The use of technologies such as phone or video calls between mental health professionals and other mental health professionals, patients, service users, family members, or carers to deliver mental health care (telemental health) has become an essential tool in recent months; importantly, it has taken a central role internationally in maintaining mental health services during the COVID-19 pandemic [1]. Policy makers and mental health professionals, along with mental health service users, now express interest in continuing some use of these technologies long term, even in the absence of pandemic-related social distancing requirements [1-3]. The potential benefits of remote technologies extend beyond adaptation to government social distancing guidelines, allowing the efficiency and flexibility of mental health services to be maximized. The mobilization of telemental health during the pandemic has happened largely ad hoc, thereby achieving remarkably rapid but highly variable implementation. This emergency response has mostly occurred without systematic references to the literature. To plan the effective and acceptable deployment of telemental health beyond the pandemic, it is crucial that we now review all the relevant evidence regarding potential impacts, challenges, and outcomes of widespread remote technology utilization and identify the key mechanisms for its acceptable integration into routine care [4].

Telemental health offers a number of potential benefits that generate a significant interest among service providers not only during the pandemic but also in the long term. For service users across a range of populations, settings, and conditions [5], potential benefits include convenience and improved accessibility, particularly where issues such as physical mobility difficulties, anxiety, or paranoia impede a face-to-face contact [1]. Potential advantages for staff include reduced environmental impact, greater convenience with opportunities for working at home, and ease of communication within and between mental health teams [2]. Although some have argued that problems with building rapport [6] along with privacy and safety concerns may hinder the implementation of remote care, service users have been found to report such apprehensions less than clinicians [7]. Several studies have also suggested that telemental health may be more cost-effective than face-to-face delivery [7].

Despite potential benefits and a substantial body of relevant research, the delivery of remote working remained very limited in most countries before the pandemic with substantial implementation barriers [8], along with the potential for inequalities to be exacerbated. Digital exclusion is an important concern for service users who lack the necessary skills, space, equipment, and monetary resources to access web-based treatment. This is more marked in marginalized groups such as people from BAME (Black, Asian, and minority ethnic) and low socioeconomic status backgrounds [9]. Further risks include the loss of privacy and deterioration in therapeutic relationships [1,10-12]. Staff participation is also impeded by technological and environmental difficulties, and they express reservations regarding the quality of assessments, deterioration of therapeutic relationships, and limitations in the extent to which there is a focus on physical as well as mental health [8,11,12].

Therefore, the potential benefits and disadvantages of telemental health are finely balanced. Risks of longer-term rollout of telemental health without close attention to intended and unintended consequences include the digital exclusion of some of those already most disadvantaged and decline in the quality of care and, potentially, of outcomes. One source with the potential to inform policy makers and service planners in their future telemental health strategies is the substantial body of research studies published before the pandemic. This paper therefore aims to provide a rapid summary of the existing literature on the effectiveness, cost-effectiveness, barriers and facilitators for implementation, acceptability, and reach of remote interventions for the assessment and treatment of mental health problems. Our objective is to identify, appraise, and synthesize the systematic reviews of literature and guidance on telemental health, including qualitative and quantitative outcomes using the umbrella review or review of reviews methodology. Umbrella reviews are useful in summarizing a broad evidence base to inform policy [13]. It is hoped that the results will help illuminate the benefits and remaining challenges while implementing telehealth technologies during the remainder of the pandemic and in the perhaps permanently changed reality that follows.

A rapid umbrella review was conducted guided by the World Health Organization practical guide for Rapid Reviews to Strengthen Health Policy and Systems [14] and adhering to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [15] and umbrella review guidance [16]. In line with the agreed rapid review methodology, this review aims to provide a timely but robust answer to the research question by accelerating some aspects of the systematic review process while maintaining transparency and protocol-driven decision making throughout [14]. The protocol was prospectively registered on PROSPERO (CRD42020208085).

A total of 7 reviewers (LG, CC, PB, TS, LSR, JW, and HIJ) extracted data from the included reviews using a Microsoft Excel-based form. In total, 10% of extractions were double checked by a second reviewer, and inconsistencies were discussed and corrected. The extracted data included citation details, objectives, type of review, participant details (including gender, ethnicity, age, and mental health diagnosis accompanied by staff details where relevant), types of telemental health intervention reviewed, setting and context (mental health service, community, inpatient or residential, or primary mental health care service), number of databases sourced and searched, date range of database searching, the publication date range of studies included in the review informing each outcome of interest, number of included studies, types of studies and country of origin of studies included, instrument used to appraise the primary studies and the rating of their quality; reported clinical, cost-effectiveness, and implementation outcomes; method of synthesis or analysis employed to synthesize the evidence; and conclusions of the review authors.

The quality of each included systematic review was assessed using the AMSTAR 2 checklist [17]. This is a revised version of the original AMSTAR checklist [18], which is a validated measure of systematic review quality [19]. The checklist was used to give each review an overall rating of quality ranging from high (0 or 1 noncritical weakness: providing an accurate and comprehensive summary of the results of the available studies that address the question of interest) to critically low (more than 1 critical flaw with or without noncritical weaknesses: the review should not be relied upon to provide an accurate and comprehensive summary of the available studies) [17]. The study quality was assessed along with data extraction. Table 1 presents the quality ratings.

Heterogeneity in study populations and interventions included in the review, as well as broad inclusion criteria for review study design (eg, qualitative and quantitative), prevented the quantitative pooling of syntheses. As a result, we conducted a narrative synthesis of all interventions and outcomes [39]. This allowed for a more in-depth consideration of all outcome measures and variations in the remote intervention delivery. We grouped reviews by the included population (mental health diagnosis) and further considered the variation in interventions on offer within these subgroups. This was performed for each outcome of interest. Most reviews provided a synthesis of multiple intervention types or failed to adequately differentiate them, making a more thorough comparison across formats impossible.

The search returned 1086 reviews, of which 292 potentially relevant full-text articles were identified. Following full-text checks, 19 reviews met the inclusion criteria (Figure 1), reporting 239 individual studies and 20 guidance documents. In total, 15 of the included reviews examined the clinical effectiveness of telemental health as compared with (1) face-to-face interventions and assessments (n=4), (2) treatment as usual (n=2), or (3) a variety of comparators, including face-to-face, telephone, and treatment as usual (n=9). Eight reviews reported on implementation (broadly defined), including process variables, fidelity, and uptake of interventions, and 10 reviews reported outcomes related to acceptability, including the satisfaction of both service users and clinicians. In total, 1 review focused specifically on the differences in therapeutic alliances between treatment modalities. A total of 2 reviews reported cost-effectiveness, one on this topic only and the other in combination with clinical effectiveness. One review synthesized international guidance on the conduct of videoconferencing-based mental health treatments. Table 1 presents full details of the included reviews, and Figure 1 shows the information on the search and screening process.

Some primary studies were included in more than 1 review: 26 studies appeared in 2 reviews and 27 studies appeared in 3 or more. The remaining 186 studies were included in only 1 review. The double counting of primary studies due to inclusion in multiple reviews contributing to the same outcome was only found for clinical effectiveness outcomes. However, conclusions were similar across reviews, and no review had all the same studies contributing to any synthesis. Multimedia Appendix 2 [20-36,38,40-276] and Multimedia Appendix 3 [40-276] present further details of the study overlap.

Most reviews elicited low confidence in quality appraisal because of multiple weaknesses in the study design. The most common weaknesses were a lack of explicit statements that a protocol was developed before the commencement of the review (explicit statements were reported in two reviews [20,35]), lack of duplicate study selection (duplicate selection was reported in five reviews [28,31,32,35,37]), no report of excluded studies and reasons for exclusion (exclusions were reported in two reviews [31,35]), and no report of sources of funding (sources of funding were reported in three reviews [26,27,31]). Meta-analysis was not performed in the majority of reviews, usually because of heterogeneous data or aims centering around more narrative conclusions such as acceptability (n=12) but in those that included meta-analysis [24,27,31,34-36], all except two [31,34] reviews assessed publication bias. The potential impact of risk of bias was only assessed in two reviews performing meta-analysis [31,34], but all reviews performing meta-analysis used appropriate statistical methods for combining results. The reviews eliciting higher confidence (moderate) were two Cochrane reviews [31,36]. Table 1 presents the quality ratings of reviews, and Multimedia Appendix 4 [20-32,34-38] shows the full details of quality assessments.

Clinical outcomes were reported in 15 reviews [20-24,26-31,34-36,38]. Across all patient populations, including patients with anxiety (n=3), posttraumatic stress disorder (PTSD; n=2), depression (n=4) (including in ethnic minorities, n=1 [21], and older adults, n=1 [20]), substance use disorders (n=1), and multiple disorders (n=4), videoconferencing interventions have been reported to cause significant reductions in symptom severity, with outcomes comparable with face-to-face controls where these were included. Telephone-based interventions tended to report similar significant reductions in symptom severity. However, a review of telephone interventions with older adults with depression [20] reported more mixed findings: reductions were reported in emergency room and hospital visits in one study, and in depression in another; however, a third study suggested that telephone interventions did not add to benefit gained from a web-only intervention. Follow-up treatment gains were less widely reported, and conclusions were mixed across the reviews. Although the maintenance of improvements was found at varying follow-up assessments in one review [34] and at 3-6 months in two additional reviews [26,38], two other reviews reported that evidence was inconsistent with some studies reporting that videoconference interventions may show less longevity in the maintenance of effects up to 6 months compared with face-to-face interventions [21,24]. A final review of mixed modality remote interventions suggested that although inferior to face-to-face formats at shorter follow-up, remote interventions may be more beneficial than face-to-face interventions at longer follow-ups (36 months) [20]. Table 2 presents further details on the clinical outcomes.

A total of 2 reviews presented conclusions regarding the economic impact of telepsychiatry [29,33]. One review concluded that telepsychiatry can be cost-effective as compared with face-to-face interventions, particularly in rural areas where there were lower numbers of consultations required before telepsychiatry becomes more cost-effective (combatting initial equipment costs) [29]. The second review, in which the focus was on the cost-effectiveness of telepsychiatry, reported that 60% (15/25) of the included studies found telepsychiatry programs to be less expensive than standard in-person care, due to savings such as the travel time and reduced need for patients and their families to take time off work. However, the remaining studies included in the review concluded that telepsychiatry programs were more expensive, particularly because of videoconferencing equipment costs (n=8) or found no difference in costs (n=1). The review also found a large range in reported costs, with, for example, a long-term delivery of telepsychiatry for veterans ranging from US $930 to US $2116 per patient. The cost-effectiveness analyses within the review suggested that telepsychiatry was less cost-effective than face-to-face alternatives. Accordingly, the review concluded that variation was due to a large disparity in the reporting of costs, for example, whether personnel costs or initial equipment costs were included, and that there remains a need for future efforts to determine the cost-effectiveness of different forms of telepsychiatry, particularly for different disorders and applications of remote technology (eg, consultation vs therapy). In addition, a third review [21] examined health service utilization, which impacts cost-effectiveness. They found that the rates of antidepressant and health service utilization were similar in the 3 months following both telephone and web-based counseling as compared with usual care or face-to-face controls.

Only one review [37] of guidelines for telemental health met the inclusion criteria. This review comprehensively summarizes the guidance published to date, including guidance on decisions about the appropriateness of e–mental health, ensuring the competence of mental health professionals, legal and regulatory issues, confidentiality, professional boundaries, and crisis intervention. Recommendations from 19 guidelines were characterized as either firm (50% or more of the guidelines recommended this) or tentative (fewer than 50% of the guidelines recommended this). The review identified the following as firm recommendations: ensuring that the remote interventions were appropriate for the needs of individual patients and within the boundaries of therapist competence, as well as adhering to laws and regulations; maintaining confidentiality and seeking informed consent (including for specific aspects of remote appointments such as data security); and ensuring that geographically accessible in-person clinical support is available in the case of a crisis or emergency. Guidelines suggested a higher risk of harm for people with cognitive impairments and psychotic disorders but did not provide stronger recommendations on how to adapt delivery to these populations. Furthermore, a minority of guidelines discussed remote technology in young people, with the main message being the importance of checking consent with both the patient and parent. A full summary of the recommendations from the review can be found in Multimedia Appendix 5.

Our umbrella review retrieved a variety of recent relevant systematic reviews on which the future planning of telemental health implementation can be usefully drawn. Across the 19 reviews included in this umbrella review, the results suggest that the remote forms of assessment and intervention can produce at least moderate decreases in symptom severity for people affected by a variety of mental health conditions. Arguments are strongest for videoconferencing interventions, with multiple reviews concluding that outcomes appear comparable with face-to-face interventions at the end of treatment. However, at present, conclusions regarding longer-term results remain uncertain, whereas some reviews have reported the maintenance of positive effects up to six months for both videoconference- and telephone-based interventions, other reviews have suggested that effects are less long-lasting than for face-to-face interventions and the amount of evidence on which to base this assessment is limited. An avenue for future research could be to further examine possible differences between settings in which longer-term benefits have been found compared with those which did not find this.

Reviews also suggest that remote interventions are acceptable to service users participating in studies who tend to report being as satisfied as in face-to-face interventions. This, along with reports that participants adhere to telemental health-based interventions at similar rates to face-to-face interventions, is promising with respect to adaptations during the COVID-19 crisis and for the future, but the reviews tend to relate to small-scale and carefully planned implementations of telemental health with volunteer participants, rather than to large-scale emergency implementations, as in the current crisis. Conclusions drawn from the reviews regarding specific factors that could impact patient satisfaction should also be noted. For example, a report of reduced skepticism following positive experiences [25], if confirmed, suggests that induction sessions and other methods of familiarizing patients with technology and making it accessible to them may improve acceptability and uptake. Clinician satisfaction varied more with reviews tending to conclude that although remote interventions may be acceptable, clinicians usually prefer face-to-face interventions. This may be related to the reports in some reviews that the clinician ratings of therapeutic alliance are poorer with telemental health [34,277]. Despite this, patients tend to feel that the alliance is on par with face-to-face interventions [30,34,38]. There is some suggestion that training and more experience with video and telephone-based technology for intervention delivery may alleviate this concern among therapists [277]; however, staff reports following an increased uptake in the COVID crisis appear to suggest continued concerns about rapport [2]. In the future, a thorough exploration of the exact reasons for acceptability and adherence would benefit the evidence base, for example, certain contextual factors such as appointment type, mood, and time of day may have a substantial influence on patient satisfaction and should be explored further.

Evidence yielded by reviews on the important questions of whether assessments appeared accurate and comprehensive and whether treatment was delivered as intended was limited. A total of 2 reviews examined the comparability of remote versus face-to-face assessment, with one review finding a good correlation between assessments and another finding that there was insufficient high-quality evidence published thus far to draw accurate and meaningful conclusions [27,32]. Regarding fidelity, we found one review that reported good therapist fidelity and competence in remotely delivered interventions in the context of service delivery for veterans with PTSD [38], but systematic reviews do not appear to yield evidence as to whether high fidelity and quality is achieved with telemental health interventions. Standardized training rooted in evidence will be important in ensuring high-quality intervention delivery and overcoming self-doubt among clinicians in delivering remote interventions [37,277,278].

A crucial question regarding the rapid adoption of remote technologies during the pandemic has been whether service users may drop out of or be excluded from care as a result. A minority of the reviews included relevant data, most of which were relatively reassuring. Reviews reported that remote interventions were convenient and that examining uptake reported an increase. When examined, retention was also comparable with face-to-face treatment [30,38]. Reports of technological difficulties were reassuringly few across reviews, although this may be more easily achieved with well-planned, smaller-scale implementations of telemental health that characterize research studies than with larger-scale implementation. However, one aspect of remote delivery on which reviews did not generally report is the risk of complete digital exclusion for patients who may not have the skills or resources to engage with remote therapy or assessments [1,2]. The implementation of telemental health across service systems is only likely to be beneficial if there are clear plans to prevent patients with limited access to technology from being at a disadvantage [279,280], whether by supporting them to engage with remote care or ensuring that equivalent care is available face-to-face.

Digital exclusion may result in the exacerbation of existing inequalities where already disadvantaged groups, such as older adults, people with sensory or cognitive impairment, or members of some BAME groups are at a greater risk of exclusion [1,9,281]. Some reviews have examined this issue [20,25]. A single review by Dorstyn et al [21] reported that members of predominantly North American ethnic minority communities with depression benefited from telecounseling. A more substantial evidence base is thus urgently required to evaluate the risk of exacerbating ethnic inequalities in mental health care access through telemental health adoption. Furthermore, many have argued that the shift to remote working may exclude older adults [25,281]. On the basis of one review [20] suggesting that videoconferencing interventions can be comparable with face-to-face sessions, and another review [25] finding high levels of patient satisfaction, therapeutic alliance, attendance, and convenience, this review suggests that effective remote intervention delivery may be feasible for older adults. This is encouraging as staying at home and avoiding infection during the pandemic is especially desirable for older adults. No reviews were found regarding other subgroups of potential concern, such as people with sensory or cognitive impairments, children and adolescents and their families, or people with comorbid mental and physical health conditions. We also did not find substantial evidence on settings of particular interest, such as mental health inpatient services (including the use of telemental health in compulsory detention processes) and crisis services.

The findings of this umbrella review should be considered along with a number of limitations. First, umbrella reviews aim to present an overview of findings from systematic reviews [282], making conclusions reliant on the quality and accuracy of reporting of included reviews and necessarily resulting in some loss of nuance when findings are pooled. For example, information regarding follow-up periods was not always clear in the conclusions presented in reviews, making a thorough examination of the longevity of benefits difficult. Although we included only reviews considered to be systematic (defined here as searching at least three databases and conducting a quality assessment when synthesizing quantitative data), it was apparent from our quality assessment that the majority of reviews lacked several attributes or characteristics of a high-quality review with robust conclusions, such as prespecified protocols and duplicate study selection. However, our aim was to gain a rapid overview, especially relevant to the current and future rapid implementation of telemental health, of the extent of supporting evidence available in previous literature regarding telemental health: the umbrella review method provides a useful route to achieving this. The inclusion of systematic reviews focused on methods other than randomized controlled trials and on guidance further increases the methodological variability of included reviews and studies, but it is a choice we have made to maximize the retrieval of material from which real-world important lessons can be learned regarding feasibility, acceptability, and implementation barriers and facilitators [283].

This review also aimed to summarize the outcomes related to the cost-effectiveness of remote delivery. We found only two reviews that summarized this outcome and only one that did so comprehensively. It is important that further work should be done to establish the cost-effectiveness of different forms of telemental health, for different patient groups, but there is a significant gap in the literature, despite efficiency being presented as one of the arguments made to support remote interventions [284].

Finally, this review aimed to summarize the literature published before the COVID-19 pandemic to identify evidence relevant to the current context and the recovery from the pandemic. However, the pandemic has given rise to a much more extensive switch to telemental health than previously, meaning that not all conclusions may be generalized to the new normal. In particular, the evidence retrieved in this review tends not to relate to the implementation of telemental health across whole catchment areas and does not yield much evidence relevant to currently highly salient issues such as risks of digital exclusion or exacerbation of mental health inequalities and economic disadvantage, which may well be exacerbated as a result of COVID-19 [1,2]. The conclusions of this review should be supplemented with the further scrutiny of the adoption of telemental health within the context of these societal changes, for example through discussion with people with lived experience (Textbox 1).

The research across a range of mental health conditions suggests that telemental health is potentially an effective, feasible, and acceptable tool for providing mental health treatment, at least when interventions are relatively well-designed and well-planned, as has been the case in research studies. Comparability in terms of symptom improvement and satisfaction relative to face-to-face methods suggests the move to telemental health to sustain mental health services during the pandemic has been a reasonable one; however, the context of this emergency implementation has been very different from most research studies. Further research should seek to build on existing evidence for establishing the long-term effectiveness and cost-effectiveness of telemental health in a range of groups and settings, such as including children and young people and inpatient acute services and focusing on issues of inclusion and reach. A further question on which further evidence would be highly desirable is the extent to which digital exclusion can be remedied, including the examination of interventions designed to include those with limited previous digital resources or skills. Future planning for telemental health implementation should draw on previous research evidence, often acquired in relatively small-scale studies, and on experiences of trying to engage large service user populations and most of the mental health workforce with remote technology delivery during the COVID-19 pandemic.