This study was performed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) 2020 guidelines [25]. It examined RCTs that compared the effectiveness of mHealth-based self-management programs with non-mHealth programs for patients with COPD. The protocol for drafting the systematic reviews method was registered in the PROSPERO (International Prospective Register of Systematic Reviews; CRD42020181157).

The research questions of this study were derived according to the PICOS (population, interventions, comparators, outcomes, and study designs) framework [26]. Studies targeting adults aged 18 years and older with COPD were included in the analysis with no restriction on race, ethnicity, geography, or sex to ensure an extensive population. The intervention had to include mHealth for COPD self-management programs. The study comparators contained non-mHealth interventions, including usual, conventional, routine, or standard care, written materials, or face-to-face programs. The desired outcomes for this study were the modified Medical Research Council (mMRC) Dyspnea Scale, exercise capacity according to the 6-minute walking test (6MWT), and health-related QoL by St. George’s Respiratory Questionnaire (SGRQ) score in RCT designs. Only studies in English were included. Observational studies, reviews, qualitative research, and protocols were excluded.

A bibliographic search was conducted to identify relevant original articles using electronic database systems, including PubMed, Ovid Medline, Embase, CINAHL, Web of Science Collection, Cochrane Database, and Scopus. The search strategy was designed, and a database search was performed in October 2024 (see Checklist 1for the PRISMA [Preferred Reporting Items for Systematic Reviews and Meta-Analyses] 2020 checklist). The terms included in the search strategy used Medical Subject Headings (MeSH) terms, Boolean operators, and a filter publication type “randomized controlled trials,” with the main keywords in the query box of “chronic obstructive pulmonary disease (COPD),” “mobile health,” and “self-management.” For full search strategies, refer to Multimedia Appendix 1. The search covered all data from January 2015 to September 30, 2024. mHealth, as defined by the WHO, is a medical and public health practice supported by mobile devices, such as mobile phones, patient monitoring devices, personal digital assistants, and other wireless devices [27].

A standardized identification and screening process was applied to ensure that all eligible RCTs were included. Three authors (GNP, PL, and MAS) independently screened relevant studies based on their titles and abstracts and assessed full-text articles to ensure they met the eligibility criteria. Any unclear or missing information was sought through contacting the corresponding authors via email. The results of each screening round were compared and reviewed until a consensus was reached. The studies were categorized based on their characteristics, such as the geographic distribution, types of mHealth intervention, involvement of health care professionals, etc. The corresponding author was involved in the discussion of any discrepancy until agreement was reached.

The quality of the included studies was assessed using the Cochrane Collaboration’s Risk-of-Bias tool for RCTs (version 2.0) [28] to evaluate bias arising from the randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, or selection of the reported results. Each domain was rated as “low risk of bias,” “some concerns,” or “high risk of bias” as reviewed by the authors. Publication bias across studies was assessed using a funnel plot and an Egger test for outcomes with a minimum of 10 studies included [29].

A template was developed to extract relevant data from the original studies. The authors independently extracted the data, including the authors’ information, clinical setting, intervention, comparator, sample size, demographics, and types of study outcomes. The ECHO model was applied to categorize outcomes. The interconnection of health dimensions may lead to some overlap and ambiguity, as some tools (eg, Patient Health Questionnaire-9 and COPD Assessment Test) inform both clinical decisions and patient perceptions [30-32]. This paper classified the economic outcomes referred to as the total medical care costs of treatment options, typically evaluated in the context of clinical or humanistic results (eg, cost, quality-adjusted life year [QALY], and cost-utility analysis). Clinical outcomes in this study represented health events arising from the disease or its treatment (eg, hospital admission, emergency room [ER] visit, and exacerbation), while humanistic outcomes captured functional status or quality of life assessed through self-reported measures (eg, 36-Item Short Form Health Survey and EQ-5D [EuroQol 5-Dimensions]) [7].

Studies reporting dyspnea symptoms using the mMRC Dyspnea Scale, exercise capacity using the 6MWT, and the total score from reported QoL using the COPD-specific questionnaire of the SGRQ as primary outcomes were pooled using a random effects model. Economic (QALYs and costs), clinical (exacerbation, hospitalization, ER and clinic visits, and mortality), and other humanistic outcomes (self-efficacy) as the secondary outcomes were descriptively reported.

A meta-analysis was conducted for any of the outcomes that were reported in 3 or more RCTs [6]. Results of the expected outcomes were converted to effect sizes, and point estimates were reported, such as mean, SD, SE, and mean difference in both the intervention and control groups. Statistical significance was defined as P<.05, and sensitivity analysis was conducted to assess the robustness of the primary outcomes. Heterogeneity was identified by I² statistics, with high heterogeneity interpreted for those studies with values exceeding 50% [33]. Hedges g, a variation of Cohen d for correcting possible bias, was used as the standardized effect size [34]. Subgroups based on mHealth type, sample size, duration of intervention, geographic distribution, comparator, setting, and sex proportion were subjected to parallel meta-regression in order to identify the sources of heterogeneity in the patient primary outcomes. Subgroup cutoffs for sample size and sex proportion were defined using the median values: 106 (IQR 102; range 72.5-174.5) patients and 62.75% (IQR 20.63; range 57.15-77.76) male. Analyses were performed with R software (version 4.2.2; R Foundation for Statistical Computing) along with the packages meta (version 7.0.0) and metafor (version 4.6.0) [35-37].

The initial search of the electronic databases yielded 3136 articles. After deleting duplicates and screening the titles and abstracts, 130 full-text articles were assessed, resulting in 36 studies [38-73] deemed eligible to be included in the review (see Figure 1).

This systematic review included 36 trials. Figure 2 shows the geographic distribution of the studies, with 19 studies from Europe [38-56], 10 studies from Asia and Australia [57-66], and 7 studies from the United States and Canada [67-73]. The most significant number of studies was published in 2017 (8 studies [42-44,55,59,64,65,68]), and the least was in 2019 (1 study [41]). There were 5606 patients involved in these studies, 3774 (67%) of whom were men. Most RCTs were conducted on a relatively small number of patients, with 26 studies [39-42,47,48,50,51,53-59,61-66,68-70,72,73] recruiting fewer than 100 patients in each group. Mainly, the experiment settings were programmed as home-based services encompassing 31/36 studies (86%; see Figure S1 in Multimedia Appendix 2).

Types of mHealth interventions are calculated in Figure S1 in Multimedia Appendix 2. From 36 studies, the majority (22/36, 61%) of studies used web-based and computer-based interventions, such as websites, video conferences, social media, and electronic diaries, which were connected to a television or tablet [38,41-45,49-52,55,57-59,64,67-73]. The clinical devices used included a pulse oximeter, pedometer, spirometer, pulse wave monitor, and a biometric sensor equipped with an alert system or electronic health records. Most studies were categorized as mHealth-based self-management with educational and motivational materials, physical activities, rehabilitation programs, symptom recording, feedback, and support. A detailed summary of intervention components (eg, education, monitoring, and feedback) for each study is available in the Table S1 in Multimedia Appendix 2 [38-64,66-73]. Most studies (31/36, 86%) used control groups that received usual or conventional care [38,40-43,70,71,73], with slight differences in whether a study applied PR or not (see Figure S1 in Multimedia Appendix 2).

Health care providers played roles in delivering, assessing, and evaluating the self-management programs. Overall, 13 of 36 studies (36%) reviewed were delivered by physiotherapists [38-40,42,50,51,53,55,56,59,61,64-66], while proportions of treatment delivered by physicians [43-45,52,67-69,71,73] and nurses [41,46,54,57,58,60,62,70] were almost the same at 9/36 studies (25%; Figure S1 in Multimedia Appendix 2). Only 6 studies involved health care providers’ in-home visits or monitoring [43,45,48,52,59,65], while 9 studies included smoking cessation programs [42,43,46,47,59,62,63,72,73] and 9 studies focused on breathing techniques [42,43,49,54,57,59,61,70,73]. Most studies were categorized as PR and consisted of physical exercises and walking tests, inhaler use, health coaching, and education. A total of 29 trials provided patients the ability to enter data related to COPD symptoms, clinical signs, and amount of exercise activities by themselves or receive feedback from health care professionals or educators [39-41,43-49,51-53,55,58-61,63-73]. Detailed characteristics of the included studies are listed in Table 1.

Figure 3 presents the diversity of outcomes and instruments in studies based on the ECHO model. The majority of studies reported the clinical domain as the primary or secondary outcome, which included clinical outcomes (eg, hospital admission, clinic or ER visits, and mortality), symptom detection (eg, COPD Assessment Test, dyspnea scale, Hospital Anxiety and Depression Scale, and exacerbation), and others. All reported humanistic outcomes were derived from questionnaire responses or distinctive scales to assess various parameters (eg, self-efficacy, QoL, and adherence). Only 2 articles revealed economic outcomes. The numbers in the brackets illustrate the number of articles that used these instruments/outcomes, and detailed outcomes are reported in Tables S3-S5 in Multimedia Appendix 3 [38-64,66-73].

Figure 4 demonstrates the potential risk-of-bias evaluation. A total of 5 studies were categorized as having low risk, 10 as having some concerns, and 21 studies as having a high risk of bias. The most common finding for a high risk of bias was in the measurement of the outcome and deviations from the domain of the intended interventions. This indicated the lack of blinding patients, caregivers, the people delivering the interventions, and the assessors involved. Publication bias is evident in a funnel plot, and the Egger test of studies included in the analysis of primary outcomes showed no significant evidence, supporting the absence of publication bias (see Figure 4B-D).

This systematic review included 36 RCTs from 5 databases evaluating mHealth interventions’ impact on COPD, including clinical, humanistic, and economic outcomes. It offers a thorough perspective by integrating studies from various geographical regions, clinical settings, forms of intervention, types of control groups, the health care provider involvement, and outcome metrics. mHealth interventions demonstrated promising results in supporting self-management among patients with COPD. They enhance symptom control and improve exercise capacity, which are key targets in PR. Improvements in other clinical domains were also observed, but their economic and humanistic impacts remain comparatively limited. However, the pooled analysis for QoL did not demonstrate statistically significant effects, although some individual study results showed potential. These findings highlight clinical benefits, including better access, early symptom detection, fewer hospitalizations, more sustained exercise, and rehabilitation effects, despite variations in delivery methods, study sizes, and mHealth tools.

This review offers a key strength by providing comprehensive data on the global distribution of research on mHealth interventions and highlighting inequalities in digital infrastructure for patients with COPD. Most of the studies were from Europe, the United States, Canada, and, to a lesser extent, Asia and Australia, indicating that the findings are broadly applicable to high-income health care settings, with limited representation from Africa and other regions of Asia [74-76]. While evidence demonstrated the emerging potential of digital health interventions, the primary challenge in resource-limited developing countries is obtaining sufficient funding for their implementation and long-term sustainability [77]. Furthermore, there are inequalities in digital infrastructure, a lack of technical expertise, undeveloped regulatory frameworks, and limited implementation capacity, encompassing technology ownership, privacy, and security concerns. These are significant obstacles to adopting digital health in less-developed countries [78,79]. Efforts to address these challenges should align with the WHO’s global strategy for digital health, which optimizes data use to achieve better well-being and sustainable development goals related to health [9].

This meta-analysis revealed a majority of positive outcomes observed in the clinical domain. The results appear to have clinical relevance when compared to the established minimal clinically important differences for the mMRC Dyspnea Scale (-0.5 to -1.0 points) and 6MWT (25-33 m) [80-83]. While previous studies reported contradictory results regarding dyspnea symptoms and exercise capacity [14,16,19,20,24], the findings from this meta-analysis demonstrated clear, statistically significant, and clinically meaningful improvements in both measures. Symptom reduction was identified as a primary treatment goal in the latest GOLD report, with PR recommended as a nonpharmacological intervention to enhance exercise capacity [8]. These encouraging findings may support the provision of mHealth-facilitated PR to increase patient access, capacity, uptake, and clinical effectiveness [84-87].

This review also raised the possibility of favorable critical clinical outcomes, including decreased hospital admission rates, prolonged times to first readmission, reduced mortality rates, and lower exacerbation frequencies, as evidenced by several studies [45,49,51,52,55,58]. At the same time, one of the included studies reported a potential reduction in health care costs [45]. This observation is in line with the vision outlined in the WHO global strategy on digital health, which emphasizes the potential of digitalization to enhance the efficiency and cost-effectiveness in the health sector, while supporting innovative business models in service delivery [9]. Inocencio et al [88] and Stecher et al [89] indicate that cost reductions result from mechanisms such as remote monitoring, timely feedback, therapy optimization, improved adherence, lower hospital admission costs, and exacerbation events. However, robust evidence on the clinical and economic impacts of mHealth care use remains scarce and of low quality, underscoring the need for more rigorous and comprehensive research. While the findings appear favorable, they must be viewed in light of the study design weaknesses, particularly the high risk of bias in most included trials.

The GOLD report recognizes self-management as a strategy to improve QoL, with technological advancements offering benefits to both patients and health care professionals. Individual studies included in the analysis demonstrated that patient adherence and the content of the intervention program influenced the effectiveness of the mHealth-based self-management program. These are noteworthy factors for maintaining short-term training advantages over time [39,55,59,66]. Building on earlier research by Shaw et al [18] and Janjua et al [90], which reported uncertain outcomes regarding self-efficacy, our review presents early evidence that PRAISE scores for self-efficacy may improve, and this deserves further investigation [64]. According to the study’s pooled analysis, the SGRQ total score did not show a significant effect of mHealth treatments for self-management in patients with COPD. Likewise, the result did not meet the minimal clinically important differences for the SGRQ total score, which is 4 units [91,92]. It is generally challenging to show a substantial improvement in SGRQ scores, as earlier reviews pointed out [14,19,24,93]. When considerable effects of QoL are seen, they are typically documented in studies that used a variety of questionnaires [94] and have brief observation periods (≤6 months) [14]. As a self-reported tool, the SGRQ score is often influenced by baseline group characteristics and patient engagement with the intervention [14,43,46,67,68].

The emergence of digital health technologies in clinical practice has demonstrated impacts across the ECHO model, as reflected in numerous studies [45,75,94]. The COPD mHealth technologies clearly improved clinical outcomes, including mMRC and 6 MWT. Current mHealth evaluations in COPD lack robust economic data and show limited humanistic effects. Their multidimensional impact highlights the need for comprehensive outcome studies in the future. This reinforces the relevance of the ECHO model for real-world evaluation, as it captures benefits that extend beyond symptom reduction, including health care use, patient experience, and broader system-level value. Consistent with WHO’s Global Strategy on Digital Health, using ECHO-based approaches can strengthen decisions on policy adoption, reimbursement, and scale-up of COPD mHealth programs [9].

The study’s findings should be interpreted with caution. The sensitivity analysis revealed that the pooled SGRQ result was sensitive to the inclusion of individual studies, particularly Cerdán-De-las-Heras et al [53] and Kessler et al [49]. This indicates limited robustness of the findings. However, the initial SGRQ forest plot revealed a borderline significant effect (P=.07), with several individual studies showing promising trends in favor of the intervention. Furthermore, meta-regression revealed that comparator type significantly moderated 6MWT outcomes (P<.001), indicating its influence on observed exercise capacity effects. Most studies exhibited a high risk of bias, small sample sizes (<100), and variability in outcome measures, resulting in high heterogeneity that limited the generalizability of efficacy findings. Additionally, the majority of interventions were delivered over a relatively short duration (<6 months), which may have affected the ability to observe sustained clinical outcomes. The inability to blind patients, caregivers, and service providers as a natural aspect of digital health research [50] led to a high risk of bias in most of the studies included. In addition, the risk-of-bias assessment relies on partly subjective tools that depend heavily on the evaluator’s judgment. The instruments used across studies were also heterogeneous (eg, 36-Item Short Form Health Survey and EuroQol-5D to measure quality of life), limiting comparability of outcomes between studies. The study’s limitations also include the absence of mHealth-specific reporting standards such as the WHO-recommended mHealth Evidence Reporting and Assessment checklist [95], as well as the lack of assessment of app quality [96] and its impact on user health outcomes. In the future, this represents an opportunity to conduct studies that adhere to more specific mHealth reporting and evaluation standards. Overall, there was a lack of studies evaluating the broader impacts of mHealth on COPD, including medication adherence and cost-effectiveness [97-99].

The safety considerations in the use of mHealth self-management interventions must also be addressed. Several potential adverse effects of mHealth interventions include the misinterpretation of self-reported data, challenges related to privacy and data security, and the risk of overreliance on technology, which may delay emergency interventions [100-103]. Furthermore, excessive dependence on technology could negatively impact mental health and unintentionally strain the patient-provider relationship by reducing human interactions [104]. Therefore, while mHealth offers promising benefits, it is crucial to address these psychological and personal aspects in the design of mHealth interventions, ensuring the support, rather than replacement, of holistic and balanced health care practices. In addition, future studies are encouraged to explore long-term interventions to better understand the sustainable impact of mHealth in patients with COPD.

Findings of this review align with the GOLD 2025 recommendation, suggesting that mHealth interventions can serve as supplementary resources in clinical practice [8]. Their practical implementation necessitates comprehensive patient education, adherence to ethical guidelines, maintenance of confidentiality, and acquisition of the patient’s informed consent. Further high-quality, large-scale research is needed to develop accessible mHealth tools that offer virtual education and active feedback, use standardized outcome measures, and are tailored to diverse age groups, disease severities, and socioeconomic backgrounds. Carefully designed studies are required to comprehensively evaluate the efficacy of mHealth across economic, clinical, and humanistic domains, while also assessing its long-term safety.