We extensively searched 6 electronic databases: PubMed, Cochrane Central Register of Controlled Trials, Embase, Web of Science, PsycINFO (via EBSCO), and CINAHL (via EBSCO). This search spanned from the databases’ inception until October 1, 2024. We subsequently updated our search until April 28, 2025. For each database, we adapted Medical Subject Headings terms and database-specific keywords, including “chronic obstructive pulmonary disease,” “internet-based intervention,” “oxygen therapy,” and “randomized controlled trial.” To ensure comprehensive coverage, we manually reviewed the reference lists of previous systematic reviews and relevant publications to identify additional studies of interest. Furthermore, 2 gray literature sources were consulted to complement the database search: the World Health Organization Institutional Repository for Information Sharing and Open Grey. In addition, we screened studies in languages other than English, if available (with English titles and abstracts), to ensure no relevant literature on the topic was missed. Details of the search strategy are provided in Table S1 in Multimedia Appendix 1.

Our inclusion criteria were (1) population: patients diagnosed with COPD; (2) intervention: the experimental groups received telehealth-supported home oxygen therapy, which involved internet-based platforms and tools. Home oxygen therapy included long-term oxygen treatment trials, home noninvasive ventilation, and home continuous positive airway pressure; (3) control: patients in the control groups received usual home oxygen therapy without telehealth support; (4) outcomes: studies needed to report at least one of the following indicators—adherence, hospital readmission, or health-related quality of life; and (5) study design: only randomized controlled trials published in English were included.

The exclusion criteria were (1) reviews, conference abstracts, case studies, protocols, or qualitative studies; (2) studies lacking complete data; (3) studies where full text was unavailable; and (4) duplicate publications.

We used EndNote version 20 (Clarivate) to remove duplicate studies, and the remaining records were screened using the web application Rayyan (Qatar Computing Research Institute) [19]. In addition, 2 authors (CH and X Liao) independently reviewed studies based on titles and abstracts, followed by a full-text review of eligible articles. Any disagreements were resolved through discussion or, if needed, consultation with a third author. Interrater agreement during the study selection process was assessed using Cohen kappa coefficient, calculated in IBM SPSS version 27.0. The level of agreement was interpreted as follows: slight (κ=0.00-0.20), fair (κ=0.21‐0.40), moderate (κ=0.41‐0.60), substantial (κ=0.61‐0.80), and almost perfect (κ=0.81‐1.00) [20].

Data extraction was also performed independently by 2 authors (CH and YF) using a standardized form. The extracted data included (1) the first author, publication year, and the country where the study was conducted; (2) participants’ age, sample size, disease stage, and baseline forced expiratory volume in 1 second; (3) the type, format, and duration of the interventions; (4) comparator interventions for the control groups; and (5) the results and outcome measures. Furthermore, data extraction focused on the effect size at the final follow-up time point reported in each study. Any discrepancies were addressed through discussion.

A descriptive synthesis was performed to summarize the characteristics of the included studies. Meta-analysis and heterogeneity testing were carried out using the meta package in R software (version 4.2.2; R Core Team). Given the variation in outcome measures for adherence, hospital admissions, and health-related quality of life across the included studies, the standardized mean difference (SMD) with corresponding 95% CIs was calculated to assess the intervention effects compared with control groups. Furthermore, to ensure consistent interpretation of outcomes, scores from the COPD Assessment Test were reverse-coded in accordance with the Cochrane Handbook (version 6.5) [21], so that higher scores across all studies consistently indicated better health-related quality of life.

Heterogeneity across studies was quantified using the I² statistic and P value, as high heterogeneity was anticipated due to differences in specific interventions, participant characteristics, and follow-up durations [22]. Consequently, we adopted a random-effects model to obtain more conservative effect estimates. Subgroup analysis and publication bias assessments were not conducted due to the limited number of included studies [23]. Sensitivity analysis was performed using a leave-one-out approach to test the robustness of pooled results. Statistical significance was set at a P value <.05 in this study.

The revised Cochrane risk of bias tool (version 2) [24] was used to evaluate the methodological quality and bias risk of the included randomized controlled trials across 5 key domains: the randomization process, deviations from intended interventions, incomplete outcome data, outcome measurement, and selective reporting of results. Each domain was assessed based on a series of signaling questions, which helped guide the assignment of bias ratings for each domain as low risk, some concerns, or high risk. The overall risk of bias for a study was determined as follows: “low risk” if all domains were rated as low risk; “some concerns” if at least 1 domain raised concerns without any being rated as high risk; and “high risk” if one or more domains were rated as high risk [24]. The assignment or intention to treat was the outcome of interest. All included studies were independently evaluated by 2 authors (CH and X Liao).

Two reviewers (CH and X Liao) independently assessed the quality of evidence for adherence, health-related quality of life, and readmission, rating each as high, moderate, low, or very low in accordance with the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) method [25]. Furthermore, 5 categories were carefully evaluated: risk of bias, imprecision, inconsistency, indirectness, and publication bias. Given that all included studies were randomized trials, the initial quality rating was set to “high” and subsequently downgraded to moderate, low, or very low if any category was judged as “serious,” “very serious,” “likely,” or “very likely [26].” We used the web application GRADEpro Guideline Development Tool (McMaster University and Evidence Prime Inc) to create the GRADE evidence profile [27].

All core review processes, including searching, screening, eligibility assessment, data extraction, and synthesis, were conducted in accordance with the registered protocol. However, the predefined outcome measures were refined during the review process. Initially, outcomes were broadly defined to include hospitalizations, exacerbations, health-related quality of life, depression, and anxiety. After study screening and data extraction, adherence, hospital readmissions, and health-related quality of life were selected as primary outcomes for meta-analysis. This refinement was based on several considerations. First, adherence, hospital readmissions, and health-related quality of life are among the most commonly reported outcomes in recent trials and reflect the growing research focus in this area. Second, these outcomes are clinically relevant in the context of telehealth-supported home oxygen therapy for COPD, as they relate to treatment compliance, disease control, and patient-centered benefits. Third, they are closely aligned with the intended functions of telemonitoring, including improving adherence, reducing preventable hospitalizations, and enhancing overall well-being.

Initially, 682 records were retrieved from 6 databases, with 2 additional eligible studies identified through manual searching. After removing 267 duplicates, 417 articles remained for title and abstract screening. Next, we conducted a thorough full-text review of 22 studies, resulting in the inclusion of 8 studies in this review. The interrater agreement between the 2 reviewers (SZ and X Lan) was almost perfect, with κ scores of 0.85 (P<.001) for title and abstract screening and 0.94 (P<.001) for full-text screening. The PRISMA flowchart is shown in Figure 1.

A total of 1275 individuals with COPD participated in the studies, with 732 receiving telehealth-supported home oxygen therapy and 543 receiving only home oxygen therapy. The mean age of participants ranged from 61 to 79 years [32,33]. Furthermore, 3 studies focused on patients with stable COPD (no recent exacerbation) [9,15,28], while 1 study included patients with recent exacerbation [29]. The baseline forced expiratory volume in 1 second (% predicted) ranged from 25 to 39 [9,32].

The telehealth interventions were organized by delivery format, including internet of things [9], applications [15,28,33], telephone calls [31,32], call centers [29], and websites [30]. Intervention durations varied from 3 to 12 months. Regarding specific types of home oxygen therapy, 3 studies examined the effects of long-term oxygen therapy [15,28,31], while the remaining studies focused on noninvasive ventilation [9,29], continuous positive airway pressure [30], and home mechanical ventilation [32].

Generally, the telehealth-supported home oxygen interventions in the included studies aimed to provide real-time remote monitoring of each patient’s clinical information and ventilator parameters. In cases of alarms or alerts, health care professionals would contact patients by telephone or other internet-based tools, offering preliminary assessments [9,15,28,29,32], promoting adherence [9,29-31,33], enhancing symptom management [9,15,28-33], addressing problems on a case-by-case basis, and arranging emergency home visits when necessary.

Participants in the control groups received home oxygen therapy without telehealth support, along with standard home oxygen management throughout the study period in all included studies [9,15,28-33].

The outcomes assessed were adherence, readmission, and health-related quality of life. Health-related quality of life was measured using 3 scales: the Severe Respiratory Insufficiency Questionnaire [9], the COPD Assessment Test [33], and the EuroQol 5 Dimensions Questionnaire [15,28]. Data on adherence and readmission were objectively obtained from medical records.

In total, 3 studies reported on the cost of telehealth-supported home oxygen therapy interventions [9,28,32]. Jiang et al [9] conducted a cost-effectiveness analysis comparing internet-of-things–based management of home noninvasive ventilation for patients with COPD to standard management. The study found a marginal increase in quality-adjusted life years for the intervention group (0.45, 95% CI 0.39‐0.52) compared with the control group (0.44, 95% CI 0.38‐0.51). Jódar-Sánchez et al [28] performed a cost-utility analysis from the health care payer perspective, which demonstrated an incremental cost-effectiveness ratio of €223,726 per quality-adjusted life year for the telehealth group compared with the control group (a currency exchange rate of €1=US $1.17 is applicable). Vitacca et al [32] also conducted a cost-effectiveness analysis, including the costs of telemedicine services, health care services, and private expenses. The study found that after accounting for the cost of tele-assistance, the average overall cost per patient was 33% lower than for usual care, indicating significant cost savings without compromising care quality. The details of the economic evaluations are presented in Table 2 [9,28,32].

According to the revised Cochrane risk-of-bias tool, 2 studies was judged to have a low risk of bias [9,33]. Furthermore, 5 studies were rated as having “some concerns” due to incomplete reporting of intervention allocation details, raising concerns about deviations from intended interventions [15,28,32]. In addition, missing outcome data and selective reporting were identified as sources of bias [29,30]. In addition, 1 study was appraised as high risk for missing adherence data due to incomplete oxygen equipment worksheets and follow-up records [31]. The summaries of risk of bias are provided in Figure 3 and Figure S4 in Multimedia Appendix 1.

According to the GRADE approach, the quality of evidence was rated as low for adherence, high for hospital readmission, and high for health-related quality of life. The downgrading factors primarily stemmed from the risk of bias in included studies, imprecision of results, and heterogeneity across studies. The evaluation details are presented in Table S2 in Multimedia Appendix 1 .

To our knowledge, this study is the first systematic review and meta-analysis of randomized controlled trials evaluating the clinical and cost-effectiveness of telehealth-supported home oxygen therapy in terms of adherence, hospital readmissions, and health-related quality of life. This systematic review of randomized controlled trials included a total of 1275 patients with COPD. Compared with control groups, telehealth-supported home oxygen therapy interventions reduced hospital readmissions and improved health-related quality of life but did not significantly increase adherence. Given that the overall methodological quality of most included studies raised some concerns or indicated a high risk of bias, and that the quality of evidence for adherence was low, more high-quality randomized controlled trials are needed to draw robust conclusions.

This review analyzed the cost-effectiveness of telehealth-supported home oxygen therapy for patients with COPD. Furthermore, 3 studies highlighted the significant economic burden that COPD imposes on health care systems [9,28,32]. Jiang et al [9] found that the average total 1-year cost per patient for stable COPD in the intervention group was ¥9524, compared with ¥7451 in the control group. However, the overall costs in the intervention group were lower, with the cost difference primarily attributed to the expenses related to health care professional contacts, telemedicine tools, and the effectiveness of internet-of-things–based management of home noninvasive ventilation. Jódar-Sánchez et al [28] adopted a health care payer perspective, meaning that cost savings would be fully absorbed by the health care system. However, this perspective overlooks out-of-pocket costs borne by patients with COPD and their families, such as caregiving expenses [45]. Therefore, future intervention studies need to conduct a more comprehensive cost analysis. Vitacca et al [32] concluded that in patients with chronic respiratory failure who are on oxygen or home mechanical ventilation, nurse-centered tele-assistance prevents hospitalizations and is cost-effective. This study considered costs from a broader societal perspective, including those borne by the health care system, patients, and other relevant parties.

Overall, all studies demonstrated that although the costs of maintaining telemedicine equipment, telephone consultations, and home visits by health care professionals were higher compared with the control group, reduced hospitalizations due to exacerbations led to lower overall health care costs [9,28,32]. This suggests that, despite higher initial costs, telehealth and home oxygen therapy interventions may result in long-term cost savings by preventing hospital admissions.

This study has several limitations. First, heterogeneity may have been introduced due to variations in study characteristics, including intervention formats, follow-up lengths, COPD severity, and outcome measures. Second, the economic analysis was limited by inconsistent cost perspectives and the lack of consideration for indirect costs such as caregiver burden and patient out-of-pocket expenses. Future research should adopt standardized frameworks that incorporate both direct and indirect costs to more accurately evaluate the financial impact of telehealth-supported home oxygen therapy. Third, the methodological quality of the included studies varied, with several studies showing some concerns or high risk of bias due to incomplete data and selective reporting. The certainty of evidence for adherence was low, which may reduce the reliability of this outcome. Future studies should employ rigorous randomized controlled trial designs, including blinding, complete outcome reporting, and standardized adherence measurement, to strengthen the quality of evidence. Finally, although this review focused on key clinical and economic outcomes, patient-centered factors such as satisfaction and psychological burden were underexplored. The lack of patient experience data limits understanding of adherence barriers. Future studies should consider qualitative or mixed methods approaches to better reflect patient perspectives and enhance real-world applicability.

Our review revealed that telehealth-supported home oxygen therapy significantly reduced hospital admissions and improved health-related quality of life, but had no significant effect on improving adherence among people living with COPD. High-quality longitudinal studies are needed to further examine whether this intervention can enhance adherence in patients with COPD. In addition to introducing relevant policies and regulations at higher levels, tailored telehealth-supported home oxygen therapy, based on demographic characteristics of patients with COPD, is warranted to improve their well-being. Furthermore, including economic analyses in future research would help policy makers make informed decisions regarding the implementation of telehealth-supported home oxygen therapy.