Noncommunicable diseases (NCDs) are the leading cause of death and disability worldwide [1,2]. Although complex [3-5], approximately 80% of NCDs can be accounted for by modifiable risk behaviors, such as physical inactivity, unhealthy diets, smoking, harmful alcohol consumption, and stress management [3,6,7]. The comprehensive management of risk behaviors [8,9] through the implementation of self-management strategies [10-12] is critical to fostering a life-long approach to secondary prevention [13]. According to Bandura [14], to achieve successful and sustainable self-management, people have to learn to monitor their health behavior and the circumstances under which it occurs.

Secondary prevention in an NCD context usually involves referral to a structured hospital or community-based programs, such as cardiac rehabilitation [13,15,16], pulmonary rehabilitation [17], or diabetes education [18-21]. Participation in center-based programs improves health-related quality of life [12,22-24] and clinical outcomes [23] and lowers the risk of hospitalization [12,22,24], recurring adverse health-related events, and all-cause mortality [9], compared with usual care.

Despite the proven effectiveness of secondary prevention [12,22-24], uptake and adherence to structured face-to-face programs is suboptimal [25-29]. Personal factors associated with low attendance include cultural, financial, and psychological barriers (eg, readiness or willingness to attend) [17,27,30]. Key operational barriers to active participation relate to availability of program resources, including limited program enrollment capacity [29,31], restrictive hours of operation [27], and limitations in program suitability (eg, for people self-managing multiple NCDs). Accessibility barriers include inadequate transportation and a lack of services within remote rural areas [17,27,29,30,32-36]. In response, there has been a necessary shift toward flexible and convenient home-based services [32,37-41], offering an evidence-based alternative to nonattendance and reducing the distance of care [42].

Digital health is a contemporary advancement in home-based self-management of NCDs. Widespread internet connectivity to at least 55% of the global population [43], over 5 billion mobile phone users [44] and the availability of thousands of mobile health applications [45] has provided unprecedented access to the digital delivery of home-based support. Content, structures, and modes of delivery for these digital interventions are wide ranging, and their potential effects are well documented.

A systematic review and meta-analysis of internet-based interventions targeting health behavior change in NCD groups (n=43,236) showed significant improvements in risk-related behavior, such as physical activity (effect size 0.24; 95% CI 0.09-0.38), dietary behavior (effect size 0.20; 95% CI 0.02-0.37), and alcohol consumption (effect size 0.14; 95% CI 0.00-0.27) [46]. The findings showed that offering multiple modes of delivery within an intervention (eg, internet-based plus text messaging) had a greater effect on behavior (effect size 0.81; 95% CI 0.14-1.49) [46].

Reviews of communication technologies used to deliver health services and facilitate patient and health care professional interaction [47] have reported that improvements in self-management of NCDs are not inferior to the positive changes produced by structured center-based programs or usual care [42,48]. Similarly, reviews of mobile-based text messaging interventions have shown sustained health behavior change [49-53] and the potential to overcome barriers associated with traditional center-based models via simple to use, flexible, cost-effective digital health [52,53].

Collectively, the literature indicates that evidence-based digital health interventions have the potential to discretely impact the self-management of NCDs while also complementing one another. Despite the advantages of these discrete interventions, no single program meets the needs of all users, as self-management of NCDs and user preferences are complex and multifaceted in nature. Discrete digital health solutions could be complimentary when combined for use. The REMOTE-CR program [54] and CAP-CR program [55] draw on combinations of web-based and mobile-based features through smartphone-enabled software to facilitate digitally assisted self-management. Outcomes from a randomized control trial of the CAP-CR intervention (n=120) indicated that uptake, adherence, and completion rates at 6 months were significantly higher than those of center-based cardiac rehabilitation programs [55]. REMOTE-CR integrates the use of smartphones, wearable sensors, and web apps to provide real-time remote exercise monitoring. A randomized control trial (n=162) found REMOTE-CR to be an effective and cost-efficient alternative to center-based programs [54].

The constraints of disparate digital health interventions mean that an undue amount of time is required to seek out interventions offering the content, features, or delivery mode best suited to the current preferences of the individual user. Moreover, when a person’s health status or general lifestyle habits change, seeking practices would have to be repeated, which may reduce the sustainability of a discrete digital program intervention and long-term user adherence [56].

Therefore, we need flexible and versatile solutions in the digital health space. This requires a shift away from the creation of restrictive digital interventions toward a paradigm that facilitates optimal engagement through centralized choice.

We propose a digital platform that would capitalize on existing digital self-management interventions that have been evaluated for effectiveness in a specific context (eg, NCDs) through a comprehensive experimental design [57]. The platform would host a digitally supported palette of discrete, evidence-based digital health interventions and incorporate a digital guidance tool to direct users to the intervention best suited to their current individual needs and preferences, optimizing personal relevance and user experience [47].

Existing disparate digital health interventions may offer modest positive effects, but engagement can be varied. Thus, a digital platform could potentially optimize engagement while also lessening the burden of care associated with irrelevant content, user ambivalence, and time-consuming seeking processes. Such an approach not only facilitates personalization but also encourages user autonomy through the self-selection of a combination of program components, which is associated with long-lasting positive effects on disease management and patient empowerment [37,58,59]. There is significant potential for this body of work, as a plethora of underutilized evidence-based digital health interventions exists, involving rigorous development processes and gold standard evaluations.

A preliminary pilot of this review was conducted in August 2018, but no digital platforms that matched our criteria were identified. Therefore, the focus of this review was extended to include digital platform-like interventions. Details of the protocol for this review were registered on PROSPERO (International Prospective Register of Systematic Reviews) in 2018 (PROSPERO 2018, Registration Number: CRD420 18102095). A follow-up literature search was conducted in January 2020.

For this review, a digital platform-like intervention was defined as a digital solution that allows users to choose from a suite of discrete, evidence-based health behavior change features to support NCD self-management. Offers a digital tool for guidance toward intervention features that are most suited to the user’s needs and preferences. This broadened definition allowed us to best provide an assessment of the potential of digital platforms in self-management of NCDs, despite the absence of existing literature.

The aims of this study were to identify digital platform-like interventions for the self-management of NCDs and to examine the potential for digital platform-like interventions to support self-management of NCDs through effectiveness and process outcome measures.

The eligibility criteria are outlined in Table 1 according to an adapted version of the Population, Intervention, Context, Outcome approach [60,61]. Studies eligible for inclusion were published in English between January 1990 and January 2020; this period coincides with the activation of commercial internet service providers. All study designs were eligible for inclusion because of the emergent nature of the research topic.

A systematic literature search of 9 electronic databases was conducted in January 2020: EBSCOhost (Academic Search Complete, Applied Science & Technology Source, Cumulative Index to Nursing and Allied Health Literature Complete, Global Health, Health Business Elite, and PsycINFO), PubMed (MEDLINE), Scopus, and EMBASE. Reference lists of included study publications and related conference proceedings were hand searched to identify additional publications that may not have appeared in database searches. Peer-reviewed systematic reviews and meta-analyses, in which the included studies were cited, were explored to identify any similar studies that did not appear in database searches.

Individual search strategies were developed for each database (example shown in Multimedia Appendix 1) and included search terms derived from 3 main categories of interest: chronic disease, digital technology, and self-management of health behavior.

All results were exported to a reference manager (EndNote X9, Clarivate Analytics), where duplicates were removed before screening for eligibility. Search results have been reported using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement [62]. Each stage of the screening process is presented in Figure 1. Titles, abstracts, and full texts were independently assessed by 2 reviewers (ST and JR), whereas the third and fourth reviewers (KB and RM) were consulted to collectively reach a consensus on studies with questionable eligibility.

The combined search strategy identified 1582 records, of which 1568 were identified through electronic database searching and 14 through other sources. Duplicates were removed, and the remaining 507 records were screened for eligibility through titles and abstracts. Full-text was obtained for the 85 remaining records, of which 29 publications outlining 7 digital platform-like interventions met the inclusion criteria (Table 2) [38,69-95]. The primary reasons for the exclusion of full-text records are shown in Figure 1. Table 2 shows how the 29 publications were related to each of the 7 studies, sorted according to the evaluation stage [57]. Owing to the heterogeneous nature of the included studies, a meta-analysis was not possible.

This review identified 7 diverse digital platform-like interventions [38,69-95]. Of the 7 interventions, 3 targeted diabetes [85,88,93], 3 were intended for cardiovascular diseases (CVDs) [71,82,94], and only one intervention supported people with chronic obstructive pulmonary disease (COPD) [77].

Of the 7 studies, 4 evaluated the effectiveness of their interventions [71,77,85,88], including 3 randomized control trials and 1 single-arm pre-post cohort design [85]. Sample sizes ranged from 69 to 1325, and mean participant ages ranged from 57 to 65 years. Of the 4 studies, 2 reported gender imbalance within participant groups (23% [71] and 31% [88] female participants). Trial periods ranged from 3 to 12 months, and the control groups in the 3 randomized control designs received the usual care [77], accessed simple web-based information [88], or a variation of the described intervention [71].

Of the 7 studies, 3 had not evaluated effectiveness at the time of this review [82,93,94], but all 3 had completed a pilot evaluation of their intervention. Pilot sample sizes ranged from 35 to 120 and had relatively shorter trial periods of 5 weeks [93], 10 weeks [94], and 6 months [82]. One of the studies subsequently published a protocol for a planned randomized control trial [95] to be completed by December 2020 (clinical trial registration: NCT03159325; international registered report identifier: DERR1-10.2196/12322).

Parent publications for each study [71,77,82,85,88,93,94] were quality assessed; thus, the MMAT was completed once per study and not for each publication. Companion publications were assessed where a Can’t Tell response was coded [38,69-96]. Overall, 6 of the digital platform-like interventions were intended to be used ad libitum. Therefore, the criterion “2.5. Did the participants adhere to the assigned intervention?” was not appropriate. This section was alternatively appraised based on the participants’ completion of follow-up data collection.

The details of the assessment criteria can be found in Multimedia Appendix 2 [71,77,82,85,88,93,94]. Overall, the scores for the included studies ranged from 3 to 7 out of a possible maximum of 7. This translates to a varied methodological quality range of 2 to 4 stars. The mixed-method studies were of the highest quality, achieving 3 [94] and 4 stars [85]. The most frequent limitations were incomplete outcome data sets [71,77,82,88,93], poor follow-up response rates, [71,77,82,88,93], and insufficient integration of quantitative and qualitative components in mixed-method studies [94].

The key characteristics of the included interventions are outlined in Table 3. All interventions included an intermediary user interface, which allowed participants to explore and self-select optional components based on their individual needs and preferences. All 7 delivered a choice of at least two evidence-based behavior change techniques or strategies (eg, self-monitoring and goal setting) [97] targeting a lifestyle-related behavior (eg, physical inactivity and smoking). Physical activity was the most frequently targeted health behavior, addressed in 6 of the 7 digital health interventions [71,77,82,85,88,93]. Of the 7 interventions, 2 focused on disease-specific self-care as their key targeted behavior, which incorporates a wide range of health behaviors [85,94].

All 7 interventions presented algorithm-supported interventions that facilitated user guidance [71,77,82,85,88,93,94]. Specific features were included to optimize the sustainability of this automated approach, such as providing recommendations to users through computer-generated responses to questionnaires or self-reported data (ie, eDecision support tools) [73,77,82,88,93,94], smart recommendation widgets that offered advice based on individual patterns of use [85], and application of the transtheoretical model (TTM) of behavior change to select stage-matched intervention features [71].

Support from a health care professional was described in all 7 interventions: typically, cardiac rehabilitation nurse practitioners, exercise specialists, or a member of the research team. The primary role of health care professionals was to facilitate trial and data collection [71,77,82,85,88,93,94] or conduct initial familiarization sessions [71,82,88]. Overall, health care professionals did not provide digital support or enhance existing intervention features, as this approach was not considered sustainable [71,83].

All 7 digital platform-like interventions were evaluated (Multimedia Appendix 3; [71,77,82,85,88,93,94]) through either preliminary pilot investigations [82,93,94] or effectiveness trials [71,77,85,88]. Of the 7 interventions, 2 used evidence-based self-report measures for data collection [71,77]; 2 other studies collected data during researcher visits to participants [82,93] and from hip-worn ActiGraph accelerometers; and 3 collected data through a combination of existing medical records [85,94], self-report measures [85,88,94], and researcher visits to participants [85,88].

This review is the first of its kind to systematically examine the literature on digital platform-like interventions for the self-management of NCDs and health behavior change. Although no digital platforms were identified, an examination of digital platform-like interventions has contributed knowledge to the conceptualization and potential value of using digital platform architectures to support self-management of NCDs.

A total of 7 digital platform-like interventions were included in the review. Evaluations for effectiveness were disparate, and so a consensus on the overall effect could not be reached. Nevertheless, positive effects were reported for physical activity, disease-specific self-care, and self-monitoring behaviors, which is a promising finding in support of digital platform use for the self-management of NCDs. As a result of the findings mentioned earlier, 3 dominant themes emerged: development, optimizing change, and support and guidance for users.

This review found that comprehensive and systematic development processes were implemented for most platform-like interventions. Iterative, user-centered approaches are highly regarded in the field of digital health research because the proactive engagement of patients can be beneficial in the development of digital interventions [103,104]. It allows researchers to gain an in-depth understanding of the psychosocial context of the potential end users of the intervention [105] and shortens the communicative distance between the researcher and the user [106]. In this review, 4 of the 7 interventions [75,81,84,91] adopted an iterative, user-centered approach. Reporting methods across the included interventions were varied, so it was difficult to determine whether incorporating user-centered development was associated with greater effectiveness. However, research suggests that such development methodologies have the most potential for developing a sustainable solution [107]. An average of 80% adherence at follow-up stages was found in the same 4 studies [77,82,85,88], which could be attributed to the comprehensive iterative development approaches used in all 4 of the interventions. The level of user engagement with digital health technology tends to wane over time [56,108,109], which has been associated with a lack of perceived value by the user and increased burden because of irrelevant material [110,111]. Placing the user at the center of development considerations and including them in decision-making processes may have affected the creation of fitting platform-like structures, appropriate for those living with NCDs and delivering relevant, usable content or features.

All 4 of the aforementioned interventions [75,81,84,91] also specified a clear theoretical underpinning to their development (eg, TTM), which is consistent with research recommendations that intervention content and features associated with a solid theory base are more likely to be effective in changing behavior [46]. A strong theoretical framework may also have affected the generalizability and adaptability of the platform-like interventions, further adding to the potential for efficacy and longevity [66,112,113] as seen through high adherence rates. Only 3 of the studies outlined platform-like intervention components according to a taxonomy [82,88,93], which made identifying behavior change features difficult because of vague reporting. We anticipate that digital platform content would have superior clarity for users and health care professionals alike because of the stand-alone, discrete digital interventions offered.

The results presented in this review indicate the potential for digital platforms to affect behavior change, such as disease-specific self-care behavior. Significant differences were reported for sustained diabetes self-care behavior between users and nonusers of the Diabetes Online Companion intervention [85].

The self-care outcome measure covers a range of health behaviors, which address several key lifestyle risk factors (eg, dietary behavior, physical activity, glucose monitoring, and smoking) [98]. Thus, our findings suggest that modest changes in multiple health behaviors (ie, changes in self-care) may have a better overall and sustained effect on the self-management of NCDs than larger effects in one single health behavior. This suggestion supports the opinion that affecting comprehensive lifestyle change may be a better approach to the self-management of NCDs, as NCDs are complex conditions influenced by several interconnected lifestyle risk factors [3,114]. Similar to self-care, improvements in self-monitoring behavior were identified through the HAPA, which incorporates six personal determinants of behavior change.

Lifestyle-related risk factors and their corresponding health behaviors are heavily influenced by one another, and thus modifying one health behavior using a digital platform may not necessarily generate sustainable improvements in the self-management of NCDs. For example, smoking behavior has been inversely linked to physical activity behavior, meaning that a lack of engagement with smoking cessation may inhibit physical activity progress [115-117]. Similarly, poor stress management may have a negative effect on engagement with healthy behaviors, such as smoking cessation [118], physical activity [119,120], and maintaining a healthy diet [121]. Single interventions that independently target diet and physical activity behavior may improve those isolated behaviors, but sustained changes in associated risk factors (eg, weight loss and maintenance) are more probable when interventions simultaneously target both health behaviors [114]. The results of our review showed limited to no changes in isolated behaviors (eg, overall physical activity and smoking) [77,82], but modest improvements were apparent in overall disease-specific self-care and multiple self-management domains [85,93,94].

Focusing on modest overall improvements in health behavior change may produce further consequential effects on sustained NCD self-management. Self-regulatory and self-efficacy theories for promoting self-management suggest that the greatest improvements in self-care of chronic conditions typically occur following some initial success in changing behavior (ie, mastery experience) [14]. Thus, seeking a modest change in a more comprehensive and generalizable outcome measure such as disease-specific self-care may generate a cascade of behavior change improvements moving forward.

A comprehensive approach could also provide a greater scope for successful self-management throughout the unpredictable health trajectories of NCDs, which are relatively unique to each individual [122]. This hypothesis is in line with primary care digital health frameworks that dynamically adapt services to the clinical care pathway of the individual, which cannot be predetermined and changes regularly [123]. This approach supports our hypothesis that overly specific digital health interventions may not be the most appropriate solutions to comprehensive self-management of NCDs. A digital platform could accommodate not only the stage of NCD but also the current health state of the person living with that NCD, by providing the self-management tools required to suit individual circumstances. It would create an opportunity for users to experiment and tinker with evidence-based interventions [124], and experience successes and failures on their path to successful self-management, leading to new insights and skills in self-care. The evaluation of a digital platform should extend further than simply adhering to the discrete digital interventions hosted within the platform and should explore how they are used and actively incorporated into everyday life [104]. Longer evaluation trial periods may better facilitate these dynamic user health trajectories to accommodate periods of exploration and changing mindsets.

The digital platform-like interventions included in this review had good adherence for up to 9 months, which could be attributed to the freedom for users to explore an array of digital features and identify the most relevant components for them as an individual. However, given that a digital platform would provide access to an extensive choice of discrete digital health interventions, it is important to note that the intention is not to overburden users [125]. It is possible that free navigation through a platform with various features to negotiate could be a cumbersome process [110].

Research indicates that using all available components is not necessarily more effective [56], and presenting a platform not matched to user needs could be detrimental to the overall success of the platform [126]. Thus, evidence suggests that the addition of human support in digital health can enhance user engagement, as users value reassurance and expert knowledge to guide their decision making [66,127,128]. Human support is resource intensive and can increase the overall cost and burden on health care systems, which may not be a feasible solution. Therefore, it is important to establish the extent to which further value can be added to a multicomponent digital health intervention by supplementing it with human support [127].

In view of this, our review has recognized the potential of applying existing theories and knowledge to create efficient, cost-effective automated guidance for users to make informed choices about their engagement. The digitally supported guidance offered to participants within the 7 interventions provided them with a supported pathway to choose behavior change components. This is in line with behavioral research that indicates that actively engaging participants in decision-making pathways for care can improve health outcomes [129] through factors such as improved autonomy [37], empowerment, and mastery [130]. These factors are critical in supporting people to self-manage their disease and in promoting a more digitally engaged patient [104].

One reason that people value human support is a sense of accountability [127]. However, this review showed that a sense of accountability was successfully acquired by several participants using automated platform-like interventions. Providing clear associations between a digital platform and expert health care professional advice may be enough to satisfy the desire for accountability. This is reinforced by research that suggests a key factor in promoting engagement with digital health interventions is clear endorsement by respected clinicians or expert organizations [110,127]. A digital platform would seek to use existing, evidence-based digital health interventions that could be put through a rigorous selection process before inclusion, driven by an extensive list of key stakeholders (eg, health care professionals, potential end users, and family members) to further endorse the trustworthiness of the platform [110].

The comprehensive development of these evidence-based, automated guiding decision pathways is another way to incorporate expert advice and guidance to support the trustworthiness of a digital platform. Automated eDecision support tools are currently being developed and validated to assist with meaningful adherence to interventions and health behavior by providing individualized, real-time assistance [81,131]; however, no validated tools have been described in the included literature to support navigation through complex digital platforms. This may have an impact on the usability or trustworthiness of the included digital platform-like interventions. The variation in guidance systems presented in this review points to a requirement for more research on an operative level of support to optimize engagement and create a user-centered experience for the individual.

In summary, the adaptive nature of a digital platform accommodates the requirement for flexibility in self-management support, which could facilitate a diverse range of users and life circumstances [66,132]. Longitudinal and detailed evaluations of digital platforms must be carried out to influence the lifetime of positive health-related choices and behaviors. It is important not only to evaluate the longevity of participating in such an intervention but also to explore the diverse patterns of engagement by a wide range of users.

This review was prospectively registered, used comprehensive search strategies across multiple databases, and reported according to PRISMA. Searches and publication of results have been conducted in an up-to-date and timely fashion. The inclusive nature of this review accommodated a broad range of NCDs, which enhances the generalizability of the findings among a wider population.

This review was not without limitations. The novel concept of a digital platform was introduced for the first time, which made it difficult to identify relevant literature. Having broadened the inclusion criterion, discrepancies in terminologies could still have led to key studies being omitted from this literature search or the misinterpretation of intervention content. Current research is focused on the development of a comprehensive eHealth taxonomy, but this is not yet wide ranging enough [133]. In response to this, a systematic approach and extensive use of terminology in search strategies were implemented to ensure that pertinent literature was included for review.

Another limitation was that none of the interventions included in this review were readily available on the web to the reviewer, none provided a digital preservation URL, and intervention descriptions were not always sufficient to identify key features. There has been a call for improved reporting of digital health interventions to improve examination and evaluation of intervention characteristics. This limitation is an important reminder that future developments should use reputable reporting frameworks and guidelines to outline their work [64,134].

We have identified a gap in the research on comprehensive and flexible digital health for the self-management of NCDs. Thus, we proposed the contemporary concept of a digital platform, which supports the innovative use of already existing digital interventions for health behavior change. Initial searches indicated that no such digital platforms currently exist, which may indicate a missed opportunity to optimize user engagement with already developed, evidence-based digital interventions. In response, this review focused on digital platform-like interventions to provide an understanding of the development and contextual considerations required to optimally construct a digital platform.

Iterative, user-centered development may be associated with improved adherence and sustained use. Offering a palette of evidence-based interventions with an appropriate degree of guidance has the potential to facilitate disease-specific health behavior change and effective self-management among a myriad of users, conditions, or stages of care. This review and the novel concept of digital platform-like interventions contribute new knowledge to the rhetoric of digital health for the self-management of NCDs.