face-to-face delivery. This latter point contrasts with findings of a recent systematic review of randomized-controlled trials investigating factors influencing telerehabilitation delivery in stroke [11], reporting overwhelmingly positive feedback for access to and interaction with therapists via telerehabilitation. More promisingly, in the current evaluation, on deeper interrogation within this theme, many clinicians felt able to adapt and develop rapport, considering themselves able to successfully build relationships with service users using remote communication. Additionally, some evidence of forward-planning and improved self-management by people with stroke was apparent as they prepared for telerehabilitation sessions, in line with previous work exploring possibilities for self-management through the use of technology [17]. Clinicians also recognized clear efficiency gains when telerehabilitation was used appropriately. Reduced travel time, flexible scheduling, and the ability to conduct joint sessions remotely enabled more responsive working and, in some cases, increased therapy intensity without additional service burden. For people with lived experience, albeit a small sample size here, this flexibility was viewed positively when it translated into being seen more often or more quickly. These findings suggest that although telerehabilitation can make some aspects of relationship-building more difficult, it may also offer a pragmatic way of increasing rehabilitation dose within constrained workforce capacity. Further, carefully designed research is needed to understand the nuanced characteristics that underpin this paradigm shift in delivery, in concert with the views of people with lived experience.

It is unsurprising that the need for an approach to using telerehabilitation shaped by individual needs, preferences, and circumstances was a core finding of the evaluation; that a “one-size fits all” approach was not suitable. The importance of personalized care and shared decision-making throughout patient pathways is well-established in modern health services, with improved outcomes and experiences linked to people’s involvement in shaping their care [18,19]. Here, we found that the small group of people with lived experience of stroke who wanted to be fully involved with decisions about use, and that clinicians considered multiple individual factors when deciding whether to use telerehabilitation. These not only included clinical factors such as consideration of cognitive, sensory, or severe communication challenges—often seen as “red flags” for telerehabilitation to be avoided—but decisions were also shaped by factors such as people’s personal preferences and their perceived confidence with technology. A deeper exploration of that latter point revealed similarities with previous work on health technologies in stroke and neurorehabilitation—that the challenge is not to dismiss but to support people in taking advantage of the potential benefits of technological approaches without making assumptions about age or digital literacy [20,21]. In fact, here we demonstrated that age was least important in influencing decision-making by clinical teams, but access to appropriate equipment was considered critical to digital participation and reducing inequalities of provision. This theme demonstrated that thoughtful shared decision-making, supported by access to appropriate equipment and clear communication, can help ensure that remote delivery meets the needs of diverse individuals and avoids reinforcing existing inequalities.

A prevailing body of knowledge indicates that, for both technological advancement and other wide-scale service innovation in stroke and neurorehabilitation, supportive organizational cultures are central to sustained change [21,22]; the evaluation findings presented here reflect that understanding. Specifically, clinicians identified the value of a collaborative, multidisciplinary service approach and a need for meaningful, tailored training for themselves and people with lived experience. Telerehabilitation training needs were exposed on how to work with people to adapt their goals, how to assess suitability for the approach, and how to communicate effectively online—in essence, this approach is about much more than the simple mechanics of opening a laptop or smartphone and making a call, if outcomes for both individuals and the service are to be optimal. As recently captured in an extensive scoping review [23], successful use of telerehabilitation requires a range of competencies from both people with lived experience and clinical teams. Telerehabilitation is a new model of care, implemented at pace in response to urgent need, and as such, workforce transformation through tailored development and a needs-led approach to training is required to embrace its full potential.

The work met an identified need for reporting of the practicalities and challenges of use of telerehabilitation in NHS services [7] using a recognized approach to combining multiple data sources. This was a time-limited pragmatic service evaluation in a large geographical area of the United Kingdom (East of England) with community health provision across a mix of rural, suburban, and city-based populations. We acknowledge the limitation inherent in evaluating provision only within the UK NHS, but we suggest that key findings on enablers of sustained use of telerehabilitation in community stroke services have potential for application in other systems. The evaluation was supported by the existing integrated stroke delivery network, PPI advisory group, and we acknowledge that our group of participants with lived experience of stroke and their caregivers was small and therefore, while providing considerable insights, cannot represent the views of the wider service-user and caregiver populations. Despite this, we adopted an inclusive model of shared learning from service users and providers at every opportunity throughout. Where discussion groups included people with stroke, the facilitators endeavored to ensure their voices were heard. Similarly, we recognize the risk of bias inherent in surveys on a particular rehabilitation approach—that those with strong views and experiences in either direction might be most likely to respond. We made an attempt to negate this through a variety of recruitment strategies in collaboration with CD as the Stroke Quality Improvement for Rehabilitation lead. Finally, it is possible in pragmatic, rapid mixed methods evaluation that meaningful connections are lost due to limited integration or dominance of one dataset; this evaluation used recognized frameworks for study design, integration, interpretation, and reflexive approaches to minimize this risk.

Digital services have clear potential to address pressing unmet rehabilitation needs as demand ever-escalates, though understanding of implementation and sustained use in a variety of service contexts has, up to now, been lacking [23]. This evaluation has made initial steps to enhance that understanding for telerehabilitation in community stroke services and should provide a platform for further primary research.

Together, the findings presented here suggest that telerehabilitation works best not as a replacement for in-person care, but as part of a responsive, blended approach grounded in individual needs and service-user choice. The evaluation demonstrates that telerehabilitation has the potential to meet people’s needs, while supporting service responsiveness and therapy intensity in pressurized clinical environments. A nuanced approach to implementation that includes appropriate training and carefully integrates the individual needs of service users and staff alongside a supportive organizational culture is required to support successful, sustained use.