Telerehabilitation assessment domains and process in chronic diseases: a scoping meta-review

Background Telerehabilitation (TR) interventions are receiving increasing attention. They have been evaluated in various scientic areas through systematic reviews. However, there is a lack of data on how to standardize assessment and report on their domains to guide researchers across studies and bring together the best evidence to assess TR for chronic diseases. Aims and Objectives The aim of this study is to identify domains of assessment in TR and to qualitatively and quantitatively analyze how they are examined to gain an overview of assessment in chronic disease and understand the complexity of TR interventions. Methods A scoping meta-review was carried out on 9 databases and gray literature from 2009 to 2019. The keyword search strategy was based on "telerehabilitation", “evaluation", “chronic disease" and their synonyms. All articles were subjected to qualitative analysis using the Health Technology Assessment (HTA) Core Model prior to further analysis and narrative synthesis. Results We that least commonly and comprehensiveness to extend this assessment

the same rigor as a new drug or treatment program to prevent decisions about future deployments of ICT in the health sector from being determined by social, economic, and/or political circumstances rather than by robust scienti c evidence [27]. Thus, we must be able to evaluate TR interventions while they are being designed, developed and deployed [27]. Thus, in assessing TR, an extensive appraisal including these different dimensions is needed in each phase of the technology's life cycle [27][28][29][30].
The aim of this scoping meta-review [31] was thus to systematically map recent research to understand the assessment of TR from a comprehensive perspective. First, in order to identify all the domains of TR assessment for chronic diseases, we used a comprehensive evaluation framework with a multidisciplinary approach: the Health Technology Assessment (HTA) Core Model [32]. The HTA includes several domains and thus appears to be the most complete framework. In addition, we identi ed the phases of TR interventions in which assessment occurs by mapping out the content of the reviews and grouping the phases of intervention with similar objectives, activities, or results. On this basis, our 2 research questions were as follows: 1. 1. What domains of evaluation have been identi ed in the literature on TR for chronic disease? 2. 2. What are the assessment domains of the different phases of TR?
Finally, the ndings allow us to present a novel way of examining the assessment of TR interventions and could provide a reference and information for policymakers, clinicians and researchers regarding the development of assessment guidelines.

Methods
This study is based on a new method, the scoping meta-review [31]. This method combines the scoping review and meta-review methods. Scoping reviews entail reviewing the emerging literature to provide an initial indication of the size and nature of the available literature on a particular topic [33,34]. Metareviews involve synthesizing evidence from a set of systematic reviews [35,36]. We rst considered performing a simple scoping review, considering that the eld is diverse and complex [37]. However, during the initial extraction of the data, we noticed the existence of numerous systematic reviews on various aspects of TR assessment. Thus, it seemed relevant and feasible to undertake a scoping review of the systematic reviews related to TR assessment approaches [31]. The advantage of relying on systematic reviews is that they can provide a solid and reliable synthesis of work in the eld [37].

Protocol
We followed the guidelines of Levac et al. (2010), updated from the initial work of Arksey and O'Malley (2005). In addition, we wrote the protocol using the "Preferred Reporting Items for Systematic reviews and Meta-Analyses Extension for Scoping Reviews" (PRISMA-ScR) [38]. Levac et al. (2010) suggest that the protocol should not be designed as a rigid tool and strictly applied. Thus, in our study, the protocol was used as a guide, and we followed it when necessary [39]. The nal protocol is not publicly available; however, it can be made available upon request to the corresponding author.

Eligibility Criteria
The methodological approach of the scoping review allowed for the identi cation and alteration of the inclusion and exclusion criteria as articles were selected. Inclusion and exclusion were performed rst on the basis of article selection through the "title" and "abstract" lters and then by reading the articles in full.
The inclusion criteria are studies that (1) are in the systematic literature review format, (2) address at least one of the NCD categories of cardiovascular disease (cardiovascular accidents), cancer, chronic respiratory disease, diabetes, obesity, identi ed on the basis of the prevalence and importance of the common behavioral risk factors (smoking, poor diet, sedentary lifestyle, harmful use of alcohol) (according to the gures provided by the WHO), (3) address TR in the sense of the de nition given in the rationale (see the introduction), (4) include features of the de nition of rehabilitation presented in the rationale, (5) contain at least one intervention offering physical activity as part of a multidisciplinary approach, (6) date from 2009 to 2019, (7) are in French or English, and (8) use an adult population 18 + years old.
The exclusion criteria are studies (1) that present interventions without technology or limited to a telephone follow-up approach, (2) that are written in a language other than English or French, (3) that present a single study or are opinion papers, draft syntheses, abstract/conference proceedings (oral presentations and posters), chapters, discussions, letters, books available electronically, and theses, (4) where it is impossible to identify the type of intervention performed, (5) that deal with an intervention with technology limited to the physical activity dimension alone (without multidisciplinary approach) or that do not contain a physical activity dimension, or (6) that deal with methods/requirements without real evaluation related to TR or for which it is impossible to locate the full text.

Information Sources
The eld of TR must take into account the growing ICT evolution in health (connected objects, mobile devices, etc.), so we included studies published only between January 2009 and October 2019. We conducted extensive literature searches in the electronic bibliographic databases most likely to contain the type of study we are looking for. The databases are multidisciplinary, covering elds from computer science to health science: MEDLINE (PubMed), Web of Science, Cochrane Library, ABI, Business Source Premier, PsycINFO, Science Direct, Academic Search Premier, and SPORTDiscus. We conducted additional searches in the gray literature (a) by consulting the reference lists of the included studies and (b) by searching repositories of gray literature: CADTH, Directory of Open Access Journals (DOAJ), Occupational Therapy Systematic Evaluation of Evidence (Otseeker), International Prospective Register of Systematic Reviews (PROSPERO), OpenSIGLE (OpenGrey), and the New York Academy of Medicine Library's Grey Literature Report. The research strategy was planned and carried out through structured team discussions and in consultation with a university librarian so that the strategy could be re ned in light of the initial results. We exported the nal search results to Zotero (5.0.95.1, Roy Rosenzweig Center for History and New Media, Fairfax, Virginia), a bibliographic database. This software facilitated the management of the research, particularly the identi cation and removal of duplicates.

Search -Identi cation of Relevant Studies
We identi ed the keywords through, on the one hand, the medical subject headings (MesH) providing the controlled vocabulary for MEDLINE/PubMed and, on the other hand, other keywords, which we call free vocabulary. Free vocabulary was added based on the expertise of the different team members but also by reading the abstracts in 1st intention if it seemed relevant and necessary. The search strategy was based on the terms "telerehabilitation" AND "evaluation" AND "chronic disease" and all their synonyms (see Additional le 1). Each database was searched individually. The keyword search strategy, based on the use of the Boolean operators AND and OR as well as ti(title) and ab(abstract), is described below.

Selection of Sources of Evidence
To increase consistency among the 5 reviewers, pairs were created to independently evaluate article titles and abstracts for inclusion in the study. Evaluators met at the beginning, midpoint and end of the process to discuss issues and uncertainties related to the selection of potentially relevant studies and to reevaluate and re ne the research strategy if necessary. We then independently reviewed the full papers for inclusion. Disagreements over study selection and data extraction were resolved by consensus and discussion with other reviewers to make a nal decision on inclusion if necessary.

Data Charting Process
To begin, the rst author developed a data table. We used a conceptual framework to guide the data extraction (Additional le 2). Subsequently, the other coauthors discussed and validated this table to determine the variables to be extracted. Once this rst version was nalized, three members of the evaluation team tested the table by independently collecting data on 3 articles to share their perspectives concerning the dimensions to be collected. Next, two of these reviewers conducted data collection on 10% of the corpus of selected articles and discussed the results, continuously updating the data charting table in an iterative process. We then carried out a second calibration exercise testing the % agreement, with a predetermined level of agreement (70-80%) [38]. The concordance determined by the Kendall concordance coe cient (W) in SPSS software was greater than 80% (mean: 87%; Kendall's W = 0.8697).
The rst researcher thus nalized the coding alone, and any disagreements and questions were resolved through discussion between the two reviewers. Moreover, by charting the content of the reviews and grouping those with similar objectives, activities, and/or outcomes, thematic analysis could be used to determine whether the reviews focused on certain phases of the technology life cycle rather than others.

Data Items
The data extracted from the articles are as follows: author(s), year of publication, location of study, design of review (narrative review, descriptive review, scoping review, meta-analysis, systematic review, theorical review, etc.), title, type of pathology, elds of technology (m-health, e-health, etc.), de nition of technology, tools associated with technology (SMS, apps, web, etc.), end-users, number of studies included (study design and participants), whether type of evaluation allows for consideration of and focus on phases of the technology life cycle (design, pretest, pilot study, randomized trial, postintervention), key ndings, and critical appraisal of researcher (if applicable). For each review, we extracted the presence and number of evaluation domains based on the HTA model (see the following paragraph). In accordance with the PRISMA-ScR, we did not perform a quality assessment or quality evaluation, as this is not essential for scoping review methodologies. Thus, the methodological rigor of the published articles was not a criterion for inclusion or exclusion.
Critical Appraisal of Individual Sources of Evidence: The HTA (Healthcare Technology Assessment) Core Model Framework [32] We needed an analytical framework, a multidisciplinary approach that could include all domains of TR assessment, to know what was being evaluated and to break down the silos of the assessment dimensions. For this purpose, we chose the Health Technology Assessment (HTA) Core Model [32] from among the evaluation frameworks available in the literature [40][41][42][43][44] because it suggests what kinds of information one can nd in an HTA report, and its de nition encompasses the dimensions of multidisciplinarity and comprehensiveness. The de nition of a health technology assessment is as follows: "a multidisciplinary process that summarises information about the medical, social, economic and ethical issues related to the use of a health technology in a systematic, transparent, unbiased, robust manner" (p13) [32]. The structure of the information collected is as follows: the domains of assessment (the broad framework within which the technology is considered), the topics of assessment (more speci c considerations in one of the domains), and the issues raised (even more speci c considerations on one of the topics that may be similar to research questions in scienti c studies). The structure of the model is based on the combination of these three points to de ne the different assessment elements and facilitate a shared understanding of what belongs to HTA. Additional le 2 below provides an overview of the 10 domains described in the HTA core model: health problem and current use of technology, description and technical characteristics of technology, safety, accuracy, clinical effectiveness, costs and economic evaluation, ethical analysis, organizational aspects, social aspects, and legal aspects.

Methods of Handling and Summarizing the Data
A qualitative synthesis of the included studies is conducted to chart the literature on the domains of TR assessment. The data are summarized using descriptive tables of the categories developed from the HTA framework. Additionally, a qualitative inductive and content analysis approach allowed us to bring out other elements of TR assessment (completing the existing framework).

Selection of Sources of Evidence
The search of the 9 databases generated 7412 results ( Fig. 1). After elimination of duplicates, 5306 publications remained. The review of the titles and abstracts led to the exclusion of 5174 publications, leaving 132 publications requiring screening on the basis of the full text. Full text screening helped to remove an additional 54 publications, leaving 78 articles. The most common reason for excluding citations during full-text screening was that the studies did not include physical activity in their interventions (n = 24). Subsequently, 2 articles were added from the gray literature and manual research. A total of 80 publications remained (Fig. 1), all of which focused on one or more domains of TR assessment and required further analysis.

Characteristics of the Reviews Included and Reports According to the EUnetHTA Template
Year of publication and geographical distribution - Table 1 shows the number of reviews included by year bracket between 2009 and October 2019. Few reviews were published between 2009 and 2013 (n = 14, 18%). Most of the articles were published from 2014 onwards, and more than one-third of the articles (n = 26, 33%) were published from 2018-2019, the last two years studied. The majority of the reviews (n = 31, 39%) were from Europe, followed by North America (n = 25, 31%), Oceania (Australia and New Zealand) (n = 13, 16%), Asia (mainly China) (n = 9, 11%) and South America (n = 2, 3%). MA, meta-analysis; QSR, qualitative systematic review Pathologies concerned -The main categories of NCDs were cardiovascular diseases (heart attacks), cancers, chronic respiratory diseases (such as chronic obstructive pulmonary disease or asthma), obesity and diabetes. Fifty-four articles (68%) focused on a single pathology, and 26 (32%) focused on > = 2 pathologies. More than half of the articles dealt with diabetes (n = 45, 56%), followed by cardiovascular disease (n = 38, 48%). One-third of the reviews were on chronic respiratory diseases (n = 25, 31%). A quarter (n = 20, 25%) of the reviews addressed cancer, closely followed by obesity (n = 19, 24%) (see Table 1).
Areas of intervention and de nitions -In our research, the term "telerehabilitation", according to the de nition given in the rationale, was found under different names in each of the included reviews. The most-used terms were rather generic: "m-health" (n = 33, 41%), "e-health" (n = 16, 20%), and "telehealth" (n = 16, 20%) or, more rarely, "web-based intervention/rehabilitation" (n = 6, 8%), "e-health/m-health" (n = 3, 4%), "digital health intervention" (n = 2, 3%), and "telemedicine" (n = 2, 3). The term "telerehabilitation" appeared only twice (3%). Systematic reviews used different de nitions corresponding to the eld of intervention stated in the research. Despite the sometimes disparate de nitions, the numerous reviews nevertheless provided an evaluation of different studies (with different interventions) that may correspond to the de nition of TR given in the rationale. In this way, some authors raised the issue of the di culty of determining the element of effectiveness of the TR interventions evaluated [45].
Types of associated technological tools -We classi ed the technological tools mobilized into several categories (see Additional le 3). In most studies, a combination of several tools was identi ed to enable the interventions to be carried out [41,[46][47][48]. Mobile and internet/website applications were most commonly used, with 51 articles (17% for each). Short message system tools followed, with 47 articles (14%). Often, in addition to these rst three tools, phone calls (n = 39, 11%), digital devices (e.g., connected objects) (n = 36, 11%), and emails (n = 29, 9%) were added. Less frequently, it was also possible to identify the following tools: videos/images (n = 22, 6%), videoconferences (n = 14, 4%), and social networks (n = 14, 4%). More rarely, we found that studies used personal health reports (n = 4, 1%) or other technological tools, such as logbooks, virtual reality, or digital libraries.
Completeness of the reports according to the EUnetHTA template -Given the number of sources included in the scoping meta-review, the relevant data from each source are provided in Additional File 4. On average, we found a total of 3 HTA domains evaluated per review. Table 2 shows the number of HTA domains appearing in the reviews. Brie y, the most represented domains were social aspect, in 79% (n = 63) of the reviews, and clinical e cacy, in 66% (n = 53). Ethical analysis and safety aspects were both evaluated only in 3% (n = 2) of the reviews studied, and accuracy was not represented in our data. In this section, we present the TR domains that review authors evaluated. To understand these domains, we mapped them using the principles derived from the EUnetHTA HTA framework (see Methodology). After extracting the domains during the qualitative analysis, we classi ed them into 9 categories: social aspects, health problem and current use of technology, description and technical characteristics of technology, costs and economic evaluation, organizational aspects, legal aspects, ethical analysis, and safety. Table 3 shows these domains and the key aspects of their measurement. Major life areas: Technologies are seen as opportunities to nd similar user communities and share experiences among peers [49]. Effect on psychological distress [50], stress management, fatigue [50,51], knowledge of treatment and chronic disease [52][53][54][55]. Encouraging but varied effects on self-e cacy [53,56,57]. Effects on behavioral changes in PA, diet, medication adherence, and smoking [53,[57][58][59][60][61][62]. Individual: Ten reviews examine the acceptability of the intervention [63,64] mainly to the patient as an end user [58,65,66], and 15 reviews study satisfaction [58,67,68]. Facilitators and individual barriers are also studied [45,69]. Communication: Evaluation of the usability of technologies during the development process [25,46,58,63,70]. Need for targeted technology [65] and stimulation of user engagement, motivation and involvement over time [68,71,72], and the quality of patient-caregiver interaction [ [20,65,81].

Health problem a current use of technology
Target condition: Differences in the effectiveness of the intervention according to the various targeted pathologies, possible differences from one pathology to another [50,65]. Example of type 1 diabetes and type 2 diabetes [20,76]. Utilization: differences in use between countries or a lack of education in low-income countries [66,84]. Identi cation of the applicability and acceptability of telerehabilitation in primary care, general practice and hospital settings [85]. Other: Only one study focused on evaluating the actors involved in the design of the technology (i.e., a team of IT developers) [86].  [89].

Costs and economic evaluation
Unit costs: Related to the unit costs of the resources used (e.g., technology acquisition costs or the cost of speci c actions) [49,65,70,84,86,95,96]. Outcomes: Health cost outcomes by type of telerehabilitation intervention [97], compared to a control group [68], to prevent, predict or minimize exacerbation [98]. Cost-effectiveness: Intervention that can be cost-effective under certain conditions [56,95]. But little studied, "urgent" need to carry out controlled and homogeneous trials [99].

Domain assessed
Key subjects of measurement

Organizational aspects
Process: Monitor care outcome processes such as maintenance of the behavioral effects of the intervention [65], clinic attendance, the effectiveness of the chronic disease surveillance system or the compliance of tools used to improve clinic attendance (e.g., SMS reminders) [95]. Interest in having a multidisciplinary team trained in motivational feedback [100]; fund technology-oriented studies and encourage proposals from interdisciplinary groups of researchers [65]. Structure: The effects of the implementation of interventions on hospital admissions, the use of health resources [72,101,102], clinical workload and work ow, and dependence on technology for work [103]. Management: Interest in proposing multiple models of patient management (e.g., integrating alternative models), based on evidence, responding to the needs and pro le of patients [68,77]. Take into account the intentions of future caregivers to integrate technological tools into their practice [65,89].
Legal aspects End-user: Identify the various target populations [53,69,77,104] and those that are poorly studied [48,53] in order to carefully examine the possibility of generalizing new modalities of intervention and their potential dissemination. Privacy of the patient and authorization and safety: Describe procedures to ensure the security and storage of private data [64] and identify if problems occur in the private sphere [67]. Legal regulation of novel/experimental techniques: Adapt the development of new mobile applications to regulations (e.g., medical devices) [89]. Regulation of the market: Identify whether reimbursement of intervention systems is possible and by whom [65].

Ethical analysis
Principal questions about the ethical aspects of technology: Spreading the use of technologies (e.g., ethical challenges of privacy and data security) [25].

Safety
Technology-dependent safety risks: Identify potential problems with the reliability and validity of information entered into the technology by the patient or caregiver; identify the number of adverse effects of interventions in patients [63]. Use-or userdependent safety risks: Identify potential complications that may arise due to certain functionalities (e.g., misinterpretation of information sent) [65].

Additional Non-HTA TR Domain
The HTA framework was not developed speci cally for TR, and during the inductive and thematic analysis and by comparing results between reviewers, we found that 95% of the reviews (n = 76/80) evaluated the "interventional aspect". This interventional aspect combines the characteristics of the interventions or their functionalities and the application of recommendations and theoretical foundations to construct these interventions. Therefore, we decided to add this assessment domain to complement the HTA framework.
Application of recommendations and theoretical foundations for behavior change -Clinical recommendations suggest that ongoing behavioral support is necessary for lifestyle changes to be sustainable [58]. Many reviews (n = 31; 39%) present intervention characteristics based on speci c theories/conceptual frameworks for designing and optimizing TR interventions [53,84,94]. The behavior change theory (BCT) developed by Abraham & Michie (2008) is the most widely used behavior change theory in technological applications (n = 13/31) [41,60]. The most mobilized BCTs are "goal setting", "self-monitoring of behavior", "information about health consequences", "social support", and "feedback and monitoring". The transtheoretical model [105] and social cognitive theory [106] are the next most applied (n = 10/31). Many other theories are mentioned more sporadically, such as self-e cacy theory, the theory of planned behavior -reasoned action, social ecological theory, social support theory, the selfmanagement model, self-determination theory, the health belief model, and cognitive behavior theory.

Phases of the Telerehabilitation Assessment Process
To answer our second research question, we focused on how the assessment was conducted in the distinct development phases of TR: design, pretest, pilot study, randomized trial and postintervention. This led us to develop the telerehabilitation assessment process (Fig. 2), which illustrates the accumulation of evidence by crossing the assessment domains with the distinct development phases of TR. The domains of assessment (i.e., health problem and current use of technology, description and technical characteristics of technology, safety, clinical effectiveness, costs and economic evaluation, ethical analysis, organizational aspects, social aspects, and legal aspects) vary in each phase. The results show that assessment is mainly carried out in the pilot study and randomized trial phases. For example, during the pilot study phase, the focus of assessment shifts primarily to the social aspect, followed mainly by clinical effectiveness. On the other hand, assessment is rarely carried out in the design, pretest, and postintervention phases. When a TR intervention initiates with the design phase, the decisions are made based on the evaluation of the description and technical characteristics of technology, social aspects, costs and economic evaluation, organizational aspects, legal aspects and ethical analysis. The health problem and current use of technology and safety domains appear in the pretest phase. Although there are still few reviews of the postintervention phase, this comprehensive evaluation process can be used to gradually accumulate evidence that could be used to make future decisions.

Discussion
This scoping meta-review was conducted to identify the different domains of TR assessment for chronic diseases and provide a comprehensive view of TR assessment through the analytic framework of HTA.
The results indicate that many systematic reviews are generally focused on a limited number of assessment dimensions.
In identifying and summarizing the main domains of assessment, we highlighted the multidisciplinarity and comprehensiveness of the assessment of TR. Our study shows that 9 out of the 10 domains composing the HTA framework have been explored by TR reviews (social aspects, clinical effectiveness, description and technical characteristics of technology, health problem and current use of technology, costs and economic evaluation, legal aspects, organizational aspects, safety, and ethical analysis). This result re ects the relevance of this framework for our speci c analysis.
Much of the focus centered on the domains of social aspect and clinical effectiveness. Together, they represent 48% of occurrences, though they constitute only 20% of the HTA domains (n = 2/9 domains). This re ects an imbalance in the assessment of the different domains. To date, research in these two domains has relied primarily on randomized controlled trials (RCTs) to assess TR. The performance of an RCT is considered the "gold standard" in research [107], and the RCT is a unique approach of achieving lifestyle changes in patients with chronic diseases [65]. Despite this major interest, some researchers have questioned their usefulness given the complexity of assessing TR interventions [107]. Our results also show that the least frequently occurring domains were safety and ethical analysis. Although our review highlights some promising emerging results that may help commissioners, developers, and users manage risks and improve patient safety [63,65], several studies have shown that mobile medical applications (apps) could compromise patient safety [108,109]. Future research could develop a risk framework that users, developers, and other stakeholders can use to assess the likely risks posed by speci c apps in a speci c context [110]. Finally, concerning ethical analysis, despite the fast-paced growth of TR, only a few articles propose suggestions to practitioners for addressing ethical challenges such as acquiring compliant software, receiving training, creating informed consent procedures, and using an ethical decision-making model [111].
Regarding our choice to mobilize the HTA Core Model, we evaluated its relevance and operationality with respect to TR. This framework had not yet been applied in the eld of TR, so we sometimes had di culty classifying some of the data using the HTA domains. Indeed, there was ambiguity regarding some items that could be classi ed under multiple domains at the same time. For example, in this framework, the description of the social aspects domain includes the effects on behavioral changes in physical activity and diet. These elements could also be classi ed under clinical effectiveness for a health or rehabilitation expert. We can also take a critical view of the results with respect to the frequency of the appearance of certain domains (e.g., the social aspects domain is present at a greater frequency than the clinical effectiveness domain). Furthermore, with regard to the model, it would be interesting to consider its supplementation or comparison with other validated frameworks. For example, the literature proposes many (more or less comprehensive) approaches to e-health [40], m-health and even telemedicine [38,[105][106][107] assessment to assist decision makers who want to introduce and use this technology.
Moreover, we identi ed an additional non-HTA domain, the interventional aspect, which de ned several attributes of TR intervention assessment. The majority of published reviews (n = 76, 95%) examine the different characteristics of interventions that engage the patient and foster the success of TR to promote behavior change and positive health outcomes. A number of reviews provide encouraging evidence about BCTs and their bene ts for the improvement of physical activity outcomes [60]. In contrast to this literature, a recent meta-review highlighted the need for better implementation tools that support patient engagement and identi ed the necessity of optimizing the design of the self-management resources included in or with guidelines [112]. Thus, a variety of theories offer insight into how patients' perceptions in uence their behavior and can be used to design and then evaluate self-management guideline tools.
For our secondary objective, we aimed to identify the phases of TR interventions in which assessment occurs. Our results showed a marked interaction between assessment domains and the distinct development phases of TR. This allowed us to highlight which domain was assessed at which phase, suggesting that it is inappropriate to assess all domains in a single phase. This TR assessment process can capture comprehensive, dynamic and complex evidence, crossing the various domains of assessment with the development phases of TR. While many TR assessments are still quite disconnected from each other and thus fail to create a synergic effect in TR research efforts, this classi cation could truly structure research in the eld, similar to the "phases" of drug discovery and development [113]. Furthermore, through comparison with existing literature, we observed that this TR assessment process is distinct from the e-health evaluation model of Enam et al. (2018) [114]. For example, when e-health intervention is initiated at the design phase, the decisions are made solely based on the assessment of the technological and cost domains of technology development, whereas in this scoping meta-review, they also include social aspects, organizational aspects, legal aspects and ethical analysis. Therefore, our process proposes an additional speci cation not present in the evaluation model of e-health interventions in general. This TR assessment process could become cumbersome because of high resource consumption, but it is not a prescription, just a way to show the progression of evidence in TR applications in a reliable manner.

Limitations
The main strengths of this review are the use of the scoping review methodology, which enabled coverage of a very broad range of topics; the comprehensive search strategy developed; and the rigorous quality assessment of each review by two independent researchers. However, there are a number of limitations that must be highlighted. First, a scoping meta-review can only report on literature that has been included in published reviews, meaning some recently published primary research might not be included. Another limitation was that our electronic database searches may have missed relevant citations. This is potentially due (1) to restriction of the search to English and French language publications, and (2) to certain documents that may have been omitted, unknowingly and unintentionally, although we have included and analyzed many journals in this scoping meta-review. Additionally, these TR reviews included only the ve major groups of known chronic diseases that represent the highest rate of premature mortality, limiting the generalizability of the results. It would be interesting to see if this TR assessment process could be extended and applied to other chronic pathologies that require TR, such as osteoarthritis or stroke [115,116].
Finally, this scoping meta-review shows that semantically, the remote delivery of rehabilitation is not homogeneous: the terms used include "m-health", "e-health", "telehealth", "web-based intervention/rehabilitation", "digital health intervention", "telemedicine" and "telerehabilitation". There is a use of multiple 'de nitions' and an apparent lack of solidarity in de ning TR. How, then, do wecollectively -de ne TR? The confusion extends to other aspects of the TR domain [117]. According to Scott et al. (2013), it seems important to resolve the semantic issues around "e-health strategies" and identify barriers to TR, such as profession-centric nomenclature. Further discussion can then ensue to ensure that the diversity of TR is understood and that the appropriate mix of speci c solutions is brought to bear in response to de ned health needs.

Conclusions
This scoping meta-review reported on a large number of reviews that focused on assessing TR intervention for chronic diseases. By proposing and using a comprehensive assessment framework for TR, our results highlighted 10 assessment domains and a list of the main related aspects. The different domains mobilized for assessment are not all studied with the same degree of interest. Furthermore, we showed that each of these assessment domains could appear at different phases of TR development, whereas current research generally focuses on one or a few assessment dimensions. These main contributions allow us to enrich this literature on the assessment of TR and propose new crossdisciplinary and complete method for the assessment of TR interventions.
Due to the challenge of integrating TR into the management of patients with chronic diseases, this framework could guide future studies in obtaining a comprehensive view of the assessment of TR. Thus, improved validation of evaluation methods could facilitate the transferability of results among similar studies and bring together the best evidence to assess TR interventions across a broad range of domains.