Juvenile idiopathic arthritis (JIA) is the most common rheumatic disease in pediatric populations [1]. Early diagnosis and active treatment are essential for maintaining physical function and psychological well-being. The treatments aim to control the disease, promote clinical remission, and prevent long-term disability [2-5]. However, to achieve these goals, the management of JIA should be multifactorial [6]. Pediatric-specific issues need tending, such as the use of antirheumatic medications in children and young people, growth retardation, pain and coping, school attendance, psychosocial functioning, dealing with parents, and, in the adolescent years, preparing for the transition to adult care [4,7,8]. For good reason, children and young people need to be closely monitored and supported by specialized rheumatology centers that provide interdisciplinary care using a range of pharmacological, physical, psychological, and educational interventions [6,9-12]. However, several barriers have been identified that hinder this current model of support, delaying the delivery of timely, individualized, and well-coordinated care.

There is an inadequate number of experienced pediatric rheumatologists (PRs) to meet demand and oversee care [7,13-18]. This has resulted in long waiting lists, the centralization of services into tertiary children’s hospitals, and the need for many children and young people to travel long distances to access pediatric rheumatology centers [2,18] or care being delivered by a primary health care provider with no pediatric rheumatology training [2,5,14,18-20]. The World Forum on Rheumatic and Musculoskeletal Diseases clearly states that poor access to health care services can significantly impede diagnosis, appropriate treatment, and health outcomes [14], highlighting the need to overcome these barriers in the delivery of JIA management.

In addition, to achieve optimal health outcomes, children and young people need to adhere to their prescribed treatment plan [21,22], and parents need to support treatment recommendations [2,22]. However, suboptimal rates of adherence are commonly reported [22-24]. For example, a literature review of children and young people with chronic rheumatoid disease reported medication adherence rates as low as 38% and physical activity adherence rates of 40%, particularly during adolescence [22]. The primary reasons included the complexities of chronic disease management and medication schedules, time-consuming nonpharmacological treatments, lack of disease knowledge, and low satisfaction with the health care team [22]. These reasons are not surprising, as a recent systematic review identified 70 studies in which health information was inappropriately tailored to children and young people and their parents’ level of health literacy, increasing their concerns and uncertainties about their condition, treatment options, and shared care decisions [25].

For JIA specifically, further reasons for nonadherence vary across treatment modalities [24] as follows: for oral medications, forgetfulness, taste, and long-term side-effects; for parenteral medications (injectables and infusions), pain and side-effects; and for physical and occupational therapy, forgetfulness, pain, and therapy not being considered necessary [24]. Fortunately, all these reasons are modifiable.

To uphold the expectations of rheumatology care, children and young people should be empowered to take an active role in their disease management by being provided opportunities to improve their health literacy and develop good self-management skills [2,10,25], particularly when considering the long-term benefits these skills will have across their life span. The development of self-management skills is also important as parents only hold a surrogate role in children’s and young people’s health care decisions; therefore, children and young people need to be prepared for their transition from pediatric to adult health care services [26].

A resource-efficient way of supplementing JIA management and the development of self-management skills could be to harness the burgeoning eHealth and mobile health (mHealth) interventions [27]. eHealth is described by the World Health Organization as an activity that delivers health-related information, resources, and services through electronic technology and the internet [28]. mHealth is described as a subbranch of eHealth [28] that uses wireless technology to rapidly uptake, process, and communicate information to support health system efficiency and patient outcomes [29].

The number of studies exploring the potential of eHealth and mHealth interventions for chronic disease management is rapidly increasing. However, most are still at an early stage of development and are limited in their scientific rigor [30-35]; most have been conducted with adults rather than children and young people [31,35], which is interesting, considering that children and young people are experienced users of this form of technology and more likely to use a digital intervention or health app [36-38]. In fact, a recent systematic review identified that children and young people use the internet for at least 1 to 4 hours a day (9438/10,974, 86%) [37] and some type of app every day (719/719, 100%) [39]. Higher rates of use have also been reported for children and young people living with JIA who are at risk of poor psychosocial functioning compared with their peers (>1 hour a day) [36].

However, concerns have been raised about how children and young people use the internet [37,40]. Studies have established that children and young people have poor internet-searching skills, tend to use a 1-word search strategy, briefly skim through search-engine result pages [40], and lack the ability to appraise quality [33,36]. This limits their capacity to find high-quality, personally relevant health information and potentially exposes them to incorrect material [36] or results in them turning to apps and platforms not specifically developed as health resources such as YouTube, Tumblr, and Instagram [41].

Johnson et al [36] believe that for pediatric services to better support the needs of children and young people living with chronic illness, they need to be provided with accessible, developmentally appropriate, and high-quality health-related information. Children and young people with JIA (n=134) agreed, particularly those with low health-related quality of life (HRQoL), expressing an interest in being provided with supportive internet-based interventions [36]. In addition, children and young people participating in feasibility and usability studies and reporting on the delivery of health messages [42], exercise programs [43], symptom monitoring [44,45], and disease management [35,36,44] have also reported high levels of acceptability [42-44], usefulness [35], and satisfaction [43] when using these interventions. However, personal, technical, and device-related barriers have also been identified, which hinder their use [46]. Understandably, before a health care professional can prescribe a digital intervention, it has been suggested that they need at least 3 published papers demonstrating the intervention’s effectiveness [47] to see whether the intervention works in a real-world setting [48].

Internationally, pediatric services cater to children aged 0 to 12 years [49], and adolescents up to the age of 18 years (mean 18.7, SD 2.6 years), before they are transferred to adult services [50]. In this review, we use the term “children and young people” to broadly include all individuals in the age range of 1 to 18 years. We exclude neonates and infants (<1 year) [51].

This systematic review aims to identify what eHealth and mHealth interventions have proven to be effective in supporting health outcomes for children and young people (aged 1-18 years) living with JIA. We anticipate that this review may aid the clinical use of eHealth and mHealth interventions and their integration in arthritis management.

This systematic review complies with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [52]. Before commencement of this review, a protocol for this review was registered on PROSPERO (CRD42018108985) [53]. Protocol questions 1, 2, and 4 were presented in a previous study [46], whereas question 3 is presented in this review [53]:

The database search retrieved 301 studies. Of the 301 studies, 90 (29.9%) were duplicates; 145 (48.2%) did not meet the inclusion criteria based on their title or abstract; and 51 (16.9%) were excluded in the full-text screening because of study design, population, age range, outcomes, or the inability to gain the full text (eg, abstract only, conference presentations, and posters). Approximately 5% (15/301) of studies met the inclusion criteria to be introduced into this review (Figure 1) [65-79]. Of the 15 studies, only 1 (7%) was retrieved in a language other than English (Dutch), and an English version of the same study was attained through ResearchGate [72].

Using the Downs and Black [56] (modified) checklist, the overall mean quality score of the 15 studies was 18.87 (SD 1.92) [65-79]. The scores ranged from 15 to 21, providing a fair-to-good score [57] (Multimedia Appendix 5 [65-79]). There were no disagreements between SB and AF that needed to be resolved by a third author (AC). The 2 areas in which study quality was consistently limited were power and sampling bias; 87% (13/15) of studies had insufficient power to detect a clinically significant effect [65,67-78], and convenience sampling and selection bias may have prevented full representation of the JIA population [68,72,75,79]. Children and young people were selected because of pain experience [66,70,78]; level of disease activity [65,66,69,71,75,78,79]; unlikelihood of medication changes [69]; no other comorbidities or cognitive impairments [65,66,69-71,73-75,77,79]; good visual acuity [75,79]; no hand deformities [75,79]; reduced PALs [65]; access to a computer, tablet, or phone or the internet [65,69,70,72,74]; and level of comprehension and ability to speak and read English [65,66,70,73-77,79], Dutch [72], Spanish [66], or French [73,77]. Methodological concerns were also seen in internal validity because of contamination or unreliable compliance [65-67,69-71,73,75-79].

Of the 15 studies, 3 (20%) fair-to-good–quality studies recorded adverse events [65,66,69]. Teens taking charge (age range 12-18 years) reported the highest number of adverse events (n=72), mostly related to infections (18/72, 25%) and arthritis-related flares (17/72, 24%) [66]. The more serious events involved hospitalization (9/72, 13%) or suicidal thoughts (4/72, 6%). There was no significant difference in adverse events between the IG and CG groups (P=.67) [66]. MisFit Flash (age range 12.8-18.6 years) also reported illness, injury, or pain (9/28, 32%), including arthritis-related ankle and knee pain (1/28, 4%). However, no significant difference was seen in functionality (mean Childhood Health Assessment Questionnaire score), pain, or active joint count during the study [69]. Whereas Rheumates@Work (age range 8-12 years) reported arthritis-related flares, affecting more children and young people in the CG (2/17, 12%) compared with the IG (1/16, 6%) [65].

Of the 15 studies, 10 (67%) studies reported dropout rates (range 0-114) by children and young people (aged 8.7-18.6 years) [65,66,69,70,72-74,77-79]. Dropout reasons before study commencement included not being interested anymore [65,66,74], early withdrawal before allocation [66,77], not receiving allocation [65,70,77], not completing app orientation [70], technical issues [77], and no show and no reason [65]. Reasons during the study period included other health problems [65,69], school and extracurricular activities [69,78], discontinued use [66,74,77], did not complete final web-based measures [77], unable to reach [66], lost to follow-up [66,69,73,74,77,79], and removal because of lack of compliance [74]. No comparisons were made between age or gender [65,66,69,70,72-74,77-79].

Of the 15 studies, 7 (47%) studies reported both the IG and CG dropout rates [65,66,70,72-74,77]. A higher dropout rate was reported in the IG (119/455, 26.2%; range 0-76) compared with the CG (56/449, 12.5%; range 0-56) [65,66,70,72-74,77]. The Canadian Teens taking charge study reported the highest dropout rate (IG: 76/164, 46.3%; CG: 38/169, 22.5%) [73].

To the best of our knowledge, this is the first systematic review to evaluate the effectiveness of eHealth and mHealth interventions in supporting children and young people living with JIA. In total, 10 interventions were identified to support symptom monitoring, physical activity promotion, or self-management development for children and young people aged 4 to 18.6 years. These 10 interventions included 4 (40%) web-based programs [65,66,68,72,73,76,77], 3 (30%) health applications [67,70,71], 1 (10%) telecommunication application [74], 1 electronic diary (10%) [75,78,79], and 1 (10%) accelerometer compatible with a tablet or smartphone [69]. The methodological quality of the studies supporting these interventions ranged from fair [68-71,75,76,78] to good [65-67,72-74,77,79].

Pain assessment was the most common type of intervention used to support symptom monitoring. The interventions My Pain Tracker and eOuch aimed to capture real-time data through children and young people self-reporting pain [71,75,78,79]. Monitoring pain is important as pain is the most frequently reported symptom by children and young people living with JIA [12,82]. Pain can dramatically interfere with physical functioning, coping mechanisms, and quality of life [12]. Stinson et al [79], through the use of eOuch pain diaries, demonstrated a correlation between pain intensity and the impact pain can have on emotional well-being (unpleasantness) and activities of daily living (interference), reinforcing the need for ongoing comprehensive pain monitoring, which could allow the health care team to make timely recommendations and prevent poor health outcomes [83-85].

However, there is no consensus on the required number of real-time assessments, per day or week [71,86,87], to ensure the collection of high-quality data and avoid the burden of momentary reporting [71]. Instead, a large variation, ranging from 2 to 9 times a day for children and young people, has been seen [87]. In this review, real-time pain monitoring ranged from 1 to 3 times a day [71,75,78,79] or when needed [71]. Lee et al [71], through My Pain Tracker, compared these reporting frequencies, finding that children and young people preferred once-a-week or when-needed pain assessments to avoid thinking about their pain. Although more details in pain data were collected from once-a-day reporting, and, for some children and young people, adherence to once-a-day reporting was easy as it became a routine [71], more research is needed on reporting frequencies.

Real-time pain monitoring also exposed differences between real-time pain and recall pain assessments [75]. Recall pain measurements were higher and predictive compared with average real-time pain measurements, influenced by the children’s and young peoples’ most intense pain and last pain score [75]. This is known as recall bias or peak-end effect [75,88]. This nonequivalence between real-time pain assessments and recall pain assessments adds significance to previous research by Stone et al [88], highlighting methodological concerns around relying on retrospective pain assessments, especially when considering the length between rheumatology appointments.

Longitudinal variances were also seen between real-time and recall pain monitoring [75]. Stinson et al [75] and Stone et al [88] both demonstrated a weak correlation with within-person data when pain was defined as >0/10, which is the most common pediatric pain scale. This suggests that real-time and recall pain assessments cannot be compared or used interchangeably when assessing long-term changes in pediatric pain [75]. Considering that the length of the studies in this review was only 2 to 8 weeks [71,75,78,79], further research on the longitudinal effects of real-time pain monitoring is needed.

The use of real-time symptom monitoring for children and young people is also supported by previous work in reducing the recall time to days, hours, or minutes [87], and importantly, a recent systematic review, reporting on real-time monitoring using mobile technology, suggests that it can be successfully implemented from the age of 7 years [87]. In addition, a study considering adults with chronic illnesses supports real-time monitoring for the identification of exacerbations, confidence in self-management, and prevention of hospital admissions [89]. In this review, the intervention EQ-5D-Y-5L endorsed this finding, as remote HRQoL monitoring identified, with satisfactory diagnostic accuracy (P<.001), moderate to high levels of disease activity, promoting the need for adjustments with prescribed treatments and rheumatology consultation frequency [67]. Further research is now needed on this web-based PR– children and young people interaction and the impact remote monitoring may have on safety [67,90].

In this review, 13% (2/15) of studies reported positive feedback from PRs after they reviewed web-based assessments during consultation [68,76]. PRs reported that SUPER-KIDZ pain assessments increased time efficiency compared with a paper-based assessment [76]. However, ePROfile increased PR satisfaction with the care they provided as the HRQoL discussion improved and the number of psychological referrals increased [68]. Although these findings were not significant, reviewing pain and HRQoL during consultation is important as children and young people with JIA have significantly lower HRQoL compared with that of healthy children and young people, and children aged 8 to 12 years with JIA have lower HRQoL than that of children with other chronic health conditions [9].

Interestingly, the use of web-based portals in adult rheumatology has been long standing. The Feed Forward System, for example, used in Sweden generates a patient’s progress over a period and has been successfully used to guide health care provider recommendations and aid the development of patient self-management skills [91].

For JIA, feasibility studies considering web-based portals also support their use, reporting that this form of technology can increase children’s and young peoples’ (aged 5-22 years) feeling of control [92,93].

Regrettably, parents and children and young people did not report the same level of satisfaction with the ePROfile consultation as PRs [68]. Haverman et al [68] suggest that this may be as they are already happy with the quality of their care. Nonetheless, many factors that can influence children’s and young peoples’ opinions on digital assessments need to be considered. First, they can be influenced by the assessment experience; they need graphical and tailored feedback to encapsulate their results and catch their interest [94]. In addition, children and young people may not value and understand the importance of monitoring symptoms, disease, and general well-being (mood, fatigue, and functional ability) [44,95] or the need for a person-centered framework that builds partnerships between families and health care teams [92]. Further research on the use of web-based portals for children and young people is needed.

In this review, 20% (2/10) of interventions, Rheumates@Work and the wearable activity tracker, Misfit Flash, aimed at improving self-management behavior by promoting physical activity for children and young people (aged 8-18.6 years) [65,69,72]. Of these 2 interventions, only 1 (50%) Rheumates@Work, demonstrated a moderate but clinically meaningful effect on physical activity, improving endurance time, PAL, and MVPA for children and young people (aged 8-12 years) [65,72]. This finding is important as children and young people with JIA are less physically active [96,97] and spend more time in sedentary activities than their peers [96]. Improving physical activities helps to retain musculoskeletal function, muscle strength, and functional capacity [98].

In addition, increased physical activity did not exacerbate disease activity or pain in the IG compared with the CG [65,69]. In fact, no significant difference was reported [65,69], and absenteeism from school decreased [65]. These findings are encouraging, especially considering the related impact JIA can have on reducing academic performance, as depicted by Bouaddi et al [99] and Laila et al [100]. Although these findings are limited, they will add to the growing body of evidence reporting that exercise therapy is well-tolerated by children and young people with JIA [98,101,102], further supporting physical activity as a helpful and necessary treatment modality, improving adherence [24,103].

In this review, 30% (3/10) of interventions—iCanCope [70], Teens taking charge [66,73,77], and the iPeer2Peer program [74]—supported self-management development for children and young people (aged 8-18 years). These interventions (including Rheumates@Work [65]) are typical behavior change technique interventions used for children and young people [31]. They support self-management through the development of disease-specific knowledge, goal setting, self-management strategies, and social support [66,70,73,74,77].

In this review, identified in the meta-analysis, children and young people participating in the self-management programs Teens taking charge and iPeer2Peer Program reported a small but nonsignificant improvement in pain intensity, disease knowledge, and self-efficacy scores [66,73,77]. However, high statistical heterogeneity was also seen within the results. This may be because of several reasons. First, a range of comparators was used for the CG. For example, in 2 of the 3 Teens taking charge studies, the IG was compared with an active CG rather than usual care. The CG also received an eHealth intervention 12 publicly available health education websites with phone support to support care. Improvements were then seen in both the IG and CG, reducing the mean difference between the groups [66]. Most digital studies primarily focus on a single intervention to demonstrate the intervention effect rather than comparing different digital interventions. However, it is this direct comparison that can reveal a more effective intervention [104]. The CG’s intervention, the use of health care workers signposting quality health education websites to support self-management skills [105,106] and improve well-being [107], is supported in the literature. Therefore, acknowledging the improvements seen in the CG is important, as the use of publicly available websites can be a cost-effective solution for dissipating health information among the masses to support the delivery of health care [28]. For example, a study of adults living with chronic pain (n=20; aged 18-74 years) explained that if they had been provided with quality pain-related information, it might have prevented the desperation and anxiety they experienced, especially during the first few years [107].

Another explanation for the moderate to high statistical heterogeneity may have been that the studies were conducted in different countries, within different health care delivery systems, with different levels of pre-existing support [73]. Differences exist with the pediatric rheumatology workforce worldwide [7,13-18] and within publicly funded and self-funded health care systems [108,109]. Differences also exist in PRs’ opinions on pediatric self-management and the use of an interdisciplinary approach to care [108]. Successful publicly available digital interventions may be a solution to transcending these boundaries and universally improving access to care [28,106]. Further comparisons between different self-management interventions is needed, especially when considering the dropout rates in this review, which, for the self-management programs, were higher in the IG [65,66,70,73,74]. These dropout patterns were similar to a recent systematic review predicting dropout rates in adults, with dropouts occurring at the beginning and over the course of the intervention [110].

There may also be no one-size-fits-all intervention, or there may be a need for a combination of interventions. For example, the iCanCope pain self-management application combined 2 interventions (real-time pain monitoring and self-management) and then compared this combination to standalone pain monitoring. This combination demonstrated a greater decrease in pain intensity scores (>1 point, 0-10 on the numerical rating scale) [70]. Although this finding was not significant, the inclusion of self-management programs could be clinically beneficial. Improving and providing effective educational interventions early in childhood should be when children and young people are beginning to develop their health behaviors [23]. Studies have shown that a high level of health literacy can support informed decision-making [111-113]; treatment adherence, especially for nonmedication interventions [113]; and the prevention of chronic health-related problems [22].

Unfortunately, not all the results of this review are promising. Across the studies, the interventions had no effect on pain interference [66,73], HRQoL [65,66,69,70,73,74,77], anxiety, depression, pain coping [66,73], disease activity [69,72], functional ability [65,69,70], or treatment adherence [73,77]. In addition, only 20% (3/15) of studies considered long-term follow-up [65,66,73]. More research is needed to gain wider health-related benefits.

It is essential that several limitations are considered when interpreting the findings of this review. First, our search strategy was restricted to an academic context, using eHealth electronically indexed health databases that publish peer-reviewed journals, rather than apps within commercial stores. This means that our results may not provide a true reflection of the health apps available for JIA. This decision was based on the commonly reported shortcomings of health apps available to the general public that are related to data safety and lack of rigorous testing [114].

Second, the selection criteria in this review deviated from our systematic review protocol [53]. In the protocol, we outlined that the comparator or CG was to receive usual care, with no eHealth or mHealth input. Instead, we included 13% (2/15) of studies comparing an eHealth intervention to another digital intervention [66,73], as a preliminary pilot study of this intervention met our inclusion criteria to be included in this review [77]. This decision enabled us to provide the most up-to-date evidence for this intervention.

Third, our findings in this review supporting the use of real-time monitoring and web-based assessments were based on descriptive summaries. The use of a narrative, descriptive methodology to summarize, synthesize, and report the results is at risk of reporting bias. To reduce this risk, all authors internally reviewed all the stages of this review.

In addition, there were methodological concerns in the data reported by some studies because of performance bias. It was not possible to blind children and young people from the intervention, which could have resulted in a placebo effect. For example, Rheumates@Work reported improvements in both the IG and CG for MVPA and participation in physical education classes. Baseline testing may have made the CG more aware of the need to improve their physical activities [65]. The interventions’ true effects may have also been overestimated, as activity levels entered by children and young people in the exercise diaries did not match the accelerometers. Overreporting exercise is not uncommon. Various correlations have been reported between exercise diaries and accelerometer recordings in the general population (r=0.52) [115]; among children and adolescents (reliable coefficient ranges r=0.5-0.93 and validity coefficient ranges r=0.03-0.88), with children being in the lower range [116]; and for JIA (light PAL and MVPA r<0.24; rest and PAL r=0.41) [117]. Unexpectedly, a 4% inaccuracy has also been identified in accelerometer recordings for JIA in light PALs (effect size 1.2) because of nonwearable periods (aquatic activities and ball games) [117]. Awareness of these possible variations enables correction. For example, Armbrust et al [117] recommend, for research purposes, the use of accelerometer recordings (7-19 days) and an activity diary (>13 days). Another feasible suggestion may be the use of wearable forms of digital technology (ie, a smartwatch) [27]; however, more research is needed to overcome the nonwearable periods such as contact sports [118] or while attending school [119].

Finally, the generalizability of our findings may be limited. We included 40% (6/15) of studies where several children and young people were categorized as unknown, not yet diagnosed, or other (55/1438, 3.82%; range 2-37) [68,70,73,76,77,79]. Dissecting the results to target children and young people specifically with JIA was not possible (Table 2). For a fair-quality study, which reported the highest number of children and young people in this category (37/101, 36.6%), our data extraction focused on the consensus view of pediatric rheumatology providers (PRs, allied health experts, and pain experts) rather than the children and young people [76], which is an area of research that is currently limited.

Evidence that supports the inclusion of eHealth and mHealth interventions in JIA management is on the rise; however, this evidence needs to be considered cautiously. Confidence in the results is reduced because of low sample size, wide CIs, high statistical heterogeneity, and no similar effect being seen across similar studies. More rigorous research is needed that focuses on the longitudinal effects of real-time monitoring, web-based PR–children and young people interactions, comparison of self-management strategies, and the use of wearable digital technology as an objective measurement for monitoring physical activity before any recommendations informing current practice can be given.