Impact of connected health interventions on psychological wellbeing and quality of life in patients with cancer: A systematic review and meta‐analysis

Abstract Objective Connected health technologies have the potential to improve access to cancer care and support and reduce costs. We aimed to assess the impacts of interventions delivered using connected health technologies on psychological and quality of life (QoL) outcomes in people living with and beyond cancer. Methods PUBMED, PsycINFO, Web of Science, and EMBASE were searched using terms relating to (i) cancer, (ii) connected health, and (iii) QoL/psychological wellbeing. Studies were included if they evaluated interventions using connected health technologies and assessed psychological and/or QoL outcomes for adults at any stage of cancer treatment or survivorship. Results Thirty‐seven studies met the inclusion criteria with a total of 8956 participants. Connected health technologies included web‐based applications (n = 24), smart applications (n = 12), and wearable devices (n = 1). Studies were heterogeneous in terms of intervention components. We identified five clusters: (i) Psychosocial support and rehabilitation, (ii) psychoeducation and information support, (iii) symptom monitoring, reporting and self‐management, (iv) peer and social support, and (v) health coaching and physical activity training. Due to heterogeneity of outcome measures, the meta‐analysis included only seven RCTs; pooled mean estimates showed connected health interventions were moderately effective in reducing symptoms of depression (SMD: −0.226, 95% CI −0.303/−0.149) and anxiety (SMD: −0.188, 95% CI: 0.279/−0.0963) compared with usual care. Conclusion While the considerable heterogeneity observed highlights the need for more rigorous studies to improve reproducibility and efficiency, results suggest that connected health interventions have the potential to improve psychological wellbeing and QoL outcomes in people living with and beyond cancer.


| BACKGROUND
Advances in cancer screening, diagnosis and treatment have resulted in an increasing number of people living with and beyond cancer (LWBC). 1,2 Given the range of symptoms that can be experienced, 3 there is a clear need to identify ways to enhance quality of life (QoL) in this group. 2,4 Beyond the effects that cancer can have on physical health, it may have an even greater and, arguably, more significant impact, on mental health. 5,6 Feelings of uncertainty, fear, or sadness resulting from diagnosis are associated with increased psychological distress, 7 which may interfere with coping strategies. 8 While effective management of symptoms can reduce distress, enhance coping, and improve QoL, 9 psychological wellbeing remains a top unmet need for people LWBC. 10 Psychosocial interventions such as cognitive behavioural therapy (CBT) and behaviour change techniques can enhance coping skills and improve QoL in people LWBC. 11,12 However, person-to-person interventions can be costly and hard to access, especially for individuals in hard-to-reach areas, those working or with caring responsibilities. 13 Rising cancer incidence and a shrinking healthcare workforce may exacerbate challenges, 13,14 with COVID-19 most recently presenting obstacles for in-person care. 15 Technology-based approaches to care, such as connected health (CH), may help overcome challenges by facilitating increased access to individualised support. 4,16 CH is a fast-growing paradigm in healthcare innovation where devices, interventions and services are designed around patients' needs through efficient data collection, analysis, and transfer. 17,18 CH is considered an umbrella term to reduce confusion over definitions of telehealth, telemedicine and mobile health. 17 CH entails technologies that are predictive, preemptive, personalised, patient centric and participatory. 18 It differs from other technologies in that a two-way flow of information is used. 19 The impact of CH in cancer care outcomes has been evaluated in previous reviews with mixed, albeit promising, evidence. For example, CH was found effective in enhancing patient engagement, 20 self-management, 21 and reducing barriers to care. 22 Other reviews evaluated design features of CH in cancer care, 23 its utility in cancer follow-ups, 24 effects on wellbeing and QoL outcomes, 25 and the benefits in supportive care. 26 However, these reviews reported mixed findings, acknowledging a lack of quality evidence regarding the efficacy of CH technologies. A recent review noted the need for more high-quality trials, especially those using standardized outcome measures. 27 CH technologies are acceptable among cancer patients, particularly those including elements of social support, self-management, and remote access to professionals. 28 However, while CH can enhance outcomes such as self-efficacy, coping, and perceived social support, [27][28][29] there is limited evidence for its impact on severe symptoms of psychological distress, such as anxiety or depression. 25 Additionally, the impact of CH on other psychological and QoL outcomes in people LWBC remains unclear.
Previous reviews have restricted their evaluations to specific subsets of CH, [27][28][29] noting a shortage of interventions and study heterogeneity. Thus, a quantitative analysis of CH efficacy is needed.
Given the sharp increase in CH interventions over the last decade, rapid shifts in technologies, and the increased demand for remote services in the context of the COVID19 pandemic, it is critical that their efficacy is continuously evaluated. 30 The present review aimed to assess the impact of interventions delivered using CH technologies on psychological and QoL outcomes in cancer.

This review was conducted in compliance with the Preferred
Reporting Items for systematic review and meta-analysis (PRISMA) guidelines. The protocol is registered with the Prospective Register for Systemic Reviews Database (ID: CRD42021246828).

| Search strategy
Searches were completed in May 2021 to identify articles pertaining to CH interventions for people LWBC. Any study evaluating a CHfacilitated intervention and reporting psychological wellbeing and/ or QoL, either as a primary or secondary outcome, published in a peer-reviewed journal and in the English language was deemed eligible for inclusion (see Supplementary material S1). Only technologies that were 'connected' and offered a two-way communication in the flow and use of data were included. [17][18][19] Considering the technological advancements in the last decade, only studies published in the past 10 years were considered. Bibliographic mining and citation searching of studies obtained were also conducted.
Studies were identified by searching electronic databases (PubMed, PsychINFO, Web of Science, EMBASE) using terms relating to (1) Cancer, (2) QoL/Psychosocial Wellbeing and (3) CH. (See Supplementary Material S1 for full syntaxis). Search terms were developed by IG, RM and DD based on previous literature. 31,32 Boolean operators were employed to search the selected databases.
MeSH, EMTREE, PsycINFO thesaurus or equivalent terms were used and exploded.

| Screening
Results of database searches were exported to Endnote and duplicates removed. A standardized online platform Rayyan 33 was used to screen studies. Title and abstract screening was completed by two reviewers (IG and ND) independently. The remaining articles underwent full-text reviews by independent reviewers to confirm eligibility. Disagreements were discussed amongs authors until consensus was obtained. Available data for aggregation were required for inclusion in the meta-analysis.

| Data extraction
The following data were systematically extracted by IG (checked by   ND) and inputted into an Excel spreadsheet: author, year, title,   design, number/characteristics of participants, including cancer type,   intervention type, outcome measures, results obtained, and study limitations. If required data were not reported, the corresponding author was contacted to obtain this or to seek additional details.

| Methodological quality assessment
IG and ND independently conducted a quality assessment for included studies using the Mixed Methods Appraisal Tool (MMAT). 34 The MMAT is intended to critically assess the quality of quantitative, qualitative, randomized controlled trials (RCT), non-randomized and mixed methods studies. This consists of two screening questions followed by five design-specific questions. Conflicts in quality assessments were resolved through discussion until consensus was reached. The latest MMAT guidelines discourages presenting a single number to denote quality as it does not tell what specific study areas are problematic. 34 For this reason, interpretation took the following form: 4-5 criteria met = high quality, 2-3 criteria met = moderate quality, 0-1 criterion met = low quality, as per previous analysis. 34

| Synthesis of findings
Study characteristics, interventions, and outcomes were described in table form. A preliminary analysis was employed to assess the nature of data available for meta-analysis. Considering the heterogeneity in outcomes variables and measures, thematic synthesis was deemed suitable to summarise the evidence. 35 This enabled us to aggregate evidence regarding the impact of CH on psychological wellbeing/QoL and to identify patterns within data relating to these outcomes. We synthesized findings in three stages. First, data pertaining to psychological wellbeing/QoL outcomes from CH-facilitated interventions were coded. Here, the primary reviewer developed a coding frame derived from the data, which was reviewed by the other reviewers, with discrepancies resolved through discussion. Next, similarities between codes were identified. Codes were grouped into themes that captured outcomes/patterns across included studies. Each theme was entered as a separate column in a table, while coded data from each study were entered in rows to illustrate themes. This technique

| Measures of intervention effect
Only seven studies had complete data for inclusion in the metaanalysis. Outcome measures of included studies were all continuous and reported on the Hospital Anxiety and Depression (HADS) scale, 36 therefore standardized mean difference (SMD) and standard error (SE) were used to summarize estimates of effects from individual studies. 37 The magnitude of standardized mean differences was interpreted using Cohen's conventions for small (SMD = 0.2), medium (SMD = 0.5), and large (SMD = 0.8) effects. 38 Since considerable heterogeneity was expected, we chose a random-effects pooling model for all analyses a priori.

| Assessment of heterogeneity
Inconsistency between study estimates was both visually and statistically examined through inspection of forest plots and consideration of the I 2 , respectively. The I 2 was calculated to assess heterogeneity. In general, heterogeneity was categorized as low (0%-40%), moderate (30%-60%), substantial (50%-90%), or considerable (75%-100%). 39 To examine small study effects, funnel plots and the Egger's test was used. Data available for meta-analysis was analysed using R software.

| RESULTS
Database searches yielded 1446 articles for title and abstract screening following duplicate removal. After initial screening, 90 full texts were assessed for eligibility, with 54 excluded (Figure 1). An

| Quality appraisal
Variability in methodology and study quality were noted (Table 1).
Fourteen studies met 4-5 criteria (high quality) while the remaining 23 met 2-3 criteria (moderate quality). Frequent limitations related to the randomization processes, non-blinded outcome assessors, non-representative samples, and non-adherence to interventions.

| Intervention characteristics
The duration of interventions and measurement points varied.
Twelve studies had one follow-up, while the rest had multiple follow-ups (range: 2-5). Ten studies had a waitlist control, 11 had active controls, and 15 used care as usual. Of 36 controlled studies, 24 evaluated interventions that included contact with health care providers or physical activity coaches. These included nurses (n = 8), trained coaches (n = 2), oncologists (n = 4), trained therapists and psychologists (n = 10). Others were unguided or self-guided.

| Connected health technologies
In terms of CH technologies, 21 interventions utilised web programs, including web-based self-guided psychosocial interventions (n = 8), web-delivered CBT and mindfulness sessions (n = 8), and web-based psychoeducational programs (n = 5). Thirteen interventions used smart applications with symptom monitoring, self-assessment, and self-management programs (n = 9), and those that facilitated social networking (n = 4) such as WeChat. Two studies evaluated live therapist streamed sessions via videoconferencing software, while one evaluated a wearable device. Only one-sixth (n = 6) of included CH technologies were publicly available platforms/websites.

Peer and social support
One quarter (n = 9) of included studies involved CH-mediated peer support and social networking interventions. Patients could share experiences with other patients and obtain professional support. In addition, users had access to a support forum for group discussion allowing them to ask questions and share experiences in the comfort of their homes and with confidentiality. The impact of these interventions on psychological and QoL outcomes was evaluated with overall promising efficacy. One study evaluated the effects of a 12- week social networking intervention 'healthspace.net' on distress, depression, anxiety, vigour, and fatigue in cancer survivors reporting high levels of cancer-related distress. Post-intervention, the prevalence of clinically significant depression symptoms declined from 67% to 34% in both groups. 76 A WeChat-based multimodal led to significant improvements in HRQoL among postoperative breast cancer patients. 63 Additionally, a 12-month WeChat-based education program was found effective in improving wellbeing and QoL in nonsmall lung cancer patients after undergoing surgical resection. 46

Health coaching and physical activity training
Three CH interventions targeted health coaching, skills training and physical activity with a working hypothesis that improved physical activity after diagnosis may decrease recurrences and improve QoL and physical functioning. 41,62 Survivors of colon cancer using 'Survi-vorCHESS', increased their moderate to vigorous physical activity, but this was not sustained 3 months post-intervention, with no QoL or distress differences over time. 62 The ACTIVATE Trial examined the efficacy of a wearable-based intervention to increase physical activity in breast cancer survivors. 75   This concern has been consistently noted in previous reviews. 23,78 suggesting future studies should aim to assess specific CH components and outcomes separately to maximize efficiency.

A considerable number of trials investigating CH interventions
show the promising role of CH in supporting psychological wellbeing.
To our knowledge, this is the first study to attempt a meta-analysis, albeit with a limited selection of outcomes. However, like previous reviews, 23 non-users to improve uptake of CH, which holds promise in the context of increasing challenges in face-to-face interventions.
On a positive note, the quality of the included studies was generally high. This is an additional strength indicating an increasing standard of evidence for CH interventions and an improvement from a previous systematic review 27 noting a general lack of high-quality primary studies and RCTs. However, the lack of standardised outcome measures remains a major concern.

| Study limitations
There are several limitations to the present review. Firstly, considering CH is a developing concept in digital health with a somewhat broad definition, the lack of consistent terminology may have hampered article identification for analysis. Secondly, pilot and feasibility studies were excluded and considering many of them might have deployed to offer remote services during COVID19 pandemic, it is possible that additional technologies with improved efficacy have been more recently developed. On the flip side, this reflects the urgent need for further examination of CH in line with the recent call from WHO for enhanced evaluation to inform integration and use of digital technologies. 30 Thirdly, the included studies may not comprehensively represent CH technologies and cancer subtypes as they were incidental to psychological wellbeing and QoL outcomes.
As such, any conclusions should be tentative in light of the likely partial data. Finally, only reports in English were included, thus excluding studies published in other languages.

| Clinical and policy implications
Our results suggest CH can be incorporated into clinical practice to manage psychological concerns in people LWBC. CH could be clinically useful for patients experiencing mild to moderate symptoms of depression and anxiety. At a policy level, more research and investments are required from all stakeholders, including user involvement in design to improve uptake, mass rollout, and sustainability, particularly in the aftermath of the COVID-19 pandemic.
However, potential concerns of CH use may exist, such as data privacy and security, with unintended consequences such as widening inequalities attributable to the existing digital divide as a result of low education, income, and poor connectivity, 4