RPMTS Positioning in the Healthcare landscape
Even though many reviews could not neatly fit in one healthcare setting, 14 out of the 26 systematic reviews (SR) included remote monitoring systems positioned in post-hospital care settings. As can be seen in Figure 3, most of the above systems had been deployed to monitor chronic conditions, previously diagnosed within hospitals and had been deployed in the context of continuity of care including detecting signs of deterioration or improvement in chronic disease, treatment or rehabilitation, patient’s advice, support, education or training, medication adherence and cost reduction in hospitalization.
Ten articles [4], [12], [18], [29], [30], [31], [32], [33], [34] and [35] dealt with interventions that could be clearly classified as falling into the preventive, pre-clinical or hospital, emergency and-/or primary care settings. The majority of the above articles, except [30], [31] and [34] combined the above setting with other settings such as the hospital or post-hospital monitoring. And as can be seen in Figure 3, four articles [4], [12], [18] and [29] were both in the preventive and primary care settings without the involvement of any hospital or post-hospital monitoring.
For the 30 primary articles (PA) considered, 12 articles [3], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45] and [46] were classified as falling into the preventive, pre-clinical or pre-hospital, emergency and or primary care settings of the healthcare landscape. Eight of the above mentioned 12 articles [36], [38], [39], [40], [41], [42], [43] and [46], combined the above setting with other settings such as the emergency, hospital or post-hospital monitoring. Five articles [3], [37], [40], [41], and [43] discussed interventions in which, preventative measures were implemented in primary care settings and two additional interventions [44] and [45] were set in the educational, preventive, pre-primary settings. Therefore, as demonstrated in Figure 4, a total of 7 articles were both in the preventive and primary care settings.
Levels of integration within traditional healthcare systems or facilities
Integration here refers to the degree to which an intervention facilitates existing clinical work or improves existing clinical workflows [5], [34]. A significant number of systematic reviews (nine) did not contain information about the extent to which discussed interventions integrated or intended to integrate with /- into traditional healthcare systems. Of the 26 reviews, 9 [4], [9], [12], [24], [29], [47], [31], [48] and [49] provided information to suggest or demonstrate that above integration was either achieved or at least attempted.
Furthermore, in the majority of above 9 cases, ‘integration’ simply meant communication with one or more healthcare professionals via phone or video conferencing and not necessarily integration into clinical workflows. This spread can be viewed in Figure 5.
In the remaining 8 reviews [2], [5], [11], [18], [34], [35], [50] and [51] discussed interventions were partially integrated into traditional healthcare systems and highlighted challenges which hampered complete integration into clinical workflows. For example, one review [34] identified two key barriers to integration: the ‘diversity of available technologies’ and ‘lack of comprehensive guiding framework for standardizing data collection and integration’. One review [2] pointed to the use of non-scalable and silo solutions which suffer from the absence of interoperability and clinical acceptance to facilitate user engagement and self-management of chronic diseases. Other reviews highlighted additional concerns affecting integration with traditional healthcare systems.
One of the above reviews [5] indicated that healthcare practitioners view some effects of mHealth on aspects such as their credibility and autonomy as impacting their acceptance of such tools and systems; while another [35] highlighted the lack of integration of community-based HIS in formal HMIS (without complete integration, there are duplicative efforts in data collection, analysis, and reporting), and the lack of technical capacity of community workers. Finally, review [51] observed that despite the huge research effort on remote care technology, there has not been a sufficient number of successful interventions which have gone past the research environment and broadly taken up and routinely used in clinical settings.
With regards to primary articles, the picture was similar. Eleven articles did not provide sufficient details to determine the extent to which, considered interventions integrated or intended to integrate into traditional healthcare systems. As demonstrated in Figure 6, in 10 of the 30 articles [20], [37], [41], [42], [52], [53], [54], [55], [56] and [57], there was sufficient information to establish that integration with traditional healthcare facilities had at least been considered. Again, in most cases above integration was limited to audio or video communication with healthcare providers or some alert mechanisms. 9 of the 30 articles [22], [58], [38], [44], [46], [59], [60], [61] and [62] provided evidence of limited integration into traditional healthcare systems and three in particular [22], [58] and [60] highlighted the significant potential that could be realized if discussed interventions were integrated into existing EMR -/ EHR.
Functional versatility (number and nature of targeted diseases)
Systematic reviews generally discussed multiple diseases targeted with different, but often independent interventions. As demonstrated in figure 7, half of the reviews [2], [4], [10], [12], [24], [29], [31], [32], [35], [47], [49], [63] and [64] discussed multiple diseases but were largely not specific to any one disease. Five [4], [10], [24], [31] and [63] actually identified diseases they targeted by name and the remaining 8 were not specific to any particular disease. Of the above 8, three [4], [12] and [29] involved audio or video conference engagements with patients, allowing them to discuss or target a broad range of unspecified conditions or diseases. However, 8 reviews [8], [9], [11], [30], [33], [34], [50] and [51] out of 26, targeted only one disease such as diabetes, COPD, or asthma, and 3 reviews [8], [30] and [33] discussed interventions related to cardiovascular diseases. The remaining 5 reviews did not provide sufficient details to determine whether they targeted one or more diseases. Most of these simply monitored vital signs but were not clear about the targeted disease(s). Where multiple diseases were manifestly targeted in a particular systematic review, it was often not clear whether a combination of diseases was targeted by the same or different RPMTSs.
The picture was quite different with regards to primary articles (See PA in Figure 7). Of the 30 articles reviewed, 22 articles [58], [20], [36], [37], [39], [40], [41], [42], [46], [52], [53], [54], [55], [56], [57], [59], [61], [62], [65], [66], [67] and [68] discussed interventions which targeted one, single disease. Targeted diseases ranged from PTSD, mental health, Parkinson disease, COPD to IBD and malaria. The remaining 8 articles [22], [15], [3], [38], [43], [44], [45] and [60] tended to cover a combination of diseases but only three [38], [44] and [60] were specific about the combination of diseases or parameters they sought to monitor or measure (HIV/AID and TB, and parameters related to CVD or COPD).
Accessibility to the general public
While a number of systematic reviews did not provide enough details to determine the extent to which, discussed interventions were accessible to the general patient or potential patient population, the majority [4], [9], [10], [11], [12], [14], [18], [23], [24], [29], [31], [32], [34], [48], [49], [50], [51], [63] and [64] provided information which indicated that accessibility was limited as depicted in Figure 8. Among the many mentioned factors which negatively affect accessibility were usability, integration between patients and electronic health records, and giving personalized feedback [4], [63], centralized and decentralized data problem, which is a source of confusion and poses security and privacy challenges [32], performance in clinical settings is still controversial [31] and insufficient healthcare infrastructure and funding [11], [29].
Other reviews, however highlighted more systemic and historical challenges including those related to inequalities and the needs of the target use group which ought to be taken into consideration early in the design and development of mhealth tools; vulnerable, hard-to-reach, or otherwise high-risk patient populations [10], [34], [48], [49] and [64]; varying degrees of literacy, connectivity and accessibility and some patients who were concerned that their care would become dependent on technology, resulting in depersonalised care, reductions in face-to-face interaction, and increased out of pocket costs [14], [18]; characteristics of the care setting and circumstances surrounding individual patients such as rural vs urban, in or out-patient, care delivery & payment models, patient's characteristics and care goals and preferences.
Studies of telehealth should consider combinations of apps of telehealth and outcomes that are important in these new models and that evaluate the specific contribution telehealth can make in these contexts [24] and review [23] pointed out that prior to deploying a newly developed intervention into healthcare settings, its practicality, clinical effectiveness and potential commercial benefits ought to be established and backed up by concrete evidence.
Primary articles also displayed similar results. Of the 30 articles, only three [40], [43] and [45] provided information which indicated that accessibility of the intervention to the general public had been considered or was at least desired. Eight articles did not provide details related to accessibility. As depicted in Figure 8, the majority of articles [22], [58], [3], [36], [38], [39], [41], [42], [44], [46], [55], [56], [57], [60], [61], [62], [66], [67] and [68] gave various reasons why accessibility of discussed interventions was limited including scalability [3], [36] health apps and smart phones' credibility for continuous data flow, feasibility, portability, and power consumption [58], [39], [44] and [66], limited or lack of training [42], limited connectivity and internet requirement of systems [41], and failure to take into account natural variations in patient physiology or behaviour [62]. Other mentioned factors were similar to those covered by systematic reviews.
The main purpose of interventions
The large majority of systematic reviews discussed interventions which included patient monitoring for various purposes, ranging from reporting worsening symptoms of chronic diseases such as Heart failure, COPD, asthma, infectious diseases to patient triage (see Figure 9).
In some cases [4], [29], [30], [32], [33], [35] and [47], monitoring was combined either with prognosis or diagnosis of various diseases. In relatively few cases [2], [11], [12] [18] and [49], reviews discussed interventions which exclusively focused on prognosis, diagnosis or triage of patients without the requirement of continuous patient monitoring as part of the intervention. In the remaining 14 cases, reviews discussed interventions whose purpose was either or a combination of communication, wellness and emergency alerts in addition to patient monitoring.
As far as primary articles were concerned, the vast majority of articles (20) discussed a combination of monitoring, communication, wellness and emergency alerts either for assessing the severity of symptoms of pre-existing health conditions or to manage patient’s adherence to treatment. However, as shown in Figure 10, one article [3] discussed on-demand monitoring for triage purposes and only hinted at prognosis and diagnosis but did not clarify its level of integration with traditional healthcare systems. In the remaining cases [36], [37], [40], [41], [43], [45], [46], [65] and [68], prognosis and or diagnosis were mentioned along with continuous patient monitoring for vital signs. Overall, most articles were clear about their main purpose.
Design and Implementation approach
Of the 26 systematic reviews considered, only one review [9] discussed an implementation which, placed patients at its centre, providing training, educational materials and daily phone calls to support patients. In seven reviews [8], [2], [23], [32], [33], [63] and [64], the design was considered to be more technically centred, with patients and potential patients simply being expected to adopt the designed solution.
Nine reviews [5], [14], [24], [30], [31], [34], [48], [49] and [51] gave evidence of wishing to pursue a user-centred design but there was an indication that such design was either not achieved or was limited due to factors such as unavailability of mHealth apps on some operating systems [49], limited mobility and flexibility, in addition to the trustworthiness and quality of the content, and personalization possibilities through customization and adaptability [5], usability drawbacks, as well as reports of the need for more comprehensive solutions, including the provision of real-time feedback and the integration of the EHRs systems being used by the care providers [51]. The layout of design approach for systematic reviews is depicted in Figure 11.
With regards to primary articles, five articles [15], [20], [46], [58] and [62] of all articles set out to pursue a user-centred design from the outset of intervention design by broadly consulting clinicians and patients. As can be seen in Figure 12, seven considered articles [37], [39], [43], [57], [61], [66] and [68] were deemed to have pursued a technical rather than a user-centred design and several of the considered articles discussed off-the-shelf solutions which required customization. However, even though most of the articles were silent about the design approach overall, the importance of a user-centred or patient-centric approach was broadly acknowledged to facilitate adoption by end-users, in the vast majority of reviewed primary articles.