Globally, it is estimated that approximately 1 million adolescents and young adults (AYAs) between the ages of 15 and 39 years are diagnosed with cancer each year [1]. Continual advances in cancer therapies now mean that the overall 5-year cancer survival rate among AYAs has increased to more than 80% with survival among some cancer diagnoses (eg, Hodgkin lymphoma, melanoma, and thyroid carcinoma) now exceeding 90% [2]. However, young people who have been diagnosed with cancer often face a myriad of physical, emotional, and psychosocial challenges because of their diagnosis and treatments [3,4]. During treatment, young people often experience prolonged periods of hospitalization and a number of symptoms and side effects such as neutropenia, nausea, alopecia, mucositis, and neuropathy. Post treatment, in survivorship, there is substantial evidence that AYAs diagnosed with cancer are at increased risk of developing long-term health conditions and experience high levels of pain, fatigue, and poor quality of life throughout their life course [5-7]. These difficulties are challenging for AYAs living with and beyond a diagnosis of cancer to manage and are faced at a time when they, as young people, should be establishing independence and autonomy [8,9]. Continual efforts are, therefore, being made in cancer care, research, and policy to ensure AYAs diagnosed with cancer receive the specialist medical, emotional, and practical support they require both during and after their cancer treatment [4,10,11].

Electronic health (eHealth), mobile health (mHealth; interventions delivered using mobile devices), and digital health interventions apply modern computing and technology innovations in the context of health care provision (the encompassing term digital health interventions has been adopted for the purpose of this review) and have been proposed as strategies to support young people with cancer manage the challenges associated with their diagnosis and treatment [12-14]. This is significant for AYAs in the context of their digital native status; for this population, continued exposure to and integration of digital interventions is the norm [15]. In the context of cancer, digital health interventions have the potential to widen access to and reach of support available to young people with cancer, particularly those being treated as outpatients or receiving long-term follow-up care. Moreover, the delivery of self-directed interventions remotely through digital technology has the potential to ease pressures on face-to-face services and overcome typical geographical and time-related constraints faced by patients, issues particularly pertinent among young people living with and beyond a diagnosis of cancer [16,17].

As demonstrated within the narrative review by Devine et al [18], there now exists a diverse range of digital health interventions for young people with cancer, which contain a variety of elements and functions. This is positive and reflects AYAs’ preferences for information resources and self-management tools relevant to their diagnosis and experiences of cancer to be made available in digital formats [19-21].

In the digital health context, previous reviews of digital interventions have focused on health behavior change and have identified a number of key components that influence intervention outcomes. Existing reviews of digital interventions targeting health behavior change suggest user involvement in intervention design, mode of delivery (eg, Web-based, mobile based, through an advisor, telephone, or e-mail), synergistic use of behavior change techniques, and usability (ie, how easy is the digital health intervention to use and engage with) heavily influence intervention outcomes [22,23]. In this review, assessing these same components and also the quality (ie, the engagement, functionality, aesthetics, and subjective appeal) of interventions is progressive and allows the utility of digital health interventions for AYAs diagnosed with cancer to be more definitively established. Moreover, assessing factors, which influence the engagement and compliance of AYAs living with and beyond a diagnosis of cancer with digital health interventions provide important insights into the feasibility of delivering self-directed interventions to this population in digital formats [24,25]. Understanding which component features of digital health interventions are most acceptable to AYAs diagnosed with cancer and whether such components affect intervention outcomes is critical to the development and evaluation of further digital interventions for young people with cancer [26]. Such data can be used to inform the design, development, and implementation of high-quality effective digital health interventions designed for AYAs diagnosed with cancer.

The objective of this review was to identify, characterize, and fully assess the quality, feasibility, and efficacy of existing digital health interventions developed specifically for AYAs living with and beyond a diagnosis of cancer.

This review aims to address the following questions:

The full protocol for this review has previously been published [27]. To summarize, a literature search for digital health and technology interventions developed specifically for or piloted among AYAs diagnosed with cancer was conducted. Digital health interventions for the purpose of this review encompassed any eHealth, mHealth, or digital health effort, which applied modern computing and communication methods. The review was carried out following the Preferred Reporting Items for Systematic Review and Meta-Analyses guidelines [28].

Studies were eligible if they were written in English and published in a peer-reviewed journal and reported or described any existing digital health intervention designed specifically for young people, aged 13 to 39 years, diagnosed with cancer. In this review, digital health interventions include any eHealth, mHealth, or digital health effort, which applied modern computing and technology innovations in the context of health care provision. Participants of interest are those aged between 13 and 39 years, defined as teenagers, adolescents, or young adults living with or beyond a cancer diagnosis, and this was inclusive of survivors of pediatric cancer who fell within the age bracket of interest.

Studies were excluded if they had insufficient detail on the target population or included an incomplete and vague description of the digital health intervention of interest. If a study reported on interventions developed for young people with comorbid conditions other than cancer or if young people with cancer were not the main focus of the study, then the study was excluded. Studies that focused on the use of digital health interventions by parents or survivors of cancer over the age of 40 years were excluded, as were studies where the mean age of the sample was over 39 years.

Bibliographic databases (PubMed, Web of Science, and EMBASE) were searched in August 2016 and again in October 2017 for articles written in English and published to date in peer-reviewed scientific journals. A combination of Medical Subject Heading terms and keywords was used. These are available in Multimedia Appendix 1.

GP and LM screened the titles and abstracts of all studies identified during the search using the predetermined eligibility criteria of any study. The interrater agreement between both authors on the eligibility was high (Cohen kappa >0.90), and any instances of disagreement were resolved through discussion.

Data extraction was conducted by all authors using a template designed to collate details on each digital health intervention. Data included (1) study characteristics (country, design, sample size, target population, recruitment setting, aim, and methods), (2) platform development and design process (steering committee and patient and public involvement), (3) digital health intervention primary outcomes (mean change and effect size if applicable), and (4) feasibility of delivering the intervention (acceptability, compliance, recruitment response, and retention to the intervention). The mode of digital health intervention delivery was coded into automated functions, communicative functions, and use of supplementary modes based on the coding scheme used by Webb et al [23]. An adapted version of the Mobile App Rating Scale (MARS) was used to group and classify reported engagement, functionality, aesthetics, information quality, and subjective quality of each digital health intervention [29]. Specifically, the theoretical background and strategies scale of the original MARS tool was used to classify the intervention features and theoretical design of each intervention. Alongside data extraction, methodological quality of the included studies was simultaneously assessed using the Quality Assessment Criteria for Evaluating Primary Research Papers from a Variety of Fields (QualSyst) tool [30]. QualSyst includes scoring systems for quantitative and qualitative studies; the maximum summary quality score for qualitative studies is 20 (10 items), and the maximum summary score for quantitative studies is 28 (14 items). Summary quality scores have been reported as percentages of maximum total scores, ranging from 0% to 100%. The higher the percentage score, the better methodological quality of the study, but no studies were excluded based on limited or reduced methodological quality. Following data extraction, included studies were rereviewed by GP, LM, and KM. Any discrepancies were resolved by discussion.

Figure 1 outlines the search process. A total of 4731 studies were identified through the search. After screening the title of each paper, 195 were identified as potentially eligible, and the abstracts were screened. The full texts of 43 papers describing 38 studies were reviewed and subsequently selected for inclusion.

The characteristics of each of the 38 studies are summarized in Multimedia Appendix 2. The included 43 papers reporting on these studies were published between 2002 and 2017. A range of study designs was reported. Of the included studies, 12 used a cross-sectional single-group design [31-42], 11 were randomized controlled trials (RCTs) [43-55], 7 were of single-group repeated measures design [56-65], 4 were of qualitative design [66-69], 2 discussed the development of a digital health intervention [70,71], 1 used a mixed-methods approach [72], and 1 was a non-RCT [73]. Sample size ranged from 6 to 375, and age of participants ranged from 10 to 55 years (mean age <33 years). All included study participants were reported within the AYA age range (13-39 years) at the time of diagnosis. The duration of studies ranged from single-use interventions (eg, virtual reality [VR] glasses used during lumbar puncture) [73] to interventions available over long durations (>6 months; eg, Partnership for Health-2, a Web-based smoking cessation intervention, which included a 15-month follow-up) [48].

Multimedia Appendix 2 outlines the methodological quality of each quantitative and qualitative study. The QualSyst scores are reported as a percentage to allow a comparison to be drawn across study designs, as there are different assessment criteria for quantitative and qualitative studies [30]. Scores ranged from 35% to 100% and were distributed across this range, varying between and within study design. The mean score was 75% for RCTs (n=13) [43-55], 71% for the non-RCT study (n=1) [73], 70% for cross-sectional single-group studies (n=12) [31-42], 74% for repeated measures studies (n=10) [56-65], 62% for platform development studies (n=2) [70,71], 65% for the qualitative studies (n=4) [66-69], and 95% for the studies using a mixed-methods approach (n=1) [72]. Full details of the methodological quality of the papers can be found in Multimedia Appendices 3 and 4.

Of the 38 studies discussed across the 43 research papers included in this review, 14 were designed with young people’s identifiable involvement in the process, and 8 had expert input or included a steering group in the design process.

A total of 15 interventions focused on symptom management and medication adherence [33,38,42,52,53,61-63,67,72,73], 8 on self-care [33,35,38,40,44,49,57,58,60], 15 on behavior change [35,37,40,44-47,50,54,64,65,70,71] (of which 6 addressed physical activity behavior [35,45-47,50]), 7 on negative emotions [31,40,44,49,57,58,74], 7 on physical health [33,35,44,46,47,50,63], 5 on anxiety or stress [41,43,49, 50,71,73], 5 on goal setting [35,46,47,54,64], 4 on happiness and well-being [31,40,44,70], 4 on depression [31,40,44,49], 3 focused on relationships [40,66,71], 1 on education [40], and the focus of 1 intervention was to provide entertainment [40]. Across the included studies as a collective, there were multiple occasions where the platform had more than 1 target behavior, as noted throughout this section.

The features used in the interventions were assessed using the MARS classification tool and are discussed in this section. There were 11 categories including an other category. Overall, 5 studies used assessment [42,45,62,63,73], 7 used feedback [32,40,42,45,54,65,66], 16 used information or educational strategies [31-33,37,43-47,51,62-65,68,69], 12 used monitoring or tracking [35,41,44,45,54,61-65,71,72], 8 used goal setting [31,35,40,44-47,54], 9 studies included advice or tips and skills building [44,45,51,56,60,62,65,69,71], and 1 used strengths-based strategies [64]. None of the studies included for review used mindfulness and meditation, relaxation, or gratitude strategies. Furthermore, 14 used strategies characterized as other [31,33,34,38,39,48-50,53,54,57,58,60, 61,70]. Reported theories used to inform the intervention features included the theory of reasoned action [65], the Adolescent Resilience Model [49], the Hope Process Framework [39], the self-regulation model of health and illness [53], social cognitive theory [53], cognitive behavioral therapy [57,58], and the Symptom Management Theory [60].

Multimedia Appendix 5 outlines the mode of delivery used for each digital health intervention, including details on automated and communicative function features within the platform. Table 1 illustrates the different digital health interventions described in the 38 studies (43 papers) included in the review. As shown in Table 1, there were 5 interventions in the other category; these included CD-ROM, computer program, digital storytelling, therapeutic music video, and e-mail. The following section summarizes the key outcome measures and findings from the studies categorized by platform and mode of delivery used. Multimedia Appendix 5 gives more detailed insight into the automated functions, communicative functions, and supplementary modes of communication used in each study.

As shown in Table 1, there were 5 interventions in the other category: CD-ROM, computer program, digital storytelling, therapeutic music video, and e-mail [34,49,51,55,66,64]. The focus of these studies included building resilience [49], symptom management [49,55], cognitive function [34], education [51], social therapy [66], and health-promoting behaviors [64]. All studies were concluded feasible and acceptable to young people with cancer, with the majority reporting good uptake and engagement from participants.

A total of 20 studies reported young people’s subjective experience of using the digital health intervention [38,39,43,45,48,50-52,54-58,61-63,66,67,69,72]. Subjective experience was typically measured as user satisfaction or appeal of the contents of the intervention. Within 11 studies, participants reported that they would either use the intervention again or would recommend it to a friend [38,39,48,51,52,55,57,58,61,62,65,67]. Very few studies reported participant’s feedback on areas for improvement or recommendations for further platform developments. Of the studies that reported feedback, it was generally that the platform had technical problems, the visual design was too simple, or that the digital platform for communicating with other young people with cancer did not replace personal connection [66]. There was no clear pattern between intervention characteristics (delivery mode and focus of functional components) and engagement or adherence. Reasons for poor engagement or noncompliance were typically either not reported or attributed to recurrent illness [63,50]. Within 1 website-based study [69], incentives were introduced to improve compliance. Some studies reported differences between the engagement and use of different features: for example, Rabin et al [45] reported that participants viewed pages on physical activity logging pages more often than physical activity tip pages of the intervention website (11.38 days vs 0.5 days). Similarly, Mendoza et al reported differences between participants’ frequency of viewing, commenting, and liking Facebook posts within their intervention (92.3% vs 65.4% vs 50%, respectively).

Only 7 of the 43 articles reviewed provided effect size within the original manuscript. Because of the heterogeneity in outcomes and differences in the characteristics of the intervention, it is not possible to make comparisons between the studies. A table summarizing the relevant data is available on request.

The studies included in this review did not specifically report intervention uptake and reach. The total sample size of all studies gives some indication as to the number of people the interventions reached as a whole and the breakdown of where studies were conducted provides some guidance as to the characteristics of those participants. There was a total sample size of 1935 participants across the 38 studies. The majority (n=23) of the digital health interventions were designed in the United States [31,32,34,35,39-47,49,51,54,56,59-61,64-66, 68,73], 3 were designed in Canada [33,62,63,67], 2 in the Netherlands [36,38], 2 in Sweden [37,71], 2 reported multiple sites across different countries [48,52,53], and the country was not reported for 6 studies [50,55,57,58,69,70,72,75].

In this review, we have focused our attention on digital health interventions for AYAs diagnosed with cancer. We are not alone in our interest in considering digital health–driven interventions for AYA populations at this specific illness foci level [18] or indeed other relevant areas such as mental health [19], complex health care needs [75], and lifestyle behavior interventions for survivors of child and young adulthood cancer [13]. A recent narrative review of digital health interventions targeting AYA cancer survivors demonstrated the range of digital modalities used to support young people with cancer [18]. Our review has moved beyond the review of Devine et al [18] by not only identifying specific interventions but also drawing out components that contribute to appropriate digital interventions for our target population (AYA diagnosed with cancer). Our use of the Mode of Delivery [23] and MARS criterion [29], respectively, has allowed our synthesis to identify key components that may influence successful uptake of digital interventions to support this population in the future.

As stated in the Introduction section, we posed 4 key questions in this review. We have revisited these questions to frame our discussion.

Our review has a number of strengths and limitations. We focused our review attentions on digital health interventions designed specifically for AYAs with cancer, and we used broad age inclusion criteria in this regard, from 13 to 39 years. Although this may seem too broad to some, to ensure our review was inclusive as possible and of international significance and relevance, we drew on a range of relevant cancer policy context definitions of AYA [81]. There were some challenges encountered with this, particularly in terms of papers, which included the older spectrum of our target participants (>26 years). Given the international variations in definitions of AYAs with cancer, papers that included these upper ranges of AYA had to be excluded from the review, as it was not possible for us to readily identify data specifically focused on the population up to 39 years, particularly if the intervention had been developed in the context of a wider adult cancer population. Although we searched a range of databases, these were limited to the most common, and we limited our searches to peer-reviewed articles, thereby excluding gray literature. Also, as we limited our searches to papers published in English language only, this may be considered a limitation by some.

To meet our review objective of identifying, characterizing, and fully assessing the quality, feasibility, and efficacy of existing digital health interventions developed specifically for AYAs living with or beyond a cancer diagnosis, our data extraction process was long and detailed. In addition to our study characteristic extractions, we also used 2 specific rating tools relevant for a focus on digital interventions: Mode of Delivery [23] and MARS [29]. Overall, the Mode of Delivery proved a useful and straightforward tool to use, but we encountered some difficulties with the MARS tool. It became increasingly apparent that for the impact of this tool to be realized, one requires full and ready access to the particular app being reviewed and rated. As we were reviewing the published evidence of digital interventions (and not just digital interventions that were publicly and commercially available on app stores such as others who have used the MARS tool in previous reviews [82,83]), we had only narrative descriptions of the apps or interventions to go by in the papers and on occasion, supplemented with screenshots of aspects of the intervention. We were therefore limited to only reporting selected, but still useful, thematic information. Additional items of the MARS were answered if published detail allowed. Although we persevered with this extraction tool throughout our review process, we were unable to draw insightful conclusions because of omission of detail on user interactivity with the digital health intervention, functional performance, ease of use, and graphic design features within manuscripts. We reflect on this as a limitation of the MARS tool itself as much as our review.

This review is positive in that it has highlighted that multiple digital health interventions do now exist to support AYAs diagnosed with cancer. The everyday technology-driven environment in which we now live has expedited the development pathway for digital interventions in health care contexts. However, within our review, it was rare to identify innovations that are ready to be or have already been deployed and implemented at scale. We find ourselves with a case of digital health pilotitis and efforts now need to shift; therefore, the most robust evidence-based innovations are routinely implemented in clinical practice. There is insufficient evidence to state conclusively which form of digital intervention is the best approach to support young people with cancer. Currently, it is challenging to provide clinicians working directly with AYAs with cancer with definitive options for valid, reliable, and robustly evaluated digital tools and interventions to use as part of their health care services. Therefore, to really establish the impact of digital interventions on health-related outcomes of AYAs with cancer and the economic value of implementing digital interventions on service design and service delivery, future endeavors should prioritize upscaled and robust outcome-driven interventions and associated evaluations.