This study did not involve the collection of primary data from human participants. All data were derived from previously published sources and are appropriately cited. Therefore, ethical approval was not required.

This review was registered with PROSPERO (registration number: CRD420251108938). During the review process, this study strictly adhered to the registered protocol. In the initial search strategy, broad telehealth-related terms were applied to comprehensively assess the multidimensional effects of telehealth interventions on pediatric patients with cancer. However, during the systematic search and screening process, it was observed that most (over 90%) of the studies meeting the original inclusion criteria used mHealth-based interventions—such as mobile apps, web-based platforms, or smart devices—rather than traditional telehealth approaches (eg, telephone follow-ups or video consultations) [22,23]. Meanwhile, a sufficient number of RCTs were identified. To enhance internal homogeneity and ensure that the review provides precise, high-quality evidence with clear practical implications, the scope of this review was refined to focus exclusively on RCTs of mHealth interventions. The registration record has been updated accordingly, and the systematic review and meta-analysis were conducted in strict accordance with the most recent registration record [24].

The systematic review and meta-analysis were conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 statement [25] (Checklist 1). A comprehensive search was conducted across PubMed, Web of Science, Cochrane Library, Embase, CINAHL, Scopus, ScienceDirect, ProQuest, PsycINFO, OVID, Chinese National Knowledge Infrastructure, WanFang, Weip Database, and China Biology Medicine database, with each searched from inception to August 1, 2025. The search terms combined both Medical Subject Headings (MeSH) terms and free-text keywords: (cancer or oncology or tumor or neoplasm or leukemia) and (child or pediatric) and (mHealth or internet-based intervention or software application). The detailed search strategy is provided in Multimedia Appendix 1.

Two researchers (HY and YX) first removed duplicates using EndNote X21 (Clarivate Analytics) and then screened the studies in 2 steps: first by title and abstract, and then by full text. Any discrepancies that arose during the screening process were resolved through discussion between the 2 researchers, or, if necessary, by consulting a third researcher (XM). The literature screening followed the participants, intervention, comparison, outcomes, and study design [26]. Inclusion criteria were as follows:

Exclusion criteria included (1) reviews, abstracts, theses, systematic reviews, meta-analyses, and case reports; (2) unavailable full texts; (3) interventions targeting only parents; (4) publications not in Chinese or English; and (5) interventions limited to calls or videos.

Two researchers (HW and YW) independently extracted the following data using a standardized form, including the first author, publication year, country, sample size, age, sex, cancer type, intervention duration and setting, mHealth platform, BCT clusters, intervention control groups, and outcome measurement tools.

Two researchers (HY and YW) independently assessed the methodological quality of the included studies using the revised Cochrane Risk of Bias 2 tool, evaluating selection, performance, attrition, detection, and reporting biases, each rated as low, some concerns, or high [27]. The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) framework was applied to appraise the quality of evidence using GRADEpro. Results were categorized into 4 levels (high, moderate, low, and very low) according to predefined criteria. The assessment incorporated study design, risk of bias, inconsistency in population, heterogeneity of findings, statistical precision of effect estimates, and publication bias [28]. Discrepancies were resolved through discussion or, if needed, consultation with a third researcher (XM).

This study adhered to the Cochrane Handbook [29] for data synthesis and analysis, using RevMan 5.4 (Nordic Cochrane Center) and Stata 15.0 (StataCorp) for statistical procedures. Effect measures included standardized mean differences (SMD, 95% CI) and odds ratios (OR, 95% CI). Heterogeneity was assessed using I² and P values; when I² >50% and P<.1 (indicating substantial heterogeneity), a random-effects model was used; otherwise, a fixed-effects model was used. Sensitivity analyses were conducted using a stepwise exclusion, while subgroup analyses and meta-regression were used to explore heterogeneity. Publication bias was assessed with funnel plots and Egger’s test. A P<.05 was considered statistically significant.

A total of 7215 studies were identified, with 24 meeting the inclusion criteria for the systematic review and meta-analysis (Figure 1).

This study included 24 RCTs published between 2017 and 2025, the main characteristics of which are summarized in Table 1. Most of the studies (19) were conducted in China [30-48], while one each was conducted in America [49], Canada [50], Iran [51], the Netherlands [52], and Turkey [53]. These studies involved a total of 2645 children with cancer, with sample sizes ranging from 18 to 444 participants. The cancer types included hematologic malignancies, brain tumors, bone tumors, neuroblastoma, sarcomas, and other solid tumors. Participants were aged 0‐18 years. The intervention durations varied from 7 days to 2 years. The intervention settings included home-based (9 studies [30-33,35,37,41,51,52]), hospital-based (3 studies [34,50,53]), and combined home and hospital (12 studies [36,38-40,42-49]).

Regarding intervention platforms used for the experimental groups, 15 studies [30-33,35-39,42-47] used WeChat (a widely used mobile-based social media and communication app in China that supports messaging, group communication, and content dissemination[54]), 1 used a website [50], 3 used professional medical apps [41,52,53], and 5 used mixed platforms [34,40,48,49,51]. All interventions were conducted under the guidance of medical staff. Figure 2 shows the BCTs used in each study. A total of 13 studies involved goals and planning [30,31,34-36,38-40,42,43,45,46,50]; 19 studies included feedback and monitoring [30-35,38,39,41-43,45-50,52,53]; 21 provided social support [30-44,46-48,50-52]; 21 included shaping knowledge; 3 involved repetition and substitution [36,49,53]; 4 included reward and threat [30,32-34]; 3 conducted comparison of outcomes [31,39,47]; 16 incorporated natural consequences [30,31,34,36,39-41,43-49,51,52]; 2 involved regulation [36,44]. A total of 3 reported security measures such as password login and patient-specific data verification [41,50,51]. The control groups received usual care.

Among the 24 included studies (Figures 3 and 4), 3 were rated as low risk of bias [32,41,49]. One study had a high risk due to unaddressed missing data [50]; the remaining 20 had some concerns (1 study [52] due to potential bias in the intervention, 1 study [37] for not reporting allocation concealment, and the other 18 due to subjective outcome assessments potentially influenced by positive psychological suggestion).

GRADE evidence quality ratings are presented in Multimedia Appendix 2. Evidence for infection incidence, overall incidence of PICC complications, PICC phlebitis, PICC puncture site bleeding, PICC puncture site infection, PICC occlusion, PICC catheter dislodgement, and PICC catheter displacement was rated as moderate. In contrast, evidence quality for QoL, PICC thrombogenesis, health knowledge, children’s self-management ability, and treatment adherence was rated as very low.

For most outcomes, excluding any single study did not alter the effect sizes of mHealth-based interventions, indicating stable results (see Multimedia Appendix 3). Exceptions were observed for QoL, PICC complication incidence, and treatment adherence. For quality of life, excluding either study by Li et al [38] or Xie [35] led to significant changes in the overall effect size (SMD 1.52, 95% CI −0.86 to 3.90; P=.21; SMD 0.67, 95% CI −0.03 to 1.38; P=.06), indicating instability. Exclusion study by Hu et al [33] (OR 0.33, 95% CI 0.03-3.19; P=.34), substantially altered the effect estimate for thrombogenesis. In contrast, heterogeneity in treatment adherence decreased markedly after excluding the study by Ding et al [46] (P=.11; I²=45%), while the pooled effect size remained stable (OR 3.72, 95% CI 1.68-8.22; P<.001). Two outcome indicators (health knowledge and self-management ability) were not subjected to sensitivity analysis as they were only involved in 2 studies.

The assessment of publication bias in the effects of mHealth–based interventions on infection incidence, QoL, PICC-related complications, health knowledge, self-management ability, and treatment adherence in pediatric patients with cancer was facilitated by the use of funnel plots. The majority of the funnel plots demonstrated symmetrical patterns in the outcomes, with the exception of those pertaining to health knowledge and self-management ability (Multimedia Appendix 4). In addition, the Egger test was performed to evaluate the incidence of infection (t=0.56; P=.61), the QoL (t=12.12; P=.052), PICC-related complications (t=0.42; P=.69), PICC puncture site infection (t=−0.66; P=.55), PICC phlebitis (t=−1.59; P=.16), PICC thrombogenesis (t=20.91; P=.03), PICC puncture site bleeding (t=−0.39; P=.72), PICC catheter occlusion (t=−2.35; P=.07), PICC catheter dislodgement (t=−2.09; P=.08), and PICC catheter displacement (t=−0.59; P=.66), and treatment adherence (t=5.37; P=.003). The results indicated the absence of substantial publication bias, with the exception of the PICC thrombogenesis, health knowledge, self-management ability, and treatment adherence.

Subgroup analyses and meta-regression were not conducted because fewer than 10 studies were available for each outcome. This decision aligns with the Cochrane Handbook [55] guidelines to prevent false-positive results, insufficient statistical power, and unstable results.

This review aims to evaluate whether mHealth-based interventions are more effective than usual care for pediatric patients with cancer. The quality of the evidence for each outcome measure in the included studies was graded using GRADEpro, with the following results:

This is the first review to summarize the effect of mHealth on infection rates in pediatric patients with cancer, demonstrating a significant protective benefit. The interventions used 8 BCTs to establish effective bidirectional communication between clinicians and patients, offering practical insights for clinical implementation. Notably, the majority of interventions used WeChat, a social platform that dominates the Chinese digital landscape with over 1 billion monthly active users and preinstallation on more than 90% of smartphones across all age demographics. Its comprehensive functionality encompasses instant messaging (text, voice, and video messages), free audio and video calls, group chats (eg, interactions between health care professionals and patients or among patients), and private messaging for individualized consultation. In health care settings, WeChat is commonly used through official accounts (for health education dissemination), mini-programs (for data recording or questionnaire administration), and multimedia transmission (eg, image or video reports), offering advantages of low cost and wide reach [54]. Nevertheless, this app presents notable limitations. Interactions frequently depend on medical staff availability, potentially resulting in delayed responses [56]. In addition, current health education materials often require manual compilation and uploading, reducing efficiency and increasing clinician workload—potentially compromising care quality and safety. There is a clear need for more automated, dynamic content delivery systems.

The results of a meta-analysis indicate that mHealth-based interventions significantly improved the QoL for pediatric patients with cancer compared to usual care. These findings address previous gaps in quantitative measurement [15] and reconcile inconsistencies attributed to confounding factors in interventions and populations [14]. The included mHealth interventions incorporated 6 BCTs, supporting psychological, social, and symptom-related needs, thereby reducing risks of psychological distress, infections, and PICC complications. However, high heterogeneity and sensitivity analyses indicate instability in these results. All studies on this outcome measure used WeChat as the platform, differing only in sample size and measurement tools. This suggests that the instability and heterogeneity of these intervention outcomes may be influenced more by methodological factors than by mHealth. In addition, the observed large effect size of QoL may be partly attributable to the elevated effect size in Xie’s [35] study, which assessed outcomes at 2 months post intervention—a shorter timeframe than other studies (3 mo, 6 mo)—resulting in stronger short-term effects. Furthermore, the employment of subjective assessment scales in all studies may have resulted in an overestimation of results. Therefore, while mHealth interventions show potential for improving the QoL in pediatric patients with cancer, their long-term effects require further confirmation through more rigorous evidence-based studies before integration into routine clinical care.

This study is the first to summarize the effectiveness of mHealth-based interventions in reducing PICC-related complications within the pediatric oncology field. Results indicate that mHealth management significantly lowers overall complication rates compared to usual care, likely by mitigating recall bias and misunderstandings associated with traditional education methods (eg, demonstrations or printed manuals) [57]. The mHealth intervention in this outcome involved 8 BCTs, offering real-time monitoring and correction, scheduling and reminders, online assessment, and follow-up. Despite these advantages, the nonsignificant effect of mHealth on reducing catheter displacement may be attributed to children’s growth and frequent postural changes, which increase the risk of mechanical traction [58]. Hu et al [33] implemented a reward and threat system (weekly quizzes with prizes for successful venous catheterization) that generated a more pronounced intervention advantage in the intervention group compared to the control group. In contrast, the remaining 2 studies [43,47] used health education manuals in their control groups, potentially reducing the distinction from the experimental group. This suggests that certain functions of mHealth may not necessarily offer a significant advantage over health education manuals for specific outcomes. Overall, mHealth appears effective for reducing PICC complications, though efficacy may vary by complication type and specific intervention components.

Similar to previous reviews [15], treatment adherence was significantly better in the mHealth group than in the usual care group. The mHealth intervention in this study integrated 9 BCTs. These strategies collectively address core aspects of treatment adherence, such as timely medication administration, maintaining appropriate lifestyle habits, and attending regular follow-up appointments and assessments [59]. Additionally, dedicated online accounts facilitate peer interaction, reducing stigma and enhancing adherence [60]. However, these results exhibit some heterogeneity and extreme effect sizes. The observed heterogeneity may be partly attributable to the 3-arm randomized controlled design used by Ding et al [46]. The extreme effect sizes may be related to the smaller sample sizes. Notably, while mHealth interventions significantly improve adherence among pediatric patients with cancer, participant attrition remains a significant issue. This may stem from limited digital literacy among children and their parents, which can hinder the effective use of mobile devices and negatively impact adherence [61]. Therefore, clinical practice should establish timely feedback and reminder mechanisms, which may be key factors in maintaining long-term engagement among pediatric patients.

This review found that mHealth-based interventions did not significantly improve health knowledge scores, possibly due to a lack of age-appropriate educational content. The 4 BCT clusters used may not suffice for children’s cognitive levels [62]. Future interventions should incorporate age-tailored content, interactive designs, and parental involvement to enhance learning outcomes. In contrast, mHealth interventions demonstrated favorable effects on improving children’s self-management ability. This result pertains to 5 BCT clusters. Benefits likely stem from flexible, real-time education and feedback [63], empowering both children and caregivers [62]. The high effect sizes may reflect the use of subjective scales and potential Hawthorne effects.

However, Yao et al [41] used a professional medical platform for intervention, yielding effect sizes higher than those reported in studies using WeChat platforms [32,33] for both outcome measures. This discrepancy may explain the higher heterogeneity and wider confidence intervals observed, warranting cautious interpretation of the findings. Notably, Yao et al [41] was also the only study to conduct a cost-related analysis, reporting that the cost of hospital visits for PICC maintenance in the intervention group 117.50 CNY (US $17.07; IQR 105.00-132.50) was significantly lower than that in the control group 321 CNY (US $46.66; IQR 261.50-374.25), limiting the generalizability of the evidence. Therefore, balancing patient economic benefits with clinician workload remains a key challenge for health care administrators.

This review has several strengths. First, it provides a systematic review and meta-analysis of mHealth effectiveness in pediatric cancer, moving beyond usability to report quantitative clinical outcomes—including infection rates and PICC complications—for the first time. It also offers evidence on QoL, health knowledge, self-management, and treatment adherence, establishing a foundation for future research and practice. Second, by including only RCTs—the gold standard in clinical evidence—and restricting participants to pediatric patients undergoing treatment (excluding parents or survivors), it minimizes confounding and enhances translational potential. Third, it identifies and categorizes BCTs across studies, standardizing intervention components and improving comparability and replicability for future design. Fourth, the inclusion of Chinese databases supplements evidence from developing countries, addressing a gap in prior research.

Limitations should also be acknowledged. First, most included studies lacked the implementation of blinding, resulting in a current lack of high-certainty evidence. This may be related to the inherent difficulty of setting up blinding for mHealth interventions. Second, 79% of the included studies were conducted in China, with interventions predominantly delivered via the WeChat platform. However, the usage context of WeChat differs from that of mHealth tools commonly used in other countries, such as standalone mHealth apps or web-based platforms. The integration of daily life functions with health care services within WeChat may result in higher user engagement and retention compared with independent medical apps, potentially representing a key factor underlying the observed intervention effectiveness. Consequently, this WeChat-based model, characterized by high user engagement, may be difficult to directly replicate in other countries or regions with different health care systems and mHealth infrastructure, potentially limiting the generalizability of the findings to other linguistic and cultural contexts. Third, as a nonmedical social platform, WeChat has limitations in clinical applications. Specifically, it shows deficiencies in data encryption, regulatory compliance, liability definition, and ethical oversight during use. Simultaneously, medical staff may experience vigilance fatigue and blurred professional boundaries during specialized communications on such platforms due to unclear responsibility frameworks [64]. Moreover, it also lacks integration with clinical systems (eg, electronic health records), structured data support, and advanced algorithmic feedback [65]. Fourth, the limited number of included studies precluded subgroup analysis or meta-regression, restricting deeper exploration of heterogeneity sources, including the influence of BCT number on outcomes. Fifth, only Chinese and English articles were included due to the authors’ reading limitations, which may have resulted in the omission of high-quality articles in other languages. Finally, although the initial protocol registration planned to include a range of telehealth interventions and study designs, the scope of the review was refined during the implementation phase to focus specifically on RCTs of mHealth-based interventions. This refinement improved conceptual clarity and internal validity, while also strengthening the clinical and practical relevance of the findings. However, this focused scope may limit the applicability of the conclusions to other forms of telehealth (such as telephone-based consultations or video-enabled remote care) and to nonrandomized study designs that more closely reflect real-world practice (eg, stepped-wedge or cluster-based designs).

The findings of this study suggest that the extant evidence remains inadequate, and subsequent research may investigate the following domains.

First, future platforms could integrate ecological momentary assessment and smart biosensors to enable dynamic, predictive monitoring. Establishing automated, risk-stratified feedback mechanisms and exploring the incorporation of large language models could enhance efficiency through automated content updates and personalized information delivery. This would allow clinicians to focus more on data review and clinical decision-making [66]. Second, given the advantage of WeChat highlighted in this review, future developments should prioritize embedding mHealth modules into widely used ecosystem apps to minimize user burden, rather than developing isolated apps. Third, research quality should be enhanced. Adopting internationally recognized core outcome sets would facilitate cross-study comparisons [67]. Large-scale, multicenter RCTs should be conducted with extended follow-up periods to observe the temporal dynamics of intervention effects and long-term clinical outcomes. Fourth, equity and safety should be prioritized. Preuse assessments of digital health literacy and preferences among patients, parents, and health care providers should be conducted, offering tailored training to ensure resource equity during mHealth implementation. Platform users should be clearly defined to mitigate the adverse effects of inappropriate mobile device use on children’s vision, sleep, and other areas [68]. Finally, implementation research should be strengthened. Clear cost-benefit analyses should be conducted; barriers and facilitators to mHealth adoption should be evaluated, such as health care institutions’ feasibility regarding equipment investment, technical maintenance, staffing costs, and time allocation; and how mHealth interventions integrate with existing care processes and electronic information systems should be examined to avoid additional workload, duplicate documentation, or process conflicts, thereby enhancing technology usability and sustainability.

This systematic review and meta-analysis synthesizes evidence from RCTs on mHealth interventions for children with cancer. The findings suggest that mHealth holds significant value in pediatric oncology care, demonstrating greater effectiveness than usual care in reducing the incidence of infection and various PICC-related complications, including overall complications, site infection, phlebitis, thrombosis, bleeding, occlusion, and dislodgement. mHealth interventions also significantly improved patients’ QoL, self-management ability, and treatment adherence. However, current evidence does not show a significant advantage for mHealth in reducing PICC catheter displacement or enhancing health knowledge. Large effect sizes for some outcomes may be related to subjective measurement tools and small sample sizes. Limitations such as lack of blinding and considerable heterogeneity reduce the certainty of evidence and may lead to overestimation of effects; therefore, results should be interpreted cautiously. Despite these limitations, existing evidence supports the potential benefits and applicability of mHealth in pediatric cancer care. Future research should conduct higher-quality, multinational RCTs and use methods such as network meta-analysis to identify the most effective BCTs within mHealth platforms. Concurrently, integrating implementation science and cost-effectiveness analyses will be crucial, with a focus on aligning mHealth technologies with existing clinical workflows to improve their feasibility, scalability, and sustainable impact.