Assessing the implementability of telehealth interventions for self-management support: a realist review

Background There is a substantial and continually growing literature on the effectiveness and implementation of discrete telehealth interventions for health condition management. However, it is difficult to predict which technologies are likely to work and be used in practice. In this context, identifying the core mechanisms associated with successful telehealth implementation is relevant to consolidating the likely elements for ensuring a priori optimal design and deployment of telehealth interventions for supporting patients with long-term conditions (LTCs). Methods We adopted a two-stage realist synthesis approach to identify the core mechanisms underpinning telehealth interventions. In the second stage of the review, we tested inductively and refined our understanding of the mechanisms. We reviewed qualitative papers focused on COPD, heart failure, diabetes, and behaviours and complications associated with these conditions. The review included 15 papers published 2009 to 2014. Results Three concepts were identified, which suggested how telehealth worked to engage and support health-related work. Whether or not and how a telehealth intervention enables or limits the possibility for relationships with professionals and/or peers. Telehealth has the potential to reshape and extend existing relationships, acting as a partial substitute for the role of health professionals. The second concept is fit: successful telehealth interventions are those that can be well integrated into everyday life and health care routines and the need to be easy to use, compatible with patients’ existing environment, skills, and capacity, and that do not significantly disrupt patients’ lives and routines. The third concept is visibility: visualisation of symptoms and feedback has the capacity to improve knowledge, motivation, and a sense of empowerment; engage network members; and reinforce positive behaviour change, prompts for action and surveillance. Conclusions Upfront consideration should be given to the mechanisms that are most likely to ensure the successful development and implementation of telehealth interventions. These include considerations about whether and how the telehealth intervention enables or limits the possibility for relationships with professionals and peers, how it fits with existing environment and capacities to self-manage, and visibility-enabling-enhanced awareness to self and others.


Background
The ubiquitous growth in technologies and devices for illness management has been accompanied by numerous studies evaluating the effectiveness and implementation of telehealth interventions. Despite this growing body of evidence, it is difficult to predict which technologies are likely to work and will be used in practice. Existing evidence points to the need for technologies to align closely with wider networks that facilitate collective efforts to enhance individual self-management [1]. However, there is a still a need to identify the mechanisms underlying the normalisation of technologies in the life worlds of patients [2][3][4]. The objective of the evidence synthesis presented here was to review and integrate evidence about telehealth interventions, including the provision of support and care at home and monitoring patient status at a distance using audio, video, web-based, and other technologies [5]. The focus was on identifying the structuring factors likely to promote implementation that should be taken into consideration when developing and deploying telehealth interventions.
Telehealth has been found to have a positive impact on social support [6,7], compliance [8], education [7,9,10], behaviour change and better self-management [9,11,12], and reduced burden on the individual and services [13,14]. The latter is particularly the case where interventions are telephone, computer, or internet-based [15,16] because they are simple to use and familiar [13,17]. Overall, patients appear more positive than professionals about using telehealth [17,18]. In a related field, Ziebland and Wyke [19] identified a number of domains through which online patients' experiences could affect health (finding information, feeling supported, maintaining relationships with others, affecting behaviour, experiencing health services, and visualising disease). The gaps identified by previous reviews confirm the need for more in-depth understanding of how telehealth interventions mediate to improve health and become embedded in people's everyday lives [18]. While many evaluation studies focus on the technology itself or the individual recipient in explaining (un) successful implementation in a field where technologies are being developed and introduced at a rapid pace [20], there is often little generic guidance about the likely impact in terms of acceptability and appropriateness that can be considered by those at the point of devising new technologies. While it is relevant to understand telehealth interventions in context, retrospectively analysing their contribution to patient care, it is useful to find ways to identify core features that can assist with telehealth intervention design and deployment. Here, we are interested in exploring interventions that could be delivered as part of self-care support for long-term (chronic) health conditions at the interface between populations and primary care. Our aim was to move beyond more quantitative, trial-based evidence (the primary resource that currently informs development) to include qualitative evidence in thinking about telehealth. Drawing together research conducted as part of two separate but related projects, the Healthlines Study (http://www.bristol.ac.uk/healthlines/) and the EU Wise programme [21], we posed the question: how (through what processes and mechanisms) can telehealth improve the health and well-being of people with long-term conditions (LTCs)? This paper addresses implementation of evidence-based practice of telehealth which crosses an interface between formal clinical settings and peoples' everyday lives. The paper brings together two studies and aims to identify and then test the core mechanisms underpinning telehealth interventions by understanding how telehealth can improve the health and well-being of people with a range of LTCs.

Theory development and theory testing
Theory development and theory testing were conducted in two stages. In stage 1, we used a realist synthesis approach to identify characteristics of telehealth interventions that had a positive impact on the facilitation of health and wellbeing of people with LTCs. The team of researchers (AR, CP, and AC) revisited the three reviews on telehealth interventions for people with LTCs undertaken for the Healthlines Study. These included a) meta-review of 16 systematic reviews centering on the topic of LTCs, published between 2005 and 2010; b) a meta-review of 20 reviews on depression, published between 2005 and 2010; and c) a metasynthesis of 29 qualitative studies of interventions for LTCs, published between 2000 and 2010. These three reviews were conducted to provide the evidence base for a telehealth intervention targeting people with long-term conditions [17]. Our searches were of Medline, Embase/AMED, PsycInfo, Web of Science, DARE, and the Cochrane Library. The first review was designed to identify existing, broad, evidence about interventions for LTCs. The second two reviews were conducted to plug gaps in knowledge, specifically about depression (one of the target disease groups for the Healthlines intervention) and to capture the qualitative evidence which had not hitherto been synthesised. The interventions included telephone-based, telemonitoring and computerized, and web-based forms of telehealth. These included real-time (synchronous) and asynchronous (that is, email) interventions offered with and without healthcare professional input. Further methodological detail necessary to reproduce these reviews is provided in a report [17] and a working paper from the study [18] available here http://www.bristol.ac.uk/healthlines/. This literature provided an overview of the evidence for telehealth interventions aimed at adult patients in home settings. Overall, the reviews provided information about outcomes ranging from specific changes in clinical features (for example, reduced hypertension) and treatment compliance as well as on quality of life. Some also provided information on financial savings and many discussed acceptability and satisfaction with interventions. These outcomes demarcate the success or otherwise of an intervention-in effect they tell us if it 'worked'. While these earlier reviews improved our knowledge about which interventions worked and which did not, they could not address the question why and through what mechanisms they worked. Thus, subsequently, a realist synthesis approach was chosen because of its suitability for conceptual development and theory building. In analysing the literature, we followed Pawson's seven stage model: identify the question and clarify the purpose of the review, theory elicitation, search the evidence, appraisal, extract the results, synthesise findings, and draw conclusions and make recommendations. Realist synthesis was an iterative rather than a linear process where we compared findings from different studies, looking for examples which challenged, refined, or supported the theories identified [22,23].
We re-read the review literature to inductively identify potential mechanisms associated with successful interventions. We discussed these amongst our team, re-examining the literature for confirmatory and dis-confirmatory evidence and refining our ideas in a process that resembled qualitative thematic analysis. Three explanations or theories, which suggested how telehealth worked to improve health, were identified from this analytical process. These emerged from developing propositions about what might be important to the success of interventions in the context of the evidence we had read and synthesised. Our emergent theories about the core mechanisms underpinning successful telehealth interventions from this review work were as follows: 1. Relationship. Relationships or connections between people (patients, peer groups, and/or lay and professional carers) are a necessary component of telehealth interventions. 2. Fit. The extent to which a telehealth intervention can be integrated within everyday life and health care routines determines the success of deployment/ adoption. 3. Visibility. Systems which increase the visibility of symptoms or health problems to self or others impact positively or negatively on the adoption of telehealth interventions depending on, for example, whether patients might want anonymity or not.
The three mechanisms that we describe are relevant for assessing the likely implementability of existing interventions and for developing new ones that are more likely to be successfully implemented. These can be used as a basis for developing a set of sensitising concepts when consideration is being given to introducing new telehealth interventions. Stage 2 provided an opportunity to test our propositions against additional and more recent qualitative studies. We cast the net wide in stage 2 to ensure that we could include a range of different technologies, deliberately including newer innovations (for example, apps) that had not been examined in the earlier literature and populations that would help confirm or disprove our conceptual hypotheses about the mechanisms that underpinned successful telehealth interventions. If the mechanisms could be shown to stand up to this testing, then we could have more confidence in their application to any new telehealth interventions. In this paper, we report the theory testing results.

Methods
For this review, we searched papers published between 2009 and 2014 which focused on COPD, heart failure, and diabetes. These three conditions were chosen as exemplars of LTCs that had high incidence and growing prevalence, which often co-exist with other conditions and to which there are increasing aspirations to manage through telecare interventions [24]. Papers for review were identified from searches in PubMed and the Web of Science. We looked for qualitative papers focusing on telehealth, telecare, self-management, diabetes, heart failure, COPD, and related conditions and risk factors. Papers were included if they reported studies conducted in OECD countries. While most of the papers selected for this review described successful interventions, this was not one of our selection criteria. This was because our interest was in understanding the processes that contributed to successful interventions, and therefore both successes and failures could potentially be equally revealing of underlying processes.
Three reviewers (AER, AK, and IV) independently reviewed abstracts to agree on papers for full-text retrieval. Where there was doubt about a paper, the full-text paper was retrieved. Thirty one papers were reviewed for quality and fit using the CASP criteria [25] and after exclusions, 15 papers were included (see Figure 1). We designed a data extraction form that included a) background information about each paper (study setting, rationale, aims, research questions, sampling, data collection, and analysis), and b) key findings and themes identified by the authors. Each paper was analysed using the three mechanisms identified in stage 1, after which the papers were systematically compared (see Table 1). All three reviewers reviewed the full papers. The findings that were independently arrived at were then discussed and refined in an iterative process. The findings and their interpretation were then discussed within the whole team of researchers (AER, AK, IV, CP, AR, and AC).

Relationships
The importance of relationships with professionals for mediating the introduction and uptake of self-care support has been noted in long-term condition and selfmanagement literature and refers to elements that professionals bring to a therapeutic approach with patients. Empathy, acceptance, and the therapeutic alliance between therapist and client are considered essential to effective psychological therapies [26] and involvement of a health professional to provide legitimization of strategies for self-management [27]. We examined the literature to see why and how relationships were important for the success of a telehealth intervention, exploring the contexts, aspects of relationality (for example, continuity, communication, and rapport), differences between peer and patientprofessional relationships, and whether telehealth technology augmented or substituted for face to face/ personal contact.
Evidence about telehealth reshaping existing relationships: open ended, contextually embedded, dependent on additional support Telehealth interventions are sometimes framed in terms of a threat to professionals and as leading to a deterioration of previously valued relationships and roles. Segar et al. [28] note a sense of protectiveness about maintaining boundaries around established remits of managing long-term conditions when new technologies are introduced. However, the presumption that technology is a threat to existing relationships when it comes to the actual use of specific technologies for self-management is not borne out by patient accounts in the studies we reviewed. Moreover, the introduction of new technologies establishes new sources of relationality and is therefore better viewed as restructuring existing relationships in a process that is open rather than leading to a pre-determined outcome. For example, the frequency of contact and approachability of professionals engaged in telemonitoring led to building new bonds of       The video conversation was experienced as much more personal and intense than telephone conversation rather than resembling the 'real encounter'. Telecare was made interesting to the users as a necessary step for users to put in effort and make it work. Attempts to extend the use of the telekit to a wide range of users did not work and there was a high level of dropout. This was because: there was not enough space in the house, not wanting to have an extra computer, plans for moving house, not wanting to make the disease too central to their lives. Also for people who were in better health and went back to work telephone calls were disruptive and using the telekit was not practical. The telekit was based on Apple application and this created problems of usability as most people were familiar and/or had access to windows and thus had to later adapt their knowledge; the volunteers providing the training also did this on windows. The telekit obtained two different identities that were in conflict: guaranteeing the effect of the treatment and providing a window to the world. These required different ways of 'taming' the technology. The hardware was available for 3 months only and after that only training was given, but it was not seen as the       15. Perceptions of successful cues to action and opportunities to augment behaviour triggers in diabetes self-management: qualitative analysis of mobile intervention for low income Latinos with diabetes; Burner ER, Menchine MD, Kubicek K, Robles M, Arora S Five focus groups with 24 people; USA; we examined nuances of motivation, intention, and triggers to action effected by TExT-MED (trial to examine text messaging for emergency department patient with diabetes), an mHealth intervention tailored to low-income, urban Latinos with diabetes Low-income Latino patients will accept text messages as a behavioural intervention. This mHealth intervention acts as a behavioural trigger rather than an education platform. Personalization is an opportunity to enhance these cues to action Yes Recommended programme to family and friends The appearance of support, although aware this is an automated service Participants said they would feel more cared for if the messages were personalised, including the name and the specific time for taking medication (programme used the same messages across participants to Found behaviour cues and medication reminders easy to follow Sense/perception that the programme has worked for them to take control over diabetes and make the behaviour changes they believed to be of benefit for them 'better controlling our life, our way of living'. Medication reminders were the most salient and prominent benefits mentioned. This included specific prompts. Healthy  trust and introduced an intermediary function between patient and GP.

'I'd say you get better [service] because if
[telemonitoring professional's first name] comes on the phone and she'll say "I think you're needing to speak to the doctor", she's just giving me a warning that she's going to get the doctor to phone me. And they'll either say "Well, I think you're needing to have some antibiotics" or "I think maybe we should pop over and just see you.' [29] Not only can technology build trust in the absence of pre-existing relationships, the introduction of technology can potentially offer more control for users through the introduction of distanced but positively perceived relationality. For example, self-management support (SMS) was preferred to a telephone conversation because it provided more control over managing relationships.
'I liked the project and the follow-up. I could send an SMS whenever I wanted. I got an answer within half an hour. I especially liked the SMS -in the Netherlands, where I lived prior to this, I knew I could call, but I like the message system a lot better' [30].
The indeterminacy as to how telehealth interventions can reshape existing relationships is evident in relation to their impact on the division of labour between users and professionals through increasing the input of users, professionals, or network members, or changes through building a shared network of responsibility. For example, Fairbrother et al. [29] report that practitioners' attempts to encourage patient participation in self-directed medication received a mixed response, with some expressing anxiety and trepidation at the prospect of being required to exercise greater personal responsibility. Many users stated that they liked being telemonitored because they felt reassurance arising from what was perceived as the provision of continuous practitioner surveillance and support. By contrast, telehealth interventions can lead to more active engagement by users, professionals, and network members which allows for the better tailoring of illness management. In a study by Dinesen et al. [31], healthcare professionals reported that a tele-rehabilitation programme allowed the handing over of responsibility for managing their illness at the same time patients reported that family and network members became more engaged in helping them to integrate the activities into their everyday routine and to maintain the focus on exercise as a normal part of their everyday lives.
Telehealth as a substitute to the role of health professionals: legitimising and strategic role There was evidence, in our sample, to support the notion that telehealth can work without professional input. There was also evidence that telehealth interventions with no professional input were more likely to be successful if they were individually tailored: that is, including the name of the user, the specific time for taking medication, and the negative consequences of the disease specific to the stage at which each one of them was [32].
Cases where telehealth intervention could work in the absence of any professional input were rare in the papers we reviewed [18]. Thus, while professional relationships did not appear to always be essential they could improve acceptability and outcomes even where this professional input was minimal for example facilitating or linking people to using telehealth. Professional 'lite' input might also work through a process of users feeling reassured or construing professional monitoring as part of a partnership where responsibility is shared.
'In a way it was a relief thinking that I should ignore my own thoughts on getting a doctor or something like that. This organisation was going to get hold of a doctor if their readings showed I needed a doctor.' (Female patient, 47 years old, post-installation) [33]. 'If I knew that someone was looking at this information on the other end at BMG, I would definitely continue to use it.' [34] The role of relationships derived from peer support: need for tailoring and extending existing networks There was limited evidence to support the necessity of peer relationships as the defining ingredient of the success of an intervention. The presence of generic peer support was likely to be insufficient and implicated the need for tailoring to different needs and circumstances. Tailoring related to age and gender appears to be the most salient features of interventions. However users are likely to feel more comfortable if they have control over the technology, how it is used, and the circumstances under which they might disclose (or not) their involvement with a telehealth programme, thus avoiding potential embarrassment: 'if this popped up, I think people would laugh at me' [31].
Interventions that opened up opportunities for extending existing networks seemed likely to be successful and related to the inherent capacity of interventions for flexibility in support in relation to context and individual needs. Some newer social media -for example, those using video -offered richer contextual awareness enhancing people's support networks and contributing to successful interventions [35].
'Oh yes, seeing each other [over the webcam] is different from talking on the telephone. It is much more personal. And much cosier! For instance with Peter's wife, when I talk to Peter she comes along to chat for a bit. And she was there [at the clinic] a lot too, same as my husband. And he chats along too. Or the guys chat together: 'Gosh, how are you' , or; 'We are in town this Saturday, will you be home?' [35] ' Ah, over the webcam you can see the smoking cigarettes in the background. When someone says: I am doing fine! You can see from the way somebody keeps his or her body that they are not fine at all. When you can see that, you can say: your shoulders are too high, are you ok?' [35] Possibilities for extending networks are likely to be of high value in contexts where members of one's intimate network, such as family, might act as a barrier by being too cautious about making changes to aspects of illness management (for example, medication) [31]. Additionally, in a programme delivered to rural communities, Guilcher et al. [36] found that in a context of stigma the programme provided a safe and supportive environment, encouraged building new connections and avoiding isolation. The success of the programme was dependent on support by peers extending beyond the meeting site and on having supportive and knowledgeable group leaders, who created good group dynamic.

Fit
When we examined the literature, we found that the extent to which a telehealth intervention worked was in part dependent on the extent to which there was a fit with patients' needs, skills, and daily life. In addition to the importance of fit, we also found that context played a facilitative role in providing the necessary conditions for workability. For example, home blood pressure readings were evaluated by users as being more 'natural' than surgery readings (and therefore more accurate), as people were reportedly more relaxed in domestic settings and the readings were taken more carefully and under controlled conditions [37].

Fit with patient-defined needs and environment
There was evidence that interventions which were perceived as 'fixing' a problem from the patient's point-ofview might fare better than others which did not. For example, people with COPD found a tele-rehabilitation programme convenient as they could do their exercises at home at any time, could try new exercises, become more involved in the management of their illness, and could adjust the training program to their home environment and situations in their everyday life [31,38]. When successfully adapted to people's needs and environment, telehealth interventions had the capacity to enhance accessibility of health care for those who might otherwise not access traditional face-to-face care or those who were geographically isolated living in remote areas with poor transport options.
'You know if something was wrong I'd get a phone call from the surgery… they'd write a prescription and I'd get it sent to the chemist and then I'd get it delivered direct. Because if I'm unwell that's one thing I have to face is that long walk to [the surgery], because there's no bus direct from here and, you know, when I'm unwell' (Female patient, 66 years old., post-installation) [39].
The use of telehealth could also offer convenience in accessing care opening up the possibility to obtain and share data and communicate with healthcare professionals and other patients independently of time and space [31]. Fit with needs and context can be seen as an ongoing and reflexive process. Continued utilisation is likely to be constant review and dependent on users continuing sense of added value.
'If it gets it wrong, you would automatically get really irritated by it…I think the risk if getting it wrong would be really annoying and I'd probably delete the app' 'It's quite easy to lose interest really because it is quite an effort and nobody wants to spend all their life writing down what they want on their phone' [40]. There was also evidence that interventions were likely to be unsuccessful where users found them disruptive and impractical. This was the case for people in better health who went back to work, had plans to move house, did not have enough space in the house to accommodate the telecare equipment, or did not want to make the illness too central to their lives [35]. Other barriers to using telehealth were described by professionals in relation to limited functionality and the lack of interoperability between telemonitoring patient information systems and the existing systems used [29]. Patients also require time before they become comfortable with using technology, and experienced problems in accessing internet, and getting transport to the site of the teleconference [35,39].

Fit with patient skills and capacity
There was evidence that patients' capability in technology use increased their propensity to benefit from interventions. Simple technologies appeared effective suggesting that technologies that are already used routinely in everyday life may be easier to use to deliver telehealth. This is most clearly illustrated when delivering simple and easy-to-follow messages such as medication reminders or specific prompts such as healthy living challenge messages.
'the challenges you send us. One imagines that when I see the message, and when I read the challenge, those are my challenge for the day…I read them and say, I have to do this'. I motivate myself, like if I am going to go for a walk.' [32]. Telehealth interventions may require some level of basic training and need to be tailored to existing user skills and physical capacity.
'If this is for people like me, there should be adjustments for functional limitations. Here they talk mainly about the possibilities, about people who are mobile etc., but the people who cannot get out of the house, for those adjustments should be made.' [41].
Engaging with technological interventions could provide new possibilities for learning through enhancing an existing skill set. 'We teach them to write e-mails. And there was one man he had a son who lives in Japan. And in the meantime he has become a grandparent, but he had never heard of the internet. So he got this internet connection at home, and his son sent him his email address. And I helped him typing the e-mail address, and when he got an answer he got pictures and saw his grandchild for the first time. Really, if you see this older man looking at a picture with tears in his eyes' [35].

Fit with the structure of daily life
The appeal of a programme might be enhanced through embedding it into a set of familiar relationships and cultural practices. This was for both professionals and users. For example, the uptake of tele-ophthalmology by an aboriginal community improved when its delivery was made to resemble practices familiar from encounters with traditional healers and supported by an aboriginal nurse and/or a liaison assistant [42]. Professionals expressed the view that telemonitoring added to workload as it required additional time to checking online data, dealing with additional administration, and increased communication and interaction with patients [33,39]. These findings were consistent with the wider literature on professional involvement [43].
The notion of 'fit' operates within an assumption about the relative structuredness of everyday life and a value attached to it. There are however, situations and contexts where this might not be the case. There is evidence of how the use of new technology brings a new structure to disrupted everyday lives. For example, telecare can aid restructuring the discharge from other care services, for example, discharge from clinics can mean potential disorientation when back in domestic settings. This can be potentially averted by the deployment of 'telekit' which helps to structure the day [35] and to 'better controlling our life, our way of living' [44]. Technology can also help adapt to changing needs and attitudes to tele-rehabilitation. Ongoing sustainability of use requires not only about access to and investment in technologies -for example, broadband upgrades and knowledge of new 'apps' [33], but also a level of interest in making telehealth work [40].
Motives for engaging, learning, and enacting new skills fade when there is ongoing stability in measurements or if the patient is constantly symptomatic (this is a particular problem in COPD). Adjustment and interest is made possible if technology use can be extended to incorporate other network members, for example, Dinesen et al. [31] reported that family and network became more engaged in a tele-rehabilitation program of activities and helped the patient to maintain the focus and integrate the activities into their everyday routines.

Visibility
We identified earlier that telemonitoring of symptoms and vital signs were perceived by patients to have positive impact on outcomes. This led us to consider that this might have reinforcing and incentivising functions (for example, reporting vital signs encouraged selfregulation, or the belief that healthcare professionals were monitoring information encouraged patients to follow instructions, or enabled healthcare professionals to respond to patients' needs quickly). Visibility is likely to work best for some physical conditions and diseases -diabetes, heart failure -more than others. This 'making visible' to self or others seemed to have a powerful role in empowering users and opening up possibilities for engagement with their network members. Indeed, the surveillance aspect of technology was experienced as a reassurance of continuous practitioner engagement and support [33]. Nevertheless, visibility was not always welcome because some patients were worried about data getting into the hands of a third party and found some phone capabilities intrusive [40].

Visibility linked to knowledge, motivation, and empowerment
The use of glucose-monitoring web application improved visualisation of blood sugar profiles and increased patients' general awareness of their blood sugar and its changes.  [38] reported that users found it helpful to know their oxygen saturation, to learn their 'normal' range by identifying telemonitoring data trends over time, and linked indicators to level of physical effort and the onset of vital signs [31].
'I'm okay from 87% (oxygen saturation) upwards and I never get better than 92. Even when I'm very well, I never get better than 92. But I go out and about and I do what I need to do and I manage it by walking' [38] 'Seeing my data on the web portal gives me a better understanding of how to exercise and interpret the development of my symptoms when I experience the onset of an exacerbation.' [31] There was some evidence that involving patients in monitoring promoted feelings of empowerment and made users feel more knowledgeable, but it also gave them a sense of being actively involved in the management of their illness. 'I felt quite happy to be involved… instead of just being a vegetable that sat back and swallowed things.' [33] Visible reminders encouraged participation and motivated patients to continue training and competing with themselves especially when values improved over time [31]. Telemonitoring helped patients and carers to recognise the onset of an exacerbation and allowed them to better address these problems.
'You'd think you would find it easy to tell when you're ill but it's only afterwards that you know you are not well. But this technology is really brilliant.' (Female patient, 66 years old, post-installation) [39].
By contrast, patients whose measured values (blood pressure, pulse, weight, spirometry, and saturation) were stable over time, and thus were unable to observe any connection between measured values and physical training, were indifferent towards the intervention [31]. Patients with COPD who were constantly symptomatic, and thus with no standard against which to benchmark their symptoms, found a telemonitoring intervention lack utility.
'It asks "Is it higher than normal?" I don't know what the normal's supposed to be. So I don't know what … Sometimes it's not all black and white.' (Male patient, 69 years old, post-installation) [39].

Visibility linked to the engagement of others
Systems which encouraged accountability such as an expectation that patients would check their blood glucose levels frequently, coupled with feedback from the healthcare staff facilitated through technology, also worked well. For example, diabetes patients reported that 'when you have a date [upcoming videoconference] you are more likely to do something' [9]. Support by peers or professionals also enhanced visibility and perceived personal accountability. For example, through interaction with healthcare professionals and other users of the programme, patients with COPD became more aware of symptoms, learned to understanding measured values, and became aware when it was necessary to contact a doctor and seek treatment [31].
Visibility also led to user learning and restructuring of relationship with professionals. Jones et al. [37] reported that many users were surprised at how much their readings varied, which led them to question whether a GP should adjust their medications only after taking a single reading: 'I was amazed how much they varied. That was very educational. I mean okay if there's a crisis or something, you expect your blood pressure to go up, but I could take them just sitting there and it was just amazing the difference in them.' [37] Discussion In a rapidly developing field where a range of influences relating to the uptake technologies have been associated with discrete telehealth and telecare technologies, this review extended and deepened our understanding of the likely core and generic mechanisms underpinning successful interventions. These are important to identify in order that new designs can take account of this upfront and from the outset of development. The realist synthesis offered three possible mechanisms that make interventions 'work'. Stage 2 review process tested these, confirming them and allowing us to enhance our conceptualisation. In terms of 'relationships' , the evidence suggested two processes which enable successful telehealth interventions: relationships provide support (professional, peer, clinical, and social) for behaviour change, and relationships provide opportunities for professional feedback which reinforces positive or required behaviour change. This suggests that interventions which enable connections and contact, notably between patients and professionals can facilitate support and reinforcement necessary for behaviour change. Where telehealth interventions limit or remove the relationship between patients and professional, other opportunities to support and reinforce behaviour may be necessary. In terms of 'fit' , the literature pointed to the importance of acceptability and ease of use of telehealth interventions for patients and professionals. Telehealth can increase accessibility of care for some populations. Simple technologies appeared to work as well or better than more complex ones, and there are some patient groups who are less able to use some technologies (notably the Web) [11,45,46]. Fit also relates to the extent to which technologies disrupt existing environments and ways of managing. The third proposition centred on 'visibility' , which is how telehealth care makes an illness or condition apparent to the self and others. There is a connection between the visibility that technologies promote and the capacity to self-manage. Visibility brings enhanced opportunities to share and engage information and tasks with others in a person's network. Visibility facilitates and mediates knowledge and motivations which are inextricably linked to the actual tasks of self-management. The evidence suggested that visibility operated by enabling feedback which reinforces positive or required behaviour change; by providing incentives, reminders, and behaviour prompts for action; and by inducing negative feelings (fear) regarding surveillance, stigma, and punishment. Thus, visibility is not only about relations to others (anonymity, surveillance) but also refers to visibility in relation to self, via enhanced awareness and knowledge [40].

Limitations
The approach adopted in this review has advantages over quantitative systematic reviews as qualitative studies offer access to understanding and testing the underlying mechanisms through which telehealth interventions operate. The limitations of this review are that the concluding picture only offers a broad outline of the mechanisms involved in telehealth interventions, which needs further validation and testing in different contexts and in relation to other conditions. Future work needs to also examine whether and how the relative importance of each of the three mechanisms might vary depending on, for example, type of condition, multi-morbidities, structure of personal networks, and broader social and physical environment. Such insights could contribute to building a richer and more nuanced conceptual framework of relevant mechanisms and feed into developing robust methods for telehealth development, implementation, and evaluation. Such methods could be used in addition, or as an alternative, to the currently dominant trials and quantitative techniques.
We recognise that using a wider set of concepts in our search strategy might have brought papers that merited inclusion. However, while this might have potentially added further nuance to our findings, the extensive review process that led to generating the three key concepts and the papers reviewed as a way of testing them gives us confidence in the validity of our findings.

Conclusion
In this review, we tested processes that contribute to the successful implementation of telehealth interventions. The three mechanisms that we describe are relevant for assessing the likely implementability of existing interventions and for developing new ones that are more likely to be successfully implemented. The review draws attention to key aspects of assessing interventions likely implementability. These can be used as a basis for developing a set of sensitising concepts when consideration is being given to introducing new tele-health interventions.
Our review has three implications for developing and the future successful implementation of telehealth interventions based on the upfront consideration given to the mechanisms that are most likely to ensure whether and how the telehealth intervention enables or limits the possibility for relationships with professionals and or peers, fits with existing environment and capacities to self-manage and visibility enabling enhanced awareness to self and others which feeds into motivation and facilitates actions. If the intervention removes or replaces relationships, then other mechanisms for support and reinforcement may be necessary to effect behaviour change. Given the apparent importance of relationality, it seems likely that patients and professionals might resist or reject interventions which threaten or limit these opportunities. Second, successful telehealth interventions are those that can be well integrated into everyday life and healthcare routines. Interventions which enhance or improve access to care (by, for example, enabling access or making it more timely) are more likely to be acceptable to patients. Ease of use is important for adoption of technologies. Our synthesis suggests that the intervention comparatively simple technology (for example, telephone) makes for easier access and use. The intervention should be designed so that it offers minimal disruption to patients' lives and professional routines. Third, the design of the telehealth intervention should address the issue of visibility. How it does this may depend on the condition and patient group involved. Monitoring systems can offer opportunities for visible feedback and prompts to actions which serve as reinforcement of behaviour change. This may be especially important if relationships are not fostered by the intervention (that is, monitoring may be used to mitigate the loss of relationship with healthcare professional). While this review identified positive aspects of visibility, the earlier reviews we conducted indicated that visibility may also have negative dimensions. Specifically, patients with mental health may wish to remain anonymous when using the system [45,47]. The design of the telehealth intervention should consider if and how symptoms and signs are made visible by the system and how these are responded to by the technology, the patient, and the healthcare professional. Further research might need to identify factors that may help maintain these technologies over the long term, which is currently underresearched.