Chronic kidney disease (CKD) is a major public health concern [1-3]. Globally, more than 70 million individuals are affected by CKD [4]. CKD is defined as kidney damage or a measured glomerular filtration rate (GFR) of ≤60 mL/min/1.73m² for more than 3 months. CKD is classified into 5 stages based on GFR decline [5]. The level of kidney function deterioration has a direct relationship with an increase in morbidity and mortality [6], poorer patient outcomes [3], higher hospitalization rates [7], and substantial increase in health care expenditures [8]. Patients with CKD report a lower quality of life (QoL) [9] and may experience severe medical complications and cognitive dysfunction [10].

Disease self-management (hereafter referred to as self-management) is defined as “an individual’s ability to manage the symptoms, treatment, physical and psychosocial consequences, and lifestyle changes inherent to the life with a chronic condition” [11]. Adequate self-management is reported to improve patients’ health behaviors targeted by the intervention (ie, proximal outcomes) and also indirect outcomes, such as disease characteristics and progress (ie, distal outcomes) [12-14]. Although the potential benefits of self-management interventions are widely reported in the literature, extrapolating these results in day-to-day practice is difficult. Lack of efficacy in practice might be related to a suboptimal implementation of the self-management interventions [15,16]. Reported barriers were often related to intervention characteristics, such as lack of tailoring to the individual patient. Moreover, a lack of patient involvement in intervention design and insufficient care continuity and accessibility were reported to hamper implementations [17,18].

Electronic health (eHealth) technologies can help address implementation barriers by making interventions more accessible, acceptable, tailored, and interactive [19-21]. The most cited definition of eHealth is that by Eysenbach [22]:

eHealth can help patients achieve personal health goals, and it allows them to feel more responsible for their health status [23]. Moreover, eHealth facilitates remote patient communication and exchange of health data, helping to increase health care efficiency while maintaining a wide-scale, cost-effective health care approach [24]. eHealth interventions have been successfully implemented to support weight loss [25,26], promote smoking cessation [27], reduce depressive symptoms [28], and decrease mortality rates and acute admissions [29]. In addition, eHealth-based interventions have been successfully applied to manage chronic disease [30-32].

Several studies have reported the use of eHealth-based self-management interventions in CKD [33-36]. Moreover, 3 systematic reviews were published on this topic [37-39]. However, these reviews only concentrated on 1 particular eHealth application, such as telemedicine; dietary mobile apps; and automated information technology tools. Moreover, these reviews focused on a limited number of study designs and outcomes. For example, 2 reviews only included randomized controlled trials (RCTs) [38,39], and 1 review excluded studies focusing on implementation outcomes such as feasibility, validity, and acceptability [39]. Moreover, none of these reviews [37-39] reviewed the contribution of individual intervention components (eg, self-monitoring) to the effects found. These limitations of previous reviews make it difficult for researchers and intervention developers to determine which components should be employed to maximize the effectivity of eHealth self-management interventions for CKD patients.

This study, therefore, aimed to systematically review the available evidence on eHealth-based self-management interventions for CKD. In specific, we aimed to review the following: (1) study characteristics and type of eHealth applications used; (2) intervention components implemented and, if possible, their relative contribution to the effect found; (3) both process and effect outcomes; and (4) determinants of implementation.

This review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [40]. The protocol was registered in the international Prospective Register of Systematic Reviews database (Centre for Reviews and Dissemination [CRD] number: CRD 420 180 81681).

A systematic search was conducted to identify relevant articles; the search strategy was developed in collaboration with a certified librarian. In total, 8 electronic databases (PubMed, EMBASE, Web of Science, Cochrane Library, EmCare, PsycINFO, Academic Search Premier, and Science Direct) were searched in November 2017. Search terms covered 3 areas: (1) CKD, (2) eHealth, and (3) self-management (see Multimedia Appendix 1). Reference lists of the included studies were searched to identify other relevant articles. EndNote X9 (Clarivate Analytics) was used to support the review process.

Inclusion and exclusion criteria (Textbox 1) were determined using the Patients, Interventions, Comparison, Outcomes, Study design methodology [41].

After removal of duplications, titles and abstracts of the retrieved articles were screened independently by 2 reviewers (HS and XC). Articles that did not meet inclusion criteria were removed. Potentially relevant articles were obtained in full text and reviewed independently by 2 authors (HS and XC). Any disagreements between the 2 authors were resolved by consensus or consultation with a third author (RK).

Data collection was performed independently by 2 reviewers (HS and XC) using a standardized data extraction form. Study characteristics, descriptions of eHealth self-management interventions (eg, intervention components), process and effect outcome indicators, and determinants of implementation were extracted. Discrepancies in extraction were discussed until consensus was reached.

Article quality was appraised independently by HS and XC using the Crowe Critical Appraisal Tool (CCAT) [42]—a reliable, widely used quality appraisal tool [43,44]. Use of the CCAT user guide promoted validity and inter-rater reliability [43-46]. The CCAT form is divided into 8 categories and 22 items, with a total of 99 subitems. Subitems are rated on a scale of present, absent, or not applicable. A 6-point scale ranging from 0 (the lowest) to 5 (the highest) is used to assign score per category, with 40 being the maximum achievable total score.

The CCAT does not allow for a qualitative comparison of appraisal scores. Hence, we used the star score system developed by our research group to compare study quality [47]. First, we calculated a quality score based on the CCAT. Then, a mean score and standard deviation of the quality scores were calculated. Star scores were then assigned to each article: 1 star if a quality score was more than 1 SD below mean; 2 stars if a quality score ranged from 1 SD below mean to mean score. The kappa between the 2 reviewers’ scores of quality assessment was 0.63, reflecting substantial agreement [48].

Data were reviewed using narrative synthesis [49]. Study characteristics were reviewed, summarized, and analyzed in a spreadsheet. In accordance with previous categorizations of eHealth [32,39,50], eHealth self-management interventions were split into 5 major types (see Multimedia Appendix 2). eHealth functionalities used were described based on the technology functionality framework [51,52]. In addition, based on the operationalization by Mohr et al [53], eHealth-based self-management interventions included were further detailed: (1) intervention components (based on Morrison et al [54]; see Multimedia Appendix 3)—active intervention parts that support self-management behavior, including elements defined as what is provided to the user (eg, education materials, integrated alerts, and video conferencing options), how these elements are delivered (eg, plans and quizzes), and the subsequent intervention workflow defined as when they are delivered (eg, daily use)—and (2) intervention strategies—behavior change techniques [55] that underlie the intervention components (eg, role modeling if the Web-based education materials used include a video of patient who successfully manages his/her disease).

Outcome indicators were classified into 2 categories: effect outcome indicators and process outcome indicators [56]. Effect outcome indicators were outcomes related to self-management, health status, or cost-effectiveness, whereas process outcome indicators were outcomes on care process, health care delivery, or patient-health care interactions (eg, adherence and usability).

To allow for comparability, we classified the results reported as positive effect, no statistically significant effect, or mixed effect (see Textbox 2). No negative outcomes were reported in the studies included in this review. Only quantitative methods were used to measure effect outcome indicators, whereas mixed methods were used to measure some process outcome indicators. Hence, the classification of the results of the process outcome indicators slightly differs from that of the effect outcome indicators. Outcomes related to patients and care providers are reported separately.

The determinants of implementation of eHealth self-management interventions extracted were categorized following the widely cited framework by Fleuren et al [57]. This framework identifies 50 determinants of program implementation in 5 subgroups: (1) characteristics of the sociopolitical context, such as legislation; (2) characteristics of the organization, such as staff turnover; (3) characteristics of the person adopting the innovations (user of the innovation), such as knowledge; (4) characteristics of the innovation, such as complexity; and (5) innovation strategies, such as a training. For example, the study by McGillicuddy et al [36] included in our review mentioned that “six subjects did not complete the lead-in phase, 5 for technical reasons relating to poor internet at their home.” This barrier was then mirrored to the 50 determinants in Fleuren framework and classified as a determinant related to the innovation and, more specifically, added to the determinant category perceived quality of eHealth intervention is excellent. In addition, in each subgroup, we identified the influence of the patients or care providers.

Our search retrieved 3307 articles in total. After removing 1497 duplicates, 1810 relevant articles were screened based on title and abstract. A total of 123 potentially relevant articles were screened full text. Of these papers, 2 described results of the same RCT [58,59] and were assessed jointly. Finally, 24 articles (comprising 23 studies) [33-36,58-77] were found eligible for inclusion in this review (Figure 1).

All 23 studies were published between 2005 and 2017, with 19 of them being conducted between 2012 and 2017 [33-36,58,64-77]. A total of 13 studies were conducted in the United States [33-36,58,60,62-65,69,71,72], followed by 2 studies in the United Kingdom [70,74]. The research designs used varied; the majority used an RCT design [33-36,58,63,64,66,70]. Most studies focused on the usability, acceptability, and feasibility of eHealth self-management interventions [36,58,61,64,65,67,69,71,72,74-77]. Most participants are patients receiving hemodialysis [58,60,62-64,66,68,69,76,77]. Sample size at baseline ranged from 5 [67] to 601 [34]. Target population age ranged from 21 to 93 years. Recruitment mostly occurred via medical centers/hospitals [35,58,60-62,68,73]. Intervention duration ranged from 2 weeks [76] to 24 months [61]; 2 studies did not specify intervention duration [58,67]. A total of 10 studies performed a follow-up measurement [33,34,63,65,66,69-71,73,76]. Moreover, 12 studies included a control group, and 9 of those studies [33,34,36,58,61,66,68,70,76] reported usual care or no internet-delivered intervention as control condition. The study characteristics have been presented in Multimedia Appendix 4.

Quality of the included articles varied (Table 1). A total of 3 articles [70,73,75] were awarded a 4-star rating, 11 [33-36,59,63,65,66,69,74,76] a 3-star rating, and 10 [58,60-62,64,67,68,71,72,77] a 2-star rating or lower. Articles with a 4-star rating scored higher on design, sampling, data collection, and ethics compared with those with a 3-star rating or lower. Moreover, 20 articles [34-36,59-70,72,73,75-77] provided insufficient details on their study design or rationale. Sampling method used (eg, randomly and purposively) was not reported in 10 articles [35,60,62,65,67,71,73,74,76,77]. Although both the number and characteristics of participants were described in most articles, 15 articles [58-62,64-69,71,74,76,77] did not specify the method of sample size calculation. A total of 10 articles [58-62,64,65,67,71,72] did not detail methods used to ensure the quality of the data collected or to reduce bias. On average, the lowest score was obtained on the ethics section.

Major types of eHealth, functionalities, and key intervention components used are summarized in Tables 2 and 3. Most eHealth interventions evaluated included multiple components (multiple eHealth types) to improve patients’ self-management (8/23 articles). Studies included did not provide detail on the specific intervention strategies underpinning these components, such as behavior change techniques. The most frequently used intervention component was self-monitoring (17/23 articles), followed by educational material or training (15/23 articles) and counseling (14/23 articles). Less frequently used intervention components were quizzes (3/23 articles) and interactive feedback from a device (4/23 articles). In addition, 5 studies reported that intervention development was guided by a specific theory.

Tables 4 and 5 present the outcome indicators and the data collection tools used. Moreover, full details on the efficacy data reported in the included studies are included in Multimedia Appendix 5. Table 6 displays the determinants of implementation extracted. No articles reported any adverse outcomes of eHealth self-management interventions.

The main findings and implications have been presented in Textbox 3.

The evidence regarding the implementation and effectiveness of eHealth self-management interventions for CKD patients was reviewed. The 23 studies included were appraised on methodological quality, and all relevant data were extracted. Although the evidence base is still inconclusive, our review provides an indication that eHealth self-management interventions have the potential to improve CKD patients’ management and health outcomes. Furthermore, high acceptability of and satisfaction with the eHealth interventions used were reported. Owing to the heterogeneity of the intervention components and outcomes measures used, we could not determine which intervention components contributed most to the effects found. The determinant ability of health care professionals to monitor and, if necessary, anticipate on patient measurements online was most frequently reported to influence implementation. The determinants reported were not quantified, and the relative importance of each determinant could not be determined.

Most studies reported the evaluation of effect outcome indicators. The positive effects on patients’ BP controlling [36,68,70,75] and medication adherence [35,36,66] were consistently reported; no adverse outcomes were reported. These findings correspond with another review on eHealth interventions in CKD [39]. Compared with standard outpatient-based management, eHealth self-management interventions have the potential to reduce health care delivery costs [78]. Although this potential reduction in costs is essential for policy makers and clinicians to adopt eHealth self-management interventions, health care expenditures were only assessed in 3 of the studies included, with only 1 performing a cost-effectiveness analysis [70]. Hence, we cannot yet determine if and how these interventions might reduce medical costs. This finding is consistent with similar reviews, which conclude that studies on the cost-effectiveness of eHealth self-management interventions are either conflicting or lacking [32,54]. As evidence on cost-effectiveness is important to support the potential scale-up of eHealth technology, further research is needed to broaden this evidence base. Regarding QoL, only 1 out of 4 studies reported a significant improvement. A possible explanation for this finding was the short follow-up period instated to capture changes in a distal outcome such as QoL [59]. As QoL in CKD is an independent predictor of mortality and hospitalization [79,80], and thus important to evaluate, we advise further research to assess QoL with a longer follow-up period.

In general, we found that eHealth self-management interventions were reported to be highly feasible, usable, and acceptable. However, we found great diversity in the use and operationalization of outcome indicators and how they were measured. For instance, a study reported acceptability by measuring adoption, adherence to the recommended intervention use, user satisfaction, and feature usage [75]. In contrast, other studies [33,36] measured acceptability by asking patients “how acceptable they found the intervention” using a self-report scale. It is also notable that only 4 studies assessed implementation adherence, although finding no or limited intervention effects can be strongly related to patients’ nonadherence to eHealth interventions as prescribed [81,82]. Examining implementation adherence can help resolve the black box of patients’ adoption and continued use of the intervention, thereby preventing a type 3 error [83]. To tackle these issues, we advise researchers to use a standardized operationalization of process outcome indicators and measure implementation adherence to enable reliable interpretation of the intervention effect found.

Considering which outcomes are most sensitive to change is important. As eHealth interventions studies are mostly of short duration, they may not detect changes in distal outcomes (eg, QoL). Hence, effectivity might be easier to detect when proximal outcomes, close to the intervention strategies, are measured. For example, BP controlling can be an outcome sensitive to change if self-monitoring is the main intervention component. Functional outcomes (such as days needed to return to work), which can quantify patients’ subjective perceptions of the effect of treatment on their daily life, might also be very sensitive to change by eHealth interventions [84,85]. Moreover, researchers should consider if their outcomes reflect meaningful change and provide a clear rationale for their choice of laboratory parameters. For example, using serum albumin as an indicator for dietary adherence might be of limited value as it is influenced by other CKD characteristics (eg, low dialysis dose) [60].

Furthermore, improving knowledge on the effect modifiers at play in eHealth self-management interventions for CKD patients is important. None of the included studies provided detail on potentially relevant effect modifiers. We can identify some possible modifying factors based on research focusing on self-management interventions in other chronic, noncommunicable diseases (NCDs). For instance, a longer intervention duration might positively modify the effect of self-management interventions [86]. In addition, the patients’ health literacy level might modify intervention effect [87]. Self-management interventions for NCDs are mostly based on similar intervention principles and behavior change techniques. Moreover, the characteristics of patients suffering from NCDs are often similar. We, therefore, argue that the modifiers found to influence the outcomes of self-management interventions for NCDs in general might also be applicable for similar interventions targeting CKD patients. However, more research is needed to identify effect modifiers to self-management interventions targeting CKD and explore possible strategies to impact these factors.

A large variety of eHealth self-management intervention components were used in the included studies (eg, self-monitoring and education), and the results differed greatly. These findings make it difficult and possibly premature to formulate a potentially ideal palette of eHealth self-management intervention components for CKD patients. However, reviewing results make it possible to identify which intervention components might be more promising than others. For instance, self-monitoring and the use of messages or alerts to nudge patient toward displaying healthy behaviors (see Multimedia Appendix 6) were most commonly reported as the effective components to optimize patient self-management skills.

Furthermore, few of the interventions studied were theory-based. The authors recommend that a strong theoretical foundation is necessary for the planning, design, evaluation, and implementation of eHealth self-management interventions [88]. We recommend building eHealth self-management interventions based on established behavior change techniques, such as formulated in the Behavior Change Techniques taxonomy [55]. Moreover, the use of cocreation methods and appreciative inquiry (such as described in the Center for eHealth Research and Disease Management [89] roadmap for eHealth development) can improve intervention fit with the needs and priorities expressed by professionals and patients.

Ability of health care professionals to monitor and, if necessary, anticipate patient measurements online was reported as an important determinant of implementation. We argue that this ability of professionals to anticipate and act upon patient measurements might reduce patients’ feeling of isolation and/or anxiety caused by independently conducted treatments at home [77] and thereby increase patients’ adherence to implementation. In addition, availability of sufficient skills/knowledge was important for users to continue their use of eHealth technology. If participants are unfamiliar with the use of eHealth, this has been reported to limit their acceptance of eHealth interventions [58,69]. Proper training and tailored tutorials are needed to guide eHealth implementation to optimize knowledge and skills and promote intervention uptake [67,72]. The included studies used various methods to evaluate determinants of implementation. We suggest that future research should use validated tools for measuring implementation quality and related determinants, such as the Measurement Instrument for Determinants of Innovations questionnaire and Determinants of Implementation Behavior Questionnaire [90,91].

Most studies were appraised to be of low to moderate quality. There is a heterogeneity of outcome measurement tools and reporting styles used in the articles included in this review. Therefore, we advise researchers to develop a more standardized approach to the use of outcome measures, guided by, for instance, the formulation of an International Consortium for Health Outcomes Measurement standard set for CKD [92]. In addition, we argue that detailed description and a thorough analysis of study design, methods, and intervention components used, based on a published theoretical framework such as Consolidated Standards of Reporting Trials-eHealth [93], can improve reporting and provide a basis for evaluating the validity and applicability of eHealth trials.

Data on eHealth self-management interventions for CKD patients in developing countries are still lacking, which corresponds with other reviews on eHealth interventions [94,95]. The need to perform such research in developing countries is high. eHealth interventions in these countries have the potential to improve the accessibility and cost-effectiveness of local care and ensure timely delivery of care to rural areas and diverse populations [20,24,96]. Furthermore, 9 studies had an intervention duration of fewer than 6 months. Few studies conducted a follow-up measurement. Forni Ogna et al [66] reported that the positive intervention effects were maintained only during the monitoring period; these effects had vanished 3 months after interruption of the drug adherence monitoring. This finding underlines that the effectiveness of eHealth self-management interventions should be tested during a longer study period and with follow-up measurements.

Of note, 3 studies with fewer than 10 participants were included. One might argue that such studies do not provide robust, generalizable evidence and should be excluded based only on their sample size. However, high-level evidence on the effectiveness of eHealth self-management interventions for CKD patients, for instance, generated by large RCTs, is very limited. Hence, studies with less robust designs are included, as in this stage, we feel that all evidence should be accumulated and taken into account as to broaden our view and deepen our understanding of the usability, implementability, and effectiveness of eHealth self-management interventions for CKD patients. Moreover, this decision is supported by similar systematic reviews on the effectivity of eHealth interventions that also included studies with smaller sample sizes [95,97,98]. That being said, results of this review should be interpreted with some caution.

To our knowledge, this is the first systematic review to evaluate the entire spectrum of studies focusing on eHealth self-management interventions for CKD patients. Our review has some strengths. First, PRISMA guidelines were followed, and a robust search strategy was used in 8 databases. Second, a comprehensive analysis was conducted on the intervention components, outcome indicators, and determinants from the various studies. The kappa value and percent agreement obtained, and thus inter-rater reliability, showed that the validity of the appraisal could be considered fair. Finally, any discrepancies were discussed until consensus was reached.

Nevertheless, several limitations need to be addressed. First, as articles only published in English were included, some relevant articles might have been missed. Second, substantial heterogeneity of interventions and outcome measures made it difficult to draw firm conclusions about the evidence emerging from these studies, and results should be interpreted with caution.

This review provides a comprehensive overview of studies evaluating eHealth self-management interventions for CKD patients. eHealth self-management interventions show promise to improve health outcomes in CKD patients. To adequately assess eHealth intervention effect, future researchers should carefully consider their choice of outcomes (distal vs proximal) based on their sensitivity to capture meaningful change. Also, to enable the standard design and scale-up of effective eHealth self-management interventions for CKD patients, a more detailed understanding of which individual intervention components lead to health outcome improvement and which determinants of the implementation can promote adherence and satisfaction with care is needed.