This systematic review was registered with PROSPERO (International Prospective Register of Systematic Reviews; CRD42024509607). A systematic review of eHealth interventions and programs to support the psychological well-being and QoL of patients with LC was conducted between December 2023 and February 2024 following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [38]. The generation and selection of search terms was informed by preliminary scoping searches to identify commonly used terminology in the LC, eHealth, QoL, and psychological literature, and variant spellings and acronyms were taken into account. A librarian with expertise in systematic review methodology was consulted, and the draft strategies were reviewed by all authors and subject specialists. The literature search was performed using 6 databases: PubMed, PsycINFO, MEDLINE, Scopus, Web of Science, and CINAHL, using combinations of search terms designed to capture a broad range of eHealth resources, including digital and telemedicine or therapy (Table 1 and Multimedia Appendix 1). As we wanted to capture all eHealth studies that measured QoL or psychological well-being outcomes, regardless of whether the impact on well-being was a primary aim, “all text” searches were performed in all cases. Reference lists of identified studies were also examined, and a cited-by search was conducted using Google Scholar to identify further high-quality studies. No restrictions were placed on the publication date.

We included papers based on the Population, Intervention, Comparator, Outcomes, and Settings criteria outlined in Textbox 1. Studies that included participants with various cancers were eligible for inclusion, provided they broke down outcomes by cancer type and the impact on patients with LC could be determined.

Nonexperimental studies (eg, letters, reviews, case reports, guidelines, or protocols), those published outside of scientific journals, and those not published in the English language were excluded. Studies that only focus on the feasibility or technical properties of eHealth tools, or health care management using e-records or health care analytics with no reference to QoL or well-being outcomes were also excluded.

Following the removal of duplicates using review management software Rayyan.ai (Rayyan Systems, Inc), all 3 authors conducted the initial screening of titles and abstracts based on the inclusion and exclusion criteria. Given the large number of studies initially identified, each author independently screened one-third of the studies, and a random subset of 10% was reviewed by at least 1 other author. Where there was uncertainty around eligibility, the full text was reviewed by all 3 authors and discussed.

The full texts of potentially eligible studies were then retrieved and assessed by 2 independent reviewers (KJ and VH). Papers were rejected if they did not meet all inclusion criteria or if any exclusion criteria were identified. Any uncertainties or disagreements that arose were resolved through discussion. A PRISMA flow diagram illustrates the study selection process, including the number of studies identified, screened, eligible, and included in the final review (Figure 1).

Data from the included studies were extracted independently by 2 reviewers (VH and KJ) using a bespoke standardized data extraction form prepared in Excel (Microsoft Corp). The extracted data included: study characteristics (authors, year of publication, country, study design, study aim, sample size, setting, and data collection points); participant characteristics (age, gender, ethnicity, and cancer type and stage); eHealth program details (nature of the program, mode of delivery [eg, mobile app, web-based platform, telemedicine], LC specificity, degree of clinician input, number of sessions, and duration); outcomes (measures used); acceptability and engagement (patient satisfaction, fidelity, and attrition rates) and key findings. Data extraction was shared evenly, with each reviewer extracting data from 50% of the included studies. To ensure reliability, a random sample (10%) from each reviewer was extracted in duplicate. No conflicts occurred.

The methodological quality of the included studies was independently assessed by 2 reviewers (VH and KJ) using a modified version of the Downs and Black checklist [39]. This method was chosen as it can be used to evaluate both randomized and nonrandomized studies of health care interventions, and it has been ranked in the top 6 quality assessment tools suitable for use in systematic reviews [40]. It has good criterion validity (r=0.90) and good interrater reliability (r=0.75). The checklist comprises 27 items that address several methodological issues, including reporting, external validity, internal validity (bias and confounding), and power. In line with previous studies, we used a modified version of the scale that simplifies the scoring associated with statistical power. While this question carried 5 points in the original scale, more recent studies have awarded this a single point to signify whether a study had sufficient power to detect a clinically significant effect (1) or not (0) [41,42]. Thus, 26 items on the scale are rated as either yes (1), no (0), or unable to determine (0), and 1 (“Are the distributions of principal confounders in each group of subjects to be compared clearly described?”) is rated on a 3-point scale (yes=2, partially=1, and no=0). Item scores are summed to produce a total score ranging from 0-28, with higher values representing better study quality. Using these scores, studies were rated as having poor (≤14), fair (15-19), good (20-25), or excellent (≥26) methodological quality.

A narrative synthesis of the findings was conducted in line with the “synthesis without meta-analysis” guidelines [43]. Due to the heterogeneity in study designs, eHealth programs, and outcome measures, a meta-analysis was not performed. Instead, following recommendations in the Cochrane Handbook for Systematic Reviews of Interventions, vote counting based on direction of effect was used for the quantitative synthesis of results, as synthesizing P values was not possible with the available data [44,45]. For the purpose of quantitative synthesis, outcome domains of interest were defined as: QoL (including total QoL scale scores, global QoL measures, and functioning-related QoL subscales), anxiety, depression, and distress. Note that specific physical symptom burden items or subscales (eg, pain, fatigue, or nausea) were not included as independent QoL measures in the effect direction synthesis, as these are often more influenced by disease progression or treatment toxicity than by behavioral or psychosocial interventions. For each outcome domain, we compared the number of studies showing a beneficial effect of the intervention (for pre-post study designs) or a relatively beneficial effect of intervention group (IG) membership (for randomized controlled trials [RCTs] or other studies looking at between group differences in scores over time), with those showing a negative effect. In line with the guidance, we did not take statistical significance or magnitude of effect into account within this quantitative synthesis [44,45]. If a study assessed the same outcome (eg, QoL) using multiple measures (eg, across several subscales), we determined the overall direction of effect based on consistency across those measures. Specifically, when 66% or more showed the same direction of effect (eg, all indicating benefit or all indicating harm), we classified this as the overall effect direction. If fewer than 66% of the measures yielded outcome effects in the same direction, we labeled the effect as mixed. Across all studies, a sign test was used to determine whether the observed pattern of results (ie, the proportion of positive versus negative outcomes) was significantly different from chance.

Results were also grouped thematically according to the type of eHealth intervention under investigation. Unfortunately, statistically investigating heterogeneity through subgroup analyses based on intervention type was not possible due to a lack of sufficient data. Likewise, it was deemed inappropriate to perform separate sign tests for each intervention type due to the lack of power that the relatively small number of studies in each group affords. As such, the impact of the different types of eHealth interventions on patient outcomes, adherence, and QoL was explored more narratively. Greater emphasis was placed on studies that were rated as having higher quality according to the criteria above, and concerns regarding the quality of evidence were reported where necessary.

After excluding duplicates, 7065 records were screened against the inclusion and exclusion criteria. This resulted in 7022 exclusions. A total of 43 full-text reports were reviewed in greater depth for additional information, of which 31 were deemed eligible for inclusion. Two additional studies were identified as being eligible from the reference lists of the included reports, resulting in 33 studies in the final review (Figure 1).

Table 2 provides an overview of the studies included in this review, including information about the design of each study and the nature of the eHealth programs under investigation. Recruitment methods were not always reported, but where described, typically occurred in clinical settings as part of routine diagnostic or treatment appointments. Countries that contributed multiple studies to the review were the United States (n=12, 36%), China (n=9, 27%), South Korea (n=3, 9%), and the Netherlands (n=2, 6%). Sample sizes ranged from 16 to 515 participants. The included literature was published between 2014 and 2024, with the majority (n=19, 58%) being published in 2020 or later.

All but 1 study [61] included solely patients with LC, and they ranged in terms of the stage of disease that they included: 5 (15%) included only early stage [49,54,68,73,77], 8 (24%) included only advanced disease [51,52,58,64-66,69,72], 12 (36%) included all stages [46,48,55-57,59,60,62,63,67,74,76], and 8 (24%) studies failed to report disease stage [47,50,53,61,70,71,75,78]. Studies also varied in terms of their methodology: 17 (52%) were described as pilot or feasibility studies or were single-arm pre-post studies; 13 (39%) were full-scale RCTs; 2 (6%) used retrospective case-control methods; and 1 (3%) used a nonrandomized, quasi-experimental design (Table 2). Additionally, 28 (85%) studies focused solely on patient participants, while 5 (15%) studies also included caregivers or loved ones (although only patient-related outcomes are included in this review).

A quality rating was assigned to all 33 studies and is provided in Table 2. Overall, in terms of methodological quality, none of the studies were rated as excellent, 12 (36%) studies were rated as good [48-52,54,62-64,67,72,73], 12 (36%) studies as fair [46,55-59,65,66,68,74-76], and 9 (27%) studies as poor [47,53,60,61,69-71,77,78].

Most studies (n=26; 79%) focused on QoL, specific psychological symptoms (including depressive and anxiety symptoms), or psychological distress as primary outcomes [46-48,50,54-56,58-60,62-75,77,78]. These measures were secondary outcomes in the remaining 7 (21.21%) studies [49,51-53,57,61,76].

As illustrated in Table 2, QoL was measured in 25 (76%) studies. Seven different validated QoL measures were used: 11 studies used the EORTC QLQ-C30 (European Organisation for the Research and Treatment of Cancer Quality of Life Questionnaire), 8 used the Functional Assessment of Cancer Therapy Scale, 2 used the Short Form Health Survey, 1 used both the EQ-5D and the EQ-VAS, 1 used the Functional Living Index-Cancer, 1 used St. George Respiratory Questionnaire, and 1 used the single item Quality of Life scale.

In terms of specific psychological symptoms, depression was measured in 12 (36%) studies, and anxiety in 13 (39%) studies. The most frequently used measure was the Hospital Anxiety and Depression Scales (n=4) [79]. Four further scales were used to assess depression: 3 used the Patient Health questionnaire [80], 2 used Zung Self-Rating Depression Scale [81], 1 used the Patient-Reported Outcomes Measurement Information System Depression items [82], 1 used the depression item of the Edmonton Symptom Assessment System (ESAS) [83], and 1 used the Center for Epidemiological Studies Depression [84]. A further 4 validated scales measured anxiety: 3 studies used the Generalized Anxiety Disorder Questionnaire [85], 2 used the Zung Self-Rating Anxiety Scale [86], 1 used the Patient-Reported Outcomes Measurement Information System Anxiety items [82], and 1 used the State-Trait Anxiety Inventory [87]. Additionally, 2 studies used the anxiety item from the ESAS [83]; and 1 study used nonvalidated daily measures of anxiety and mood.

Psychological distress was measured in 10 (30%) studies. A total of 5 studies used the distress thermometer [88] (1 in combination with the Impact of Event Scale to assess cancer-related stress) [89], 2 studies used the Symptom Distress Scale [90], 1 used the ESAS well-being subscale [83], and 1 used the Psychosomatic Status Scale for Cancer Patients [91]. An additional study used the MD Anderson Symptom Inventory [92] to infer affective interference.

Reflecting the heterogeneous nature of eHealth programs, the reviewed studies used interventions that varied significantly from 1 another in both scope and technological basis. Of the included studies, 14 aimed to improve health literacy via education about LC symptoms, treatment-related side effects, or symptom management [46,47,54,55,57,67,69,71-75,77,78]; 13 provided psychological or well-being support [62-66,68-74,76]; 13 were used for symptom monitoring [48-53,67,68,72,75-78]; 12 aimed to improve physical ability, often as a form of pulmonary rehabilitation [54-61,73,74,76,77]; and 3 provided access to medical records or laboratory results [74,75,77]. Categories were not mutually exclusive, with many studies using eHealth to deliver multiple avenues of support simultaneously. To better understand how different intervention characteristics related to patient outcomes, results for each study have been grouped according to the interventions’ main functions (Tables 2 and 3 and Figure 2).

Across all 33 studies, acceptability of the interventions was quantitatively explored in 42.42% (n=14) of cases. Five studies [49,54,55,77,78] reported that 70%-100% of patients found the interventions to be valuable or helpful, while 4 studies [47,48,54,74] reported that 88%-100% would recommend the intervention. Five studies [58,71,74,76,77] reported 77%-100% of their participants were satisfied with the intervention, while 2 studies [46,61] measured intervention acceptability on self-reported satisfaction scales from 0-4 and found means to be relatively high (all scores were over 3.2). Furthermore, 1 study [66] found that a higher proportion of their participants preferred receiving their intervention online than in person.

Nine studies assessed participants’ engagement with the interventions through metrics such as intervention (or session) completion rates, log-ins, and usage tracking. Engagement levels varied widely. Coats found that patients attended 100% of their activity sessions when they were supervised, and 96% when they were not [58]. In contrast, Bade et al [55] found only 21% of patients met their personalized physical activity goals, although step counts were recorded in 90% of weeks. Granger et al [57] found that only 50% of patients reliably wore their activity trackers, while 62% completed exercise diaries. PRO completion rates varied widely across 3 studies from 40%-82% [48,51,78]. Engagement in psychological well-being interventions seemed generally high, with 2 studies all reporting attendance rates of over 70% [64,66]. For 1 of the multicomponent portal interventions, engagement was mixed. Although 92% of patients logged in more than once and 82% of PROs were completed, the mean number of program log-ins during the 4-month study period was relatively low (<12), as was the mean usage time (<13 minutes) [77].

Unlike traditional models of care, which are temporally and geographically constrained, eHealth interventions offer flexible, open, and interactive avenues of support. Indeed, a particular strength of eHealth is its ability to reach people in rural or underserved areas, enabling timely, remote support by overcoming logistical and health care access-related barriers related to location, staffing shortages, limited access to specialist care, etc [113]. Findings from both this review and the wider literature support the potential benefits of eHealth in cancer care. For instance, several reviews have found eHealth programs to be superior to standard care in terms of improving QoL and psychological outcomes in patients with a range of different cancers [94,100,101,114], with the caveat that the impact on psychological outcomes is less clear.

Notably, intervention modality and design appear to influence outcomes. As also reported by Li et al [94], app- and web-based interventions tend to be more effective than those delivered by telephone. Similarly, our review found that multimodal, platform-based, or nurse-led interventions were consistently more successful than single-focus interventions, aligning with findings from breast cancer populations [102,115]. Given the complex and multifactorial nature of QoL, which can be influenced by disease type, stage, treatment burden, and individual needs [116], these interventions may be particularly successful because they support patients in multiple ways and use personalized tools and solutions to address an array of QoL and well-being needs.

These findings have important real-world care implications, as a recent review found that unmet physical and psychological care needs significantly impact the QoL and well-being of people with LC, highlighting a need for targeted services and interventions to address these needs throughout the disease trajectory [117]. Furthermore, using eHealth strategies to deliver support aligns with the NHS (National Health Service) 10 Year Health Plan [118]—a strategy for England’s NHS aimed at sustainably improving patient care by harnessing the use of digital technology in health care delivery, focusing on illness prevention, the expansion of mental health and well-being support, and emphasizing the importance of community care (ie, shifting care closer to people’s homes rather than in hospitals where possible). However, despite the promise of these tools, real-world implementation remains limited. Most eHealth interventions are not yet integrated into routine care, and when they are, they are often not used as intended, reducing their potential benefit [119,120].

Barriers to implementation at the institutional level and challenges to uptake among both health care professionals and patients require further investigation and targeted solutions. One key consideration is the feasibility of delivering eHealth interventions, particularly those that involve ongoing clinician involvement. Several interventions reviewed in the current paper included scheduled, individualized contact over extended periods (eg, [73]), which raises concerns about cost and scalability and may hinder wider implementation. Thus, cost analyses need to be carried out to assess the feasibility of their delivery. From the user perspective, although acceptability within this review was generally reported as high and many participants found these interventions helpful, this was assessed in only a minority of studies and warrants deeper exploration. While prior research has shown high satisfaction with digital health tools in chronic illness populations [121], actual engagement with such programs can be inconsistent. This may be influenced by factors such as lack of trust in digital systems, poor digital literacy, and minimal tailoring to age, gender, or disease stage [112,114]. To address these issues, future research should prioritize strategies to enhance engagement, such as patient and public involvement. Only 4 studies in the current review discussed how their interventions were developed (ranging from top-down development by experts to patient and public involvement co-design), so there is limited evidence as to how these strategies may have impacted acceptability and engagement. However, previous work suggests that taking an expert-by-experience approach can hold significant potential to improve both uptake and sustained use of interventions [122].

There are several strengths associated with this review, including a rigorous and systematic search and selection process that ensured comprehensive evidence coverage of the literature and the application of a validated quality assessment tool. Additionally, previous work has emphasized the need for research to investigate the components and features of eHealth programs that lead to improved therapeutic outcomes [115]. A further strength of this review is that it aimed to achieve this by characterizing the nature of the eHealth interventions and strategies used with patients with LC and elucidating the approaches that are likely to positively impact QoL and psychological well-being. However, we acknowledge that when combination approaches are taken, it is difficult to ascertain the specific aspects of the interventions that are driving the outcomes, particularly as the included multicomponent studies did not specify which aspects of their interventions resulted in change.

Several limitations should also be acknowledged when interpreting our results. First, our search was restricted to studies published in English and did not include gray literature, potentially omitting relevant research. Second, we found significant heterogeneity across the included studies in our review, due to variations in study designs, outcome measures used, intervention types, intervention lengths, follow-up periods, and control conditions (when included). The absence of homologous methods and approaches across studies prevented a meaningful meta-analysis. As such, this review relied on a predominantly narrative synthesis based on vote counts of effect directions. While this approach can indicate whether an effect exists, it cannot estimate the combined significance of the studies or the magnitude of the effects. These limitations may reduce the strength of the conclusions that can be drawn, but also highlight the need for more standardized reporting in future primary studies.

Third, many digital tools used in health care settings (such as patient portals, electronic records, and well-being tools) are developed for broader populations and not tailored specifically to those with cancer. As the current review focused only on interventions designed for people living with or beyond cancer, we may have missed a wider range of potentially relevant tools. As such, the total number and diversity of eHealth options for patients with LC may be larger than currently represented in the review.

Finally, review methods aside, the limitations of the included literature must also be noted. The quality of studies included in the review was variable, with no studies rated as being of excellent quality. Additionally, many of the studies were pilot, feasibility, or single-arm studies with associated risks of bias, and many had small sample sizes (of n<50), limiting their generalizability. Additional large-scale RCTs using standardized QoL and well-being measures are needed to strengthen the evidence for the efficacy of these interventions. Furthermore, due to the relatively short follow-up periods in this review, evidence for the long-term effects of eHealth interventions was limited.

This review presents the first comprehensive synthesis of evidence on the role of eHealth in supporting QoL and psychological well-being among people living with LC. The findings suggest that multicomponent and nurse-led digital interventions, particularly those that are personalized, involve clinician input, and are delivered via apps or web-based platforms, consistently demonstrate positive effects across both QoL and psychological domains. Additionally, longer-duration interventions (12 weeks or more) were particularly helpful for QoL. Physical activity programs were also frequently effective for both QoL and psychological well-being. However, while symptom monitoring programs had a positive impact on QoL, their effects on well-being were less clear, potentially due to small sample sizes or implementation issues. Overall, eHealth interventions show strong promise as acceptable, scalable, and flexible tools for addressing the complex and varied supportive care needs of people living with LC; yet, their integration into routine care remains limited, and few are tailored specifically to this population.

While the findings of this review are encouraging, variability in study quality, design, and outcome measures, along with limited real-world implementation, underscores the need for more rigorous, standardized research. To fully realize the potential of eHealth in this context, future efforts must focus on the development of well-designed, targeted digital solutions, with clear strategies for implementation, sustained patient engagement, and meaningful involvement of people with lived experience. Doing so could help close existing gaps in care, improve QoL and well-being, and contribute to the broader goals of delivering accessible, personalized cancer care.