We conducted a review of peer-reviewed literature examining the application of digital health interventions for the treatment of mental health conditions described below. We searched Medline on November 1, 2021, for research published after January 1, 2010. We used Medline filters to restrict retrieved records to meta-analyses, systematic reviews, and other types (narrative and conceptual) of literature reviews. We used broad term definitions to maximize the types of digital health interventions and mental health conditions captured in the search. The search strategy consisted of combinations of Medical Subject Headings (MeSH) words and other keywords including the following: virtual reality; telemedicine; computer-assisted therapy; digital health; videoconferencing; mental health; mental health services; psychotherapy; attention deficit and disruptive behavior disorders; anxiety disorders; trauma and stressor related disorders; mood disorders; bipolar and related disorders; dementia; disruptive, impulse control and conduct disorders; dissociative disorders; feeding and eating disorders; neurodevelopmental disorders; neurotic disorders; pain; personality disorders; schizophrenia spectrum and other psychotic disorders; sleep wake disorders; and substance-related disorders (see search query in Multimedia Appendix 1).

This review is restricted to other reviews that assessed the use of digital interventions for the treatment of mental health conditions. Studies that did not report on the effectiveness of digital interventions on mental health outcomes or did not outline a study protocol were excluded from this review.

Once records were retrieved and deduplicated, TJP, NS, and AJ conducted title and abstract screening where any disagreements were resolved through consensus. Team members then proceeded with mutually exclusive full-text screening to identify articles that qualified for inclusion in the review. As with previous meta-reviews, adherence to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [78] was considered to assess risk of bias in selected studies (maximum score of 1: completely adheres to PRISMA Guidelines). To assess the quality and reliability of research within the field, one reviewer conducted data extraction using a standardized and iterative data extraction form (Multimedia Appendix 2). Extracted data included study details (author, date, and type), participant characteristics (mental health conditions), intervention details (intervention type and effectiveness), number of participants, and controls used (treatment as usual, waitlist, placebo, or not applicable). Quality and bias scores describing the primary literature reported in included studies were averaged and faithfully converted (when necessary) to a consistent 3-point scale (1=low, 2=moderate, and 3=high). Owing to the significant heterogeneity in research approaches and findings, we selected a qualitative and semiquantitative approach to summarize and present research findings.

To provide a clearer picture of how digital interventions are used in treating various mental health conditions, we separated mental health conditions based on the parent MeSH terms and Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria. Where we identified dissimilarities or similarities in treatment, we either added a subcategory or combined categories together. We removed attention-deficit/hyperactivity disorder (ADHD) from developmental disorders and added it as a separate category owing to dissimilarities in the treatment and management of this condition versus other developmental disorders. We combined the frequently comorbid anxiety, mood, and trauma disorders owing to similar treatment approaches, effectiveness, and reporting in the literature. Additionally, patients with chronic pain, chronic medical illnesses, and chronic disabilities (shortened to chronic illnesses) often experience mental health issues that are underrecognized, receive little attention within digital health intervention literature [79-83], and have unreliable treatment efficacies [79-83]. Despite the use of similar psychological treatments anxiety, mood, and trauma disorders [84-87], chronic illness treatments also involve acceptance, remediation, music, and virtual reality [79-82,84-86]. We therefore retain chronic illness as a category related to but separate from anxiety, mood, and trauma disorders. Similarly, caregivers are often untrained family members who face significant stress and anxiety in the process of providing care for loved ones. Caregivers also benefit from mental health services such as cognitive behavioral therapy and specific psychoeducation, which overlap with some mental health conditions. but also benefit from peer support, training, and acceptance therapy [70,88-95]. Substance use disorder was included since treatments include therapies based on psychological principles [32,78,96-115], and this disorder is often comorbid with other mental health disorders and is considered a mental health condition by medical associations (eg, Canadian Medical Association, American Medical Association, and World Health Organization) and diagnostic manuals (eg, DSM-5).

This review adheres to PRISMA guidelines [116] (Multimedia Appendix 1).

The PRISMA flowchart of the screening process is presented in Figure 1. We identified 3051 records and used Medline selection tools to exclude primary articles (n=2510), and studies published before 2010 (n=42). Of the remainder, 4 were inaccessible and authors did not respond to copy requests; thus, 466 studies proceeded to full-text review where 159 were excluded for not reporting on intervention effectiveness and 3 were proposals. This selection resulted in 77 meta-analyses, 84 systematic reviews, and 143 literature reviews examining the use of digital health interventions for the treatment of mental health conditions.

A summary of metadata extracted from database searches, curated secondary literature, curated primary literature, and participant numbers is provided in Table 1. Per participant, studies on substance use disorders account for a majority (n=241,377, 52%) of digital mental health research, followed by mood, anxiety, and trauma disorders (136,121, n=29%), and >5% for other mental health conditions (pain: n=24,327, schizophrenia: n=20,500, dementia: n=10,823, feeding and eating: n=10,441, developmental: n=8736, bipolar: n=3573, sleep-wake: n=3333; and ADHD: n=2428). Additionally, limited research has examined the use of digital health to provide psychological support to caregivers of people with dementia and developmental disorders. Lastly, we retrieved no records examining the use of digital health interventions to treat antisocial, avoidant, borderline, dependent, histrionic, and narcissistic personality, dissociative identity, paraphilic, and sexual health disorders. To demonstrate how the amount and reliability of research can be estimated from metadata, we correlated elements in Table 1 and report a 4D correlation (P<.05; see Table S2 and 4D illustration in Figure S1 in Multimedia Appendix 1). Overall, this illustrates a significant need for the development and testing of digital interventions for other mental health conditions. Nevertheless, the number of research publications has steadily increased since 2015 (Figure 2)—a trend that will likely continue with greater interest in digital mental health research.

To more precisely measure the amount of research available to treat specific mental health conditions using a specific digital health intervention, we superimposed the primary digital health interventions on study characteristics and conclusions drawn from our search results (Table 2). All digital health interventions are supplementary to synchronous or real-time communication.

Digital health interventions can be separated into 7 primary categories.

This review found that a majority of studies on digital health interventions are focused on substance use–, anxiety-, mood-, and trauma-related disorders. For patients with these conditions, the greater flexibility, comfort, and routine associated with digital health offered a favorable substitute for in-person visits and retained therapeutic utility. Given this finding, we expect the use of digital health interventions to persist during and after the pandemic owing to the relaxation of insurance and administrative regulations [12,373-377]. The volume and quality of research for these disorders has enabled the discovery of new treatment methods and the refinement of existing digital health tools to improve treatment efficacy.

We also found that the sudden onset of the COVID-19 pandemic led to a rapid shift toward the use of new technology and interventions without the necessary time to train or prepare practitioners and posed challenges for many health care providers. To remedy this, governments, professional organizations, and academics, have created region-specific digital health toolkits [12,378-383] to facilitate and encourage the provision of digital health services. These toolkits are extensive and provide examples of ways in which digital health can be delivered in a meaningful and effective way.

Evidence from this review also suggests that digital health interventions have implications for combatting the dual public health emergencies across North America: the COVID-19 pandemic and the ongoing overdose crisis [11,384]. Findings indicate that there is significant potential for digital health interventions in reducing the harms experienced by people who use substances [32,75,78,96-114,329-339,346-353,385]. Research into digital health interventions for substance use disorders is relatively new and demonstrates the promising use of web-based programs and social media to reach participants instead of relying solely on referrals from practitioners [100,102,111-114,171,339,349,385]. These interventions may offer timely and cost-effective solutions, where texting, moderated forums, validated web-based or computer-based programs, or mobile apps may be used for treatment, psychoeducation, managing ongoing symptoms, and preventing relapse [12,20,25,30, 79,80,85,147,157,175,351] (see Table S1 in Multimedia Appendix 1 for a list of validated tools). Nevertheless, it is important to acknowledge that there are certain instances where in-person contact with a service provider is most suitable. This is particularly important given that many homeless and street-involved populations lack access to and knowledge of technology [386-388].

Similarly, this review found indications that web-based programs in anxiety-, mood-, and trauma-related disorders are poised for similar expansion. Since anxiety and depression symptoms have risen in the general population during the pandemic [10-16], several interventions can be useful for short-term symptom management, such as synchronous communication (videoconferencing or telephone calls) with a therapist [68,96,123-126,129,130,133-139,141,142] and referral to validated web-based [1,30,68,69,84-86,96,118,124, 125,129,130,133-135,137,138,142,145-187], computer-, or mobile-based applications [1,68,96,185-187,193-200] such as those listed in Table S1 in Multimedia Appendix 1. However, we would like to emphasize that interventions were far less successful without practitioner guidance [30,69,73,76,77,155,157,252].

While this review also identified promising developments in digital programs for ADHD, developmental, dementia, eating, schizophrenia, and chronic illness, we found that digital health interventions for these conditions are nascent. Negative findings in sleep-wake and bipolar disorders suggest that significant retooling is necessary for treating these conditions. Furthermore, no reviews on the use of digital health tools for dissociative, elimination, sexual, and personality disorders were identified. The positive outcomes reported for digital health interventions in a wide range of mental health conditions suggest that there may be merit to exploring these interventions in additional clinical contexts during and after the COVID-19 pandemic. Caution is also warranted with patients with schizophrenia, psychosis, or bipolar disorder as technology may be triggering or could exacerbate existing symptoms [147,232,360,361].

Review findings also suggest that synchronous digital contact is an effective substitution for in-person treatment and assessment for many mental health conditions. Considering successes in most mental health conditions, these findings can be generalized to other conditions where less research is available, such as bipolar, sleep-wake–related, and personality disorders. While some health care providers have expressed concerns regarding their ability to build a therapeutic alliance with their patients, research shows that this is not significantly affected by synchronous communication [26,389,390]. Interestingly, synchronous digital health may be beneficial for autism spectrum disorders [269] and social anxiety since it reduces social interaction–related stress, need for eye contact, oversensitivity, and overstimulation. Evidence from this review indicates that synchronous platforms are associated with significant cost and time savings. First, this transition is also beneficial by reducing commutes to work, the ability to organize one’s working day and tasks [391-393], and protects therapists from the risk of physical confrontations [394,395].

Digital health tools have also been found to allow practitioners to reduce the time they spend with each patient, where evidence suggests that spending 10 minutes with patients through synchronous platforms, and providing referrals to asynchronous platforms (eg, web-based, mobile-, or computer-based therapy and cognitive training) is sufficient [30,69,73,76,77,155,157,252,308]. Some patients (eg, children and elderly) may face other barriers to using or accessing technology [396,397], which can be resolved by specific training on using the application [59,62], obtaining help from a caregiver, and could even be accomplished through remote desktop applications (such as Microsoft Teams: Remote Desktop Protocol). Nonetheless, transferring this responsibility to a family member increases caregiver burden and may lead to suboptimal results over the long term. However, the proliferation of untested applications (especially mobile apps) raises concerns around the quality of existing platforms [1,74,154,180,185,187,194-196,198,200,231,237,280,333]. More specifically, these applications often lack validation, reliability, and are not always built on sound psychotherapeutic principles [1,74,154,180,185,187, 194-196,198,200,231,237,280,308,333].

Digital health interventions are also less effective at mitigating the impacts of social isolation, particularly in the context of the COVID-19 pandemic, where public health orders and the requirement of physical distancing is expected to drastically impact peoples’ mental health. Human connection contributes significantly to one’s mental health; therefore, it is important that digital health interventions maintain their human aspect as this is associated with increased efficacy [68,69,84,98,109,111,112, 124,135,151,159,169,170,178,179,199,273,348,349,362]. Findings demonstrate that asynchronous platforms, such as web-based forums, social media, and other digital communities, likely increase patient engagement and adherence to treatment across all mental health conditions [87,232,304,305]. Additionally, preventative education can be disseminated via asynchronous platforms (eg, social media, groups, forums, and schools) for all mental health conditions, as seen in substance use disorders [100,102,111-114,171,339,349]. Owing to increased demand and lack of availability of services during the COVID-19 pandemic, many patients have transitioned to mobile apps and web-based programs without the guidance of a practitioner [13]. Hence, the absence of sufficient research into these venues, their impact on mental health, and the lack of practitioner guidance and support [1,74,154,180,185,187,194-196,198,200,231,237,280,333] raise concerns that these platforms may cause harm. Indeed, government intervention to increase the prominence of validated region-specific tools and resources in web-based and app-related searches may be required.

Another emerging asynchronous technology that can be used for the treatment of mental health conditions are virtual reality tools. Greater accessibility, comfort, and normalcy of the technology will encourage the development of virtual reality interventions on site or at home. Nevertheless, there are also barriers to providing and expanding virtual reality tools. For example, the high cost of equipment acts as a significant barrier, however, lower priced equipment or mobile phones can be used as substitutes [137,220,247]. Additionally, virtual reality tools are based on recent technological advancements, and there is little quality research on the use of industry-standard equipment and even less so for low-cost virtual reality options. Despite these limitations, virtual reality addresses a particular niche of therapeutic tools (eg, exposure therapy) [65-67,98,115,134,183,205,206,208-210,212-215,219,222,311,369] and is an effective tool for pain management [81,83,290,291,294,295], indicating that as technology and research advances, it may become a central component of any comprehensive mental health treatment strategy.

Owing to the social distancing and quarantine requirements posed by the pandemic, patients with mental health disorders already face social isolation in addition to increased stress and anxiety [5-14]. Additionally, patients surviving COVID-19 may experience lingering symptoms and post–intensive care syndrome long after discharge from intensive care units [398,399]. Mental health challenges for these patients include anxiety, depression, posttraumatic stress disorder (PTSD), cognitive decline, and chronic illness [398,399]. Along with previously mentioned interventions to deal with symptoms of anxiety, depression, and posttraumatic stress, virtual reality can be used to reduce stress, distract from pain, and retrain functional movement in patients who experience chronic illness after COVID-19.

Health care providers are also at risk of feeling social and professional isolation as well as burnout [26,394,400], which must be properly managed by managers, the professional organization, and practitioners themselves. Given the anticipated impact of the pandemic on the mental health of health care providers [11,18,401-403], health care organizations will benefit from specialized synchronous, web-based, and mobile therapy and moderated discussion forums to alleviate this burden. Similar interventions have been used with family caregivers [70,88-90] and health care providers [18,401,402,404,405] to treat anxiety, depression, PTSD, and burn out. Therefore, such interventions can help manage health care providers’ mental health.

Over the last two decades, research on the use of digital health interventions to deliver mental health care has increased significantly. Lessons learned from highly studied fields (eg, substance use–, anxiety-, mood-, and trauma-related disorders) can guide the implementation of digital health interventions to treat other mental health conditions. Starting at the most basic level, where practitioner guidance for 10 minutes was essential and often sufficient for the treatment of anxiety-, mood-, and trauma-related disorders, web-based, computer-based, or mobile programs or apps developed for these conditions could be adapted, improved upon, and evaluated to treat other conditions with overlapping symptomatology. For example, one could consider the overlap in symptomatology among mood-related, anxiety, bipolar, sleep-wake–related, and some personality disorders [406]. Thus, digital interventions for the former two conditions could be adapted to include journaling, behavioral modification prompts, and other psychotherapeutic treatments akin to these conditions, and finally be re-evaluated. Nonetheless, for conditions where no treatments exist, the development and digitization of novel treatment strategies is required [233,323,326]. Indeed, the digital nature of these programs enables the collection of regular assessment data, input from patients, and evaluation by health care providers to develop decision trees and machine learning algorithms to instantly improve and personalize treatment plans, require less practitioner time, and provide greater flexibility in treatment delivery.

The rapid pace of technological advancements also poses significant challenges. For example, treatment program implementation has evolved from computerized delivery with CDs to web platforms to mobile apps in the last two decades. Significant technological shifts have forced researchers to completely rebuild the programs despite apparent similarities between these modes of delivery. Many validated programs identified (Table S2 in Multimedia Appendix 1) are outpaced by technological advancements and lack recent updates. First, easy-to-use development and cross-platform tools (eg, React Native and Xamarin) will enable researchers to make, evaluate, and maintain programs despite rapid technological advancements. Second, health care policies and evaluation may need to be modified so that validated tools can evolve over time and across platforms when the underlying therapeutic principles remain consistent.

Existing research on digital interventions rarely covers comorbid conditions, emergency situations, or complex socioeconomic factors. For example, research on people experiencing homelessness is limited to commentaries and policy recommendations based on available research in the general population [32-42]. This is of particular concern when considering that those of lower socioeconomic status or with complex life circumstances show reduced benefits from digital health interventions [148,165,407]. Additional research considering individuals experiencing various psychosocial complexities or comorbid conditions is required.

Further research must endeavor to use appropriate controls and more rigorous design to improve overall study quality assessed in Table 2. Blinding patients to the digital nature of the treatment is difficult, but creative solutions (eg, unrelated cognitive tasks in lieu of treatment) are recommended. Additionally, standardized rating scales (ie, DSM-V criteria) should be used instead of nonstandard assessments or a participant’s opinion on the treatment. Most studies are restricted to treatment duration and lack long-term follow-up (>6 months). Considering digitization of treatments and records, practitioners can automatically request follow-up surveys and assessments via email or text. Follow-up surveys must also consider whether patients have pursued other treatment programs, as these could confound any pertinent treatment effects. Lastly, following successful remission, there is limited research on the use of digital health interventions (eg, email, text, social media, and forums) to prevent relapse, which can be accomplished via email or text [74,110,157,198,232,274,279,330,361,364-367].

Web-based peer support is dependent on human interaction, which can be unpredictable and include uncontrolled variables. For example, since any large number of people can participate in forums for intermittent periods of time, the inevitable turnover can cause cultural shifts. This would therefore require moderation by practitioners. Evaluation is further complicated by the lack of objective and quantifiable pre-post measures in open social media groups and forums. Indeed, practitioner moderated forums or groups may fare better and could automatically request participants to fill out monthly surveys. Further research is needed to address these hypotheses.

There are several limitations to this review. First, to rapidly inform health care stakeholders responsible for managing treatment of a broad range of mental health disorders, we took a comprehensive approach. As a result, we limited the scope to secondary literature sources and utilized a systematic methodology designed for meta-reviews [167]. Since we primarily report on the effectiveness, feasibility, and reliability of digital delivery in lieu of face-to-face treatment, we did not attempt to compare different forms of therapy. This included drawing comparisons to nondigital interventions reported within identified studies, when available (27% of primary studies compared digital health interventions to TAU). Owing to the urgency of this endeavor and to limit the already substantial number of references, we focus exclusively on reports obtained from Medline. This is not atypical as many of the included reports use a single database but can miss some reports.

Additionally, the metadata collection procedure described only approximates the state and volume of research. Reliance on secondary research articles implies that we likely missed recent relevant primary research articles. Nevertheless, our correlative analysis (Table S2 and Figure S1 in Multimedia Appendix 1) suggests that metadata and secondary research can be used to estimate the relative amount and reliability of primary research.

Owing to differences in quality and bias reporting between included literature, we could not report this for individual studies and instead relied on the included literature sources to dictate the quality of research in the field. We observed considerable variability in quality assessments between reviews (Table 2). Potential explanations include the specific selection and inclusion or exclusion criteria of reviews or lower stringency in early discovery studies versus later RCTs. Nevertheless, this raises concerns regarding interreview reliability, which we did not assess here. To enable policy makers and researchers to reliably compile all amassed data, reliably rate studies, and reduce time lost to re-evaluating studies we recommend an update to Cochrane and PRISMA requirements to include the adoption of a single consistent bias and quality assessment reporting methodology and consistent reporting of study details in all reviews. In addition to ensuring similar quality and bias assessment between reviewers within a review, we recommend comparison with previous reviews to ensure greater reproducibility of quality and bias assessments between independent reviews. Nevertheless, living systematic reviews are likely to accelerate research and development in digital mental health interventions and may if designed accordingly upend the systematic review process. Living reviews stem from the ability to continuously update web-based articles with the latest developments in the field. These are a way forward for rapid evidence-based development, collaboration, standardization of digital health tools, and a necessary step forward to improve treatment options.

Although digital delivery of mental health treatment has been in clinical use for a long time, the available research on the topic is far from comprehensive or consistent. New guidelines to increase reliability and consistency of reporting, evaluation, and quality and bias assessments would enable faster literature synthesis and increase confidence. Living systematic reviews for bipolar, personality, developmental, dementia, and sleep-wake disorders would also be very useful to guide and organize novel digital treatment strategies.

Overall, digital treatment strategies paired with synchronous practitioner contact are as effective as nondigital alternatives. However, in offering digital treatments, it is essential to consider feasibility of treatment, caregiver burden, patient-specific symptoms (eg, paranoia), and patient-specific parameters. More research is especially needed in marginalized populations who face greater barriers to mental health treatment access. Thus, to maintain treatment quality and efficacy, patients should have the option for face-to-face interventions, despite the challenges posed by the COVID-19 pandemic. Nonetheless, digital treatments offer many benefits such as increased patient engagement, accessibility, and availability paired with reduced practitioner workload. Additionally, the drastic shift to digital health is likely to encourage further developments in treatments for many mental health disorders and expansion into other digital modalities, such as virtual reality, social media, and web-based forums. These developments promise significant advances in mental health treatment via global collaboration and investment.