The review followed a predetermined protocol, developed in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines [31,32] and the Joanna Briggs Institute Methodology for Scoping Reviews [33]. A scoping review method was used to examine the extent, range, and nature of work on this topic; identify gaps; and provide suggestions to improve future directions for research and practice on e-mental health [34]. Quality assessments were not undertaken, as the aim was to examine the full breadth of the literature, consistent with the general aims and methodology of scoping reviews [35].

Four academic databases (ie, MEDLINE, Embase, PsycInfo, and CINAHL) were searched from December 31, 2019, to March 31, 2021. The search strategy consisted of terms pertaining to eHealth (eg, “eHealth” and “telehealth”), more specific terms related to e-mental health (eg, “m-mental health”), common mental disorders (eg, “anxiety” and “depression”), and COVID-19. The search strategy was adapted for each database as necessary (see Multimedia Appendix 1 for the complete search strategy, using Ovid MEDLINE as an example). The search strategy was developed in consultation with an academic research librarian and was reviewed by all authors prior to execution.

Articles were included if they were (1) in the English language, (2) peer-reviewed journal articles, (3) discussed the application of e-mental health (theoretical or applied), and (4) were published following the onset of COVID-19 (from December 31, 2019). No restrictions were placed on the target population or setting. Studies were excluded if they did not focus on e-mental health, they were book chapters or conference proceedings, or the full text was not in English.

Reference details, including abstracts, were downloaded into the reference management software EndNote X9 (Clarivate), exported to Microsoft Excel, and divided among the research team for title and abstract screening. Two reviewers (IM and LAE) independently reviewed 5% of the titles and abstracts; interrater reliability between the two reviewers was high (κ=0.95, 95% CI 0.77-1.00; P<.001). Full-text screening was conducted by four independent reviewers (LAE, IM, KC, and Tamasha Jayawardena). A total of 5% of full texts were reviewed by four reviewers, with the interrater reliability assessed to be sufficiently high (κ≥0.80) [36].

A customized data extraction workbook was developed in Microsoft Excel. The workbook was piloted by each of the four reviewers with a subset of papers (n=10). Issues in consistency of data entry and usability of the template were then discussed, and modifications were made accordingly. Key information extracted included the following: article characteristics (ie, authors, date of publication, country of residence of the corresponding author, and journal name); article keywords, as supplied by the authors of the paper; and article type (ie, empirical, nonempirical, protocol, or review). For empirical studies (ie, studies that present an analysis of primary or secondary data in their results), information on study design (ie, quantitative, qualitative, or mixed methods) and methods (eg, surveys and interviews) were also extracted. In addition, article titles, abstracts, and keywords were searched for terms to identify papers with specific subpopulations of interest (eg, children, adolescents, veterans, and the elderly) and specific mental disorders (eg, depression, anxiety, posttraumatic stress disorder [PTSD], and substance abuse), as well as articles with a focus on issues or challenges related to the use or uptake of e-mental health in the context of COVID-19 (eg, barriers, privacy, and ethics).

Altmetric data were selected as the most appropriate measure of an article’s impact or influence because of the recency of this literature [37]; while citations have a lag effect, the real-time update of social media metrics makes Altmetric events around research outputs visible within a short amount of time (eg, hours or days). The Altmetric Attention Score, a composite score of social interest that includes mentions in newsfeeds, Twitter, Facebook, and Google, among other sources, was collected for each included study, where available, from Altmetric Explorer in June 2021. In addition, we recorded the CiteScore for the journal in which each article was published, taken from the Scopus database in June 2021. CiteScore metrics were used rather than the Journal Impact Factor, as CiteScore values give “a more comprehensive, transparent, and current view of a journal’s impact” (page 941 of Roldan-Valadez et al) [38], and they are provided for many more journals than the Journal Impact Factor. Although previous research indicates that an article’s Altmetric score within its first year of publication reliably predicts its future citation count, we followed the recommended optimal approach of using both Altmetric scores and CiteScore values as an indicator of article impact [39].

Articles were grouped together based on common attributes in the data (eg, article type). The country of the corresponding author was coded by income classification based on the World Bank’s definitions of gross national income per capita per year. The three categories were low (<US $1045), middle (US $1046-$12,695), and high income (>US $12,696) [40].

Key topic areas were identified through an analysis of article keywords. These were extracted by the research team and were then cleaned and checked for consistency. Derivative terms (eg, “health care” and “healthcare”) were amalgamated. Each keyword was reviewed and inductively classified by two authors (LAE and IM) into key topic and subtopic areas using the structure and definitions outlined in Multimedia Appendix 2. The keyword data were analyzed for frequency and co-occurrence and were graphically presented using Gephi (version 0.9.2) [41]. Our presentation of keyword data and discussion of key topic areas focused on the most influential papers as representatives of salient topic areas. To select these papers, we used a combination of Altmetric and CiteScore data, including publications with journal CiteScore values of ≥5, representing the top one-third of journals in our included articles, and/or Altmetric Attention Score values in the top 25% of all publications scored by Altmetric.

The search retrieved a total of 2418 publications. After removing duplicates, 1506 remained for title and/or abstract review. Following title and/or abstract screening, 1023 publications were discarded as they did not meet the inclusion criteria. Based on the full-text assessment, a further 127 publications did not meet the inclusion criteria, resulting in 356 publications included in this review (see Multimedia Appendix 3 for included articles). Figure 3 demonstrates the inclusion and exclusion of papers at each stage of the screening process.

A summary of the key characteristics of the included articles is presented in Table 1. Of the 356 articles, most were nonempirical commentaries or opinions (n=225, 63.2%), around one-quarter were empirical publications (n=99, 27.8%), 20 were literature reviews (5.6%), and 12 were protocols (3.4%). Figure 4 shows the trends in article types published during the pandemic, with nonempirical publications peaking in June 2020 and empirical publications emerging later and becoming more frequent as the pandemic progressed.

Of the 225 nonempirical publications, most were commentaries or opinion pieces (n=106, 47.1%), with a further 49 (21.8%) classified as descriptive case studies, 36 (16.0%) as letters to the editor, 22 (9.8%) as unstructured reviews, 10 (4.4%) as editorials, and 2 (0.9%) as unstructured protocols (Table 1). Of the 99 empirical publications, the majority used quantitative methods (n=68, 69%), 21 (21%) used a mixed methods design, and 10 (10%) employed qualitative methods alone. Most of the quantitative and mixed methods studies (n=89, 39.6%) were cross-sectional (n=60, 67%), with only 4 (4%) randomized controlled trials [42-45] and 3 (3%) nonrandomized controlled trials identified [46-48].

Overall, the empirical studies focused on the uptake and effectiveness of e-mental health initiatives in providing mental health care during COVID-19. The nonempirical descriptive case studies outlined the processes and challenges of the rapid conversion to telehealth for a particular mental health service, while nonempirical commentary or opinion pieces more broadly outlined policy changes, issues, and challenges to e-mental health and offered guidance to clinicians (see Table 2 [49-52] for exemplar papers for each category).

Corresponding authors were predominantly from high-income countries, with almost half of the 356 included outputs coming from the United States (n=160, 44.9%), followed by Australia (n=28, 7.9%), the United Kingdom (n=20, 5.6%), and Canada (n=17, 4.8%). Notably, our study identified a number of publications from middle-income countries (n=59, 16.6%), with most of these being from India (n=29, 7.6%) and China (n=9, 2.5%) (Figure 5).

The 356 included articles were spread across 165 different journals, which were primarily focused on mental health, but varied in scope. Most of the journals had a reported CiteScore in Scopus (n=138, 83.6%), with scores ranging from 0.3 (Psychiatric Times) to 25.2 (The Lancet Psychiatry). A total of 3 articles (1.8%) were published in The Lancet Psychiatry [53-55], of which 2 were from middle-income countries [53,54]. The largest number of included articles were published in JMIR Mental Health (n=19, 11.5%; CiteScore=1.3), followed by the Asian Journal of Psychiatry (n=14, 8.4%; CiteScore=4.7) and the Indian Journal of Psychological Medicine (n=13, 7.9%; CiteScore=1.8), with one-quarter of articles (n=40, 24.2%) being published across the JMIR suite of journals (mean CiteScore 3.37). The mean CiteScore value across articles (ie, CiteScore for each article based on its journal CiteScore and then averaged across articles) was 4.22 (SD 3.53); however, CiteScore values varied by article type, with protocol papers having a lower average CiteScore (mean 2.1, SD 1.55) than empirical studies (mean 3.8, SD 2.06), nonempirical articles (mean 4.6, SD 4.16), and reviews (mean 4.4, SD 1.91).

Of the 242 publications with reported Altmetric data, almost half were ranked in the top 25% of all outputs scored by Altmetric (n=118, 48.8%). Altmetric Attention Score values ranged from 1 to 282 (mean 13.33, SD 30.04); though similar to CiteScore values, there was variation in Altmetric metrics by article type, with empirical studies (mean 14.10, SD 29.22), nonempirical articles (mean 13.6, SD 32.77), and reviews (mean 11.9, SD 14.11) having higher Altmetric scores than protocol papers (mean 5.4, 4.39). The publication with the highest Altmetric score was a nonempirical contribution by Kozloff et al, which outlined the adverse mental health consequences and virtual mental health service delivery options for people with schizophrenia during COVID-19, with an Altmetric Attention Score of 282, placing it in the 98th percentile of outputs of the same age [56]. The top five articles with the highest Altmetric Attention Score values are shown in Table 3 [24,42,55-57].

The topmost influential publications—defined as those with Altmetric Attention Score values in the top 25% of all research scored by Altmetric, and/or CiteScore values of ≥5—were identified. Based on this definition, close to half of our 356 included publications were classified as influential (n=165, 46.3%). The 165 influential articles were broadly reflective of the countries and country incomes previously reported (high-income countries defined here: n=143, 86.7%).

Among the 165 most influential publications, 124 had author-contributed keywords, with a total of 273 unique keywords in this subset of papers. The frequency with which these keywords were used together on a publication is visually depicted in the network of co-occurring keywords (Figure 6). In this network, each circle (node) represents a keyword, and each line (edge) indicates co-occurrence on a publication. The most common keywords (highest in-degree score), as reflected in size in the figure, were as follows: COVID-19 (n=90), mental health (n=37), telehealth (n=32), telemedicine (n=30), and telepsychiatry (n=17). The sociogram includes 124 papers (grey nodes), 50 keywords (colored nodes), and 390 edges. The size of each node is filtered by the in-degree score, where bigger nodes would have larger in-degree scores (ie, the number of edges directed to the node). Our inductive analysis of keywords identified a number of key topic and subtopic areas: (1) COVID-19 pandemic; (2) mental health, including specific mental disorders; (3) e-mental health, including specific eHealth modalities and issues or challenges; (4) specific populations of interest; and (5) study field (eg, cancer and pediatrics). These are colorized separately in the figure to indicate their relation to one another.

Although many of the 356 articles focused on mental health broadly, describing concerns relating to the effect of the pandemic on stress (n=78, 21.9%), loneliness or isolation (n=34, 9.6%), and general well-being (n=23, 6.5%), other studies focused their attention on specific mental disorders. A total of 71 of the 356 included articles (19.9%) were concerned with anxiety disorders (see Table 3 for examples), including the use of e-mental health to treat PTSD during the pandemic. A total of 48 articles were concerned with depression (13.5%). Other disorders of specific focus identified were addiction and substance abuse disorders (n=17, 4.8%), psychotic disorders (n=16, 4.5%), suicide (n=13, 3.7%), and eating disorders (n=11, 3.1%).

A large proportion of the 356 included articles described the application of e-mental health via telephone and/or videoconferencing during COVID-19 (n=142, 39.9%), as well as the use of smartphone apps (n=43, 12.1%) and support groups (n=6, 1.7%) in assisting people during times of need. For example, the review by Strudwick et al identified 31 smartphone apps and 114 web-based resources, including telephone support, virtual peer support groups, and discussion forums, that could be used to support the mental health of the Canadian public [58]. A smaller number of studies empirically tested the efficacy of specific e-mental health modalities during COVID-19. For example, a trial from Canada tested the efficacy of delivering videoconferencing psychotherapy for people with panic disorder and agoraphobia, demonstrating that cognitive behavioral therapy delivered via videoconferencing is no less effective than face-to-face delivery on all outcome measures, and provides important information to guide the delivery of e-mental health services during and after the COVID-19 crisis [47]. Less mentioned were virtual reality, wearables, and artificial intelligence, though these modalities were noted by Vadlamani et al as being the “future of telepsychiatry” [27].

A number of articles identified particular at-risk populations who would require greater care and/or more resources to overcome barriers to access e-mental health during the pandemic. A total of 90 articles out of 356 were focused on children and/or adolescents (25.3%) (see Table 3 for examples), and 39 articles focused on older adults or the elderly (5.6%) and veterans (4.8%) as at-risk populations.

There appears to be a general consensus in the literature that younger children and older adults, who may be less familiar with technology, may find it more difficult to access the benefits of e-mental health [59]. However, the flexible and remote nature of e-mental health may offer a more viable option to certain populations who may have historically fallen through the cracks or struggled to engage with traditional psychology service provision, for example, marginalized populations, such as those identifying as LGBTQ+ (lesbian, gay, bisexual, transgender, queer, questioning) [60], and young adults who prefer to seek help anonymously and at a time that suits them via smartphone apps [61].

A further 15 articles (4.2%) were specifically concerned with the mental health impact of COVID-19 on health professionals themselves. For example, Cheng et al outlined a peer support project offering support from mental health professionals overseas to frontline health care workers in Wuhan, China, via a popular smartphone app [62].

Although it has been only 18 months since the start of the pandemic, this review identified that the total number of articles on e-mental health quickly thrived, with nonempirical articles peaking in June 2020 and empirical articles emerging later and becoming more common as the pandemic progressed. The location of articles was predominantly in high-income countries (ie, the United States, Australia, the United Kingdom, and Canada), reflecting general trends previously identified in the e-mental health literature [78,79]. However, a higher number of articles from medium-income countries were identified in this review in comparison to previous e-mental health reviews [78], with most of these articles coming from India and China. Increased outputs from China may be attributed to having the earliest outbreak of COVID-19, and for other middle-income countries, such as India, this could be reflective of the particular concerns and urgent solutions needed for people living in these countries [80].

Our bibliometric analysis indicated that the e-mental health literature during COVID-19 has had above-average influence, with almost half of the articles being ranked in the top 25% of all output scores by Altmetric and having an average CiteScore across articles of 4.22. Nonempirical articles (eg, commentaries or opinion pieces) received a notable amount of attention; articles of this type surged early in the pandemic, so elevated influence may reflect researchers and clinicians searching for new information on e-mental health, guidelines, evidence-based practice, and other relevant developments at an uncertain time and when empirical studies were limited.

The network analysis of author-supplied keywords identified key topic areas, which were explored further via our full-text analysis. Many articles focused broadly on mental health and well-being, with specific attention being paid to anxiety disorders, depression, substance abuse disorders, and psychosis. Children and adolescents were the most frequently identified in articles focused on specific populations, with other identified at-risk populations, including the elderly, veterans, and health professionals themselves. It was noted that younger children and older adults, who may be less familiar with technology, may find it more difficult to access the benefits of e-mental health. Frontline workers directly involved in the care of patients with COVID-19 have been identified as being at particularly high risk for mental health issues as a result of excessive workloads and work hours, insufficient protective equipment, feeling inadequately supported, as well as the high infection rate among medical staff [81]. Studies from past pandemics (eg, SARS) suggest that those in emergency departments, intensive care units, and infectious disease wards are at highest risk of developing adverse mental health outcomes [81].

Modalities of e-mental health largely focused on the application of “telehealth” via telephone or videoconferencing. However, there were also a number of articles reporting on the adoption of smartphone apps [82-85] and virtual support groups [62,86-88]. Self-directed treatment apps may be particularly helpful for those preferring to seek help anonymously and seeking flexibility or for those who have historically struggled to engage with traditional psychology service provision. Less mentioned were virtual reality, wearables, and artificial intelligence, though these modalities are widely believed as becoming more important in the future [27].

Since the start of the pandemic, both clinicians and patients have been required to gain both the skills and experience to adopt e-mental health solutions out of necessity, with a number of descriptive case studies being identified that inform the rapid conversion to e-mental health to continue to deliver care [89,90]. Various included articles articulated the main issues and challenges to this rapid conversion, as well as highlighting the fast-tracking of more relaxed regulations and supported reimbursements to overcome some of the barriers related to practicing virtual care. Included articles suggested that the experience has been largely valuable for patients and health care professionals alike; however, a number of concerns remain, with complexities identified at the levels of the patient, clinician, technology, treatment, ethics, organization or service, and broader system. A “one-size-fits-all” approach will not suit the needs of all patients, clinicians, or services [74]. Future research and interventions in e-mental health should investigate how to overcome or improve upon these barriers.

After any subsequent waves of COVID-19 subside, the key question is whether we are likely to remain in the “new normal,” in which telehealth remains a prominent vehicle for mainstream mental health treatment delivery [66]. The ability of health care systems to continue to provide telehealth depends on the continuation of relaxed regulations and supported reimbursements to incentivize the use of telehealth [66]. It is unclear what rationale the regulatory bodies and insurers will employ to decide which prior limitations, if any, should be reinstated [66]. At the same time, given the large scale at which these tools are now being used, there is renewed urgency to assess and address ethical issues associated with e-mental health tools, such as maintaining privacy and patient data protection, and whether the privacy and safety regulations that have been relaxed need to be tightened to increase accountability [14].

Despite the swift adoption of the multitude of e-mental health apps and platforms that emerged during the pandemic, little is known about their immediate-, medium-, or long-term clinical value or the barriers to, and facilitators of, their uptake [18]. Given that many of the digital tools used in e-mental health have been developed outside of the health care system [91,92], there is also little understanding of how to optimize their integration into existing health care models and their various governance, funding, planning, and accountability frameworks. To date, there continues to be little translational research (ie, the implementation of e-mental health clinical trial research into practice settings) [93], limited use of quality evaluation measures [94], as well as little or no long-term follow-up [95]. This research-to-practice gap suggests there is much to learn about how e-mental health can be best incorporated into real-world settings to reap its benefits [96], with the current rapid uptake of e-mental health solutions providing a stimulus to addressing this gap.

Some key research questions identified in the literature include the need to compare specific outcomes for in-person versus remote care related to specific mental health conditions, such as those with paranoia or other psychotic disorders and substance use disorders. Such conditions may pose unique challenges to management via e-mental health programs and services [97,98], conditions that could also benefit from the greater flexibility of this modality [66,99]. Other questions have also emerged in the current literature: What types of patients respond particularly well to virtual rather than in-person visits? [14,72]; Are there differences in quality of care by socioeconomic status, health literacy, technology literacy, and other individual factors outside of psychiatric conditions?; What are the impacts of telehealth on the therapeutic relationship and on clinicians themselves?; and What are the financial implications of widespread adoption of telehealth? [14]. Now is the time to accelerate e-mental health research to ensure the continued success of virtual modes of care beyond the pandemic.

The strengths of this review are the inclusion of a broad range of articles, including nonempirical articles, reviews, and protocols. In addition to looking at the countries of origin, we also examined article influence and provided a novel keyword analysis with inductive thematic coding to identify key topic areas. Our thorough analysis of the key issues and challenges identified gaps in research to guide the next wave of research on e-mental health. The limitations of this review are primarily methodological. We chose to use CiteScore data because almost 85% of the identified journals had CiteScore information and almost 70% had reported Altmetric data, but not all journals have CiteScore data, nor do all articles have Altmetric Attention Score values available. Similarly, the keyword analysis could only be conducted on journal articles that provided keywords. To supplement this, we undertook an analysis of specific subpopulations, mental disorders, and issues or challenges within titles and abstracts for all publications. Although we included a broad range of journal articles (eg, commentaries, reviews, letters to the editor, and protocols), we did not include a grey literature search, which may have broadened our understanding of the e-mental health field since the pandemic began. Although we identified a relatively high proportion of articles from medium-income countries, our restriction to records in English and published works may have underestimated the true amount of literature emerging from low- and middle-income countries. Although, the inclusion of non-English studies can add substantially to the resources required to complete a review, we believe it could be important to include them in an updated search of the literature in this field in the future. The inclusion of conference abstracts in the updated search would also be beneficial, given the rapidly emerging nature of the topic.

Arguably, the COVID-19 pandemic is the defining moment for e-mental health adoption, with virtual care being likely to remain a prominent vehicle for mainstream mental health service delivery postpandemic. This review identified that many of the emerging e-mental health studies focused on the application of “telehealth” via telephone or videoconferencing, with notable interest in, and concern for, vulnerable populations, such as children, veterans, and health professionals as well as those with pre-existing mental health conditions, including anxiety disorders, depression, substance abuse disorders, and psychosis.

The rapid expansion of e-mental health services during the pandemic has been enabled by existing technology and swiftly implemented policies but has been challenged by numerous issues at the patient or clinician level and at the technology or treatment level, or due to factors at the ethical, organization or service, and broader system levels. Ensuring that the field advances beyond simply the sharp increase in publications documenting the use of e-mental health—and the concomitant interest by readers—to further accelerate access and quality of care beyond the pandemic will be the next big challenge.