We performed a scoping review, following the framework stages proposed by Arksey and O’Malley [33] and further refined by Levac and colleagues [34,35]. The reporting adhered to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines [36] (Multimedia Appendix 1). This review approach is designed to address an exploratory research question with the aim of mapping the existing literature and key concepts in a defined field. It serves to identify gaps and types of evidence in research [35].

We performed a systematic search of the PubMed, MEDLINE, Embase (via Elsevier), PsycINFO, and PsycArticles databases. The search terms included the following keywords: (1) treatment ((blended) AND (therapy)); (2) broader digital health ((internet) OR (mobile) OR (online) OR (digital)); and (3) population (“severe mental illness” OR “serious mental illness” OR “schizophrenia” OR “psychosis” OR “psychotic” OR “psychotic disorder*” OR “schizoaffective” OR “schizo-affective” OR “bipolar disorder*” OR “mania” OR “manic” OR “bipolar” OR “depression” OR “major depressive disorder” OR “antipsychotic” OR “personality disorder” OR “inpatient”). The search strings were converted for each database. We also manually searched the bibliographies of relevant studies and reviews using the snowball principle to identify further eligible articles. We included publications from the date of inception up to March 23, 2023, in the search.

In accordance with the PRISMA-ScR guidelines for conducting scoping reviews [36,37], formal quality appraisal was not incorporated into the study selection process. Nevertheless, we evaluated the study quality of the selected articles using a quality assessment checklist developed by Ellis and colleagues [38]. This was done to ensure that the included studies adhered to standard research norms and to provide insights for future research. The scores of the studies ranged from 9 to 11 (out of a maximum of 12), signifying a high level of quality for the selected articles. Further details can be found in Multimedia Appendix 2 [39-61].

Data charting was undertaken as a collaborative and iterative process. A customized data extraction worksheet was created and consistently adjusted to capture all data that could be pertinent to addressing the outlined research questions. The ultimate data chart underwent scrutiny for validity and data accuracy by 6 members of the research team. The comprehensive data extraction form is available in this publication (refer to Multimedia Appendix 3 [30,39-62]).

Following the removal of duplicates, a total of 587 articles underwent screening, and 59 papers were assessed in full for eligibility. We identified 25 articles that were relevant to the scope of the review. Among these, 23 papers [39-61] reported as main studies on aspects of feasibility, adherence, and effectiveness of bPT interventions, while 2 [30,62] were substudies analyzing aspects of adherence only. For the quantitative analysis of the review, these 2 substudies will not be included. We specifically utilized the information for the analysis of adherence. The outcomes of the article selection process are illustrated in Figure 1.

Tables 1 and 2 outline the characteristics of the included studies (n=23) and patients (n=2554). The papers incorporated into the review were predominantly from research groups in Germany (8/23, 35%) [43,44,46-48,53,57,58], followed by the United Kingdom (5/23, 22%) [39,40,51,54,59], the Netherlands (4/23, 17%) [49,50,55,61], and Australia (3/23, 13%) [42,45,56]; 1 study was conducted in the United States [60], and 2 were structured as transnational collaborative research [41,52]. We identified articles published between 2012 and 2021.

Among the bPT studies, the majority (12/23, 52%) utilized randomized controlled trial designs [43-49,51,52,57,59,61]; 2/23 studies (9%) were nonrandomized controlled trials [42,55], and 1 used a quasi-experimental 4-group design [39]. The remaining studies comprised 6/23 (26%) observational pilots [50,53,54,56,58,60], 1 observational study [40], and 1 qualitative evaluation [41].

Among the bPT trials with control groups (15/23; 65%) [39,40,42-49,51,52,57,59,61], 11 utilized treatment as usual (TAU) groups [40,45-49,51,52,57,59,61], 3 used an active control group [39,43,44], and 1 study had a waiting list control group [42]. The bPT interventions covered a range of SMI study populations, involving a total of 2554 patients with SMI. Notably, for the trials conducted by Zwerenz and colleagues [43,44], the same study population was utilized to generate different data sets—1 on inpatient treatment and another for follow-up on the same population after inpatient treatment.

We identified 14/22 (64%) disorder-specific study populations, comprising patients with psychotic disorders (n=5), severe major depression (n=3), borderline personality disorder (n=2), anorexia nervosa (n=2), bulimia nervosa (n=1), and hoarding disorder (n=1). Additionally, we identified 8/23 (35%) transdiagnostic or symptom-oriented bPT interventions: 3 targeting affective symptoms, 2 addressing psychotic symptoms, and 2 designed for patients meeting the criteria of SMI as defined by the Dutch consensus group on SMI [29].

We identified a variety of bPT intervention formats (Tables 3-6), primarily focusing on outpatient treatment (21/23, 91%). The duration of the analyzed interventions ranged from 2 sessions to 18 months, varying based on the specific objectives of each intervention. Among the 23 interventions, 17 (74%) used an integrated blending approach [39,43-45,49-61], with 12 of them adopting a concept of transformational blending, and 5 [42,45-48] utilizing an additional blend. Additionally, 6/23 (26%) studies designed the bPT interventions in a sequential manner [40,42,45-48]. One study incorporated the digital component before intensive face-to-face psychotherapy [40], while 5 studies [42,45-48] included the digital component as part of an after-care concept following intensive face-to-face treatment. Among the sequentially designed interventions, 2 used a transformational blend [40,42], and 4 utilized an additional blend [45-48].

Concerning psychotherapeutic approaches, the majority of bPT interventions (16/23, 70%) used principles and methods derived from cognitive behavioral therapy (CBT) with various focuses. These ranged from disorder-specific CBT for depression [43,44,49,50], eating disorders [47,48,51], or psychosis [41,51,52] to competence training for social skills [53], social cognition skills [54], or voice coping [52]. Some interventions (4/23, 17%) centered on recovery [30,45,55,56] or maintenance therapy and relapse prevention [42,46-48]. Klein et al [57] and Jacob and colleagues [58] based their bPT interventions on the principles of schema therapy, and Craig and colleagues [59] used an interpersonal approach as the foundation of an AVATAR-based bPT program for the treatment of patients with SMI with psychotic symptoms [59]. Zwerenz and colleagues [43] integrated face-to-face psychodynamic inpatient treatment with a CBT-based online application (deprexis).

Among the 23 included bPT studies, a total of 20 different digital components were used. Some of the trials used the same digital component for different studies. Specifically, Jacob and colleagues [58] and Klein et al [57] used the priovi digital component, Kooistra and colleagues [49,50] used a self-developed i-CBT program, and Zwerenz et al [43,44] used the deprexis digital component.

The majority of the digital components (13/20, 65%) were based on online programs, while 5 (25%) used smartphone-based apps. One intervention incorporated an app that was compatible for both online and mobile use [51]. Two (10%) non-online–based digital components were included in the review. One study used vodcasts, which are short video clips aimed at encouraging reappraisal and acceptance of food in the blended treatment of patients with anorexia nervosa [39]. Another study combined an AVATAR program with face-to-face psychotherapy and psychosocial care to improve coping with distressing auditory verbal symptoms in psychosis [59].

A total of 7/35 (35%) digital applications provided opportunities for peer contact, either with known peers from the same treatment group [47,48,55,60] or with a broader peer community for mutual support and inspiration [30,45,46]. Therapist assistance for the digital component was implemented in 11/20 (55%) studies [41,42,45-50,55,58,61], either through individual feedback on patients’ completed assignments or by providing the option for patients to contact therapists via email or chat. Six applications offered electronic platforms or message boards for contacting and communicating with program peers [45-48,55,60], sometimes facilitated by a therapist (2/20, 10%) [47,48]. Tables 3-6 present a detailed account of the bPT intervention details.

We defined feasibility as whether the bPT study intervention was suitable for the given setting and the respective study population. Adherence addresses participants’ engagement and compliance with the bPT treatment [62,63]. We found that 22/23 (96%) studies concluded that their bPT interventions were feasible. One study reported nonfeasibility due to participants’ lack of computer skills, trainers’ hesitation toward the program, and the program’s inflexibility [30]. Feasibility aspects focused on acceptability, patient and therapist satisfaction, and overall ratings, which were reported to be good to excellent (see Multimedia Appendix 3). Safety considerations were addressed in 7 studies [41,45,48,54,56,57,60], with 1 study [58] reporting adverse emotions triggered by self-use of the digital application.

Regarding treatment adherence, all but 1 study (22/23, 96%) reported on it, mainly focusing on patients’ adherence to the digital component. Various indicators were used, such as completed tasks, frequency and duration of usage, log-ins, self-reported usage, and participation in live chats. Four studies analyzed the impact of treatment adherence on intervention effects [47,48,58,60]. Two studies found no correlation [48,58], while 1 reported a positive association between app engagement and symptom reduction [60], and another identified a significant effect of treatment adherence on participants’ body weight [47].

Adherence to the face-to-face component was examined in 1 study [48], which found that patients without regular face-to-face treatment were more likely to not complete the online program. The overall treatment adherence rates varied between 0% and 37%, and in controlled trials, lower dropout rates were generally observed in the intervention treatment compared with control groups. In controlled trials, 12/15 (80%) reported lower dropout rates in the intervention treatment compared with control groups, although differences overall appeared small and heterogeneous, ranging between 0.5% [46] and 15% [55].

To provide a concise overview of the treatment effects of bPT interventions for SMI, the main outcomes of the controlled studies are summarized in Table 7.

Among the 15/23 (65%) studies that used control group designs to examine the efficacy or effectiveness of bPT interventions, 4 (17%) reported significant effects in favor of the intervention [43,44,46,59], with 2 of them using the same study population [43,44]. Three of these studies focused on bPT effects in patients with affective disorders [43,44,46], while Craig et al [59] investigated the treatment of distressing auditory verbal hallucinations as part of psychosis. Two trials compared bPT with TAU groups without a digital component [46,59], and Zwerenz et al [43,44] compared the intervention with an active control group with face-to-face treatment and different digital interventions.

Among the trials without significant main intervention effects, 5 reported significant effects on secondary outcomes [45,47,48,51,52], indicating potential efficacy. Klein et al [57], in a study involving patients with borderline personality disorder, found a significant intervention effect in a per-protocol analysis, specifically for participants from the intervention group who regularly used the digital component (priovi). Additionally, Cardi et al [39], in their quasi-experimental bPT study on the positive impacts of listening to vodcasts on food appraisal in patients with anorexia nervosa, found an interaction effect with the context: outpatients benefited more from the vodcast intervention, while inpatients reported more benefits from the control condition, which involved listening to classical music.

As many as 3/15 (20%) of the controlled studies [42,49,55] concluded that there were no group differences between intervention and control groups. Kooistra et al [49] in their investigation of the efficacy and cost-effectiveness of a manualized bPT intervention for patients with severe depression found no cost-effectiveness from a societal perspective but an acceptable probability of being cost-effective from the health care provider perspective.

Beentjes et al [61] reported significant effects in favor of the intervention for self-management, recovery, and general health perception. However, due to low treatment and study adherence, the authors deemed the results inconclusive.

In this scoping review, we provided a comprehensive summary of the current evidence on bPT for SMI, with a specific emphasis on feasibility, adherence, and effectiveness. Our review encompassed 25 research papers, consolidating data from 23 bPT interventions that involved a total of 2554 patients with SMI. Our findings offer preliminary evidence supporting the feasibility of bPT treatment for patients with SMI. In examining adherence, our analysis identified it as a crucial factor in exploring and advancing bPT treatment for SMI. However, it is important to note that empirical research in this area is currently limited. The potential positive impact of bPT on adherence in SMI treatment is indicated by the attrition rates reported in the summarized studies. These rates ranged from 0% to 37%, demonstrating favorable outcomes when compared with both guided and unguided technology-based interventions (TBIs). For instance, Musiat et al [64] reported average completion rates of 64% for guided TBIs and 52% for unguided TBIs. When considering attrition rates for mobile health (mHealth) interventions, with a recent meta-analysis reporting an average of 24% [65], the attrition rates found in our studies appear comparable. Nevertheless, the accumulated evidence on the effects of bPT for SMI remains sparse and challenging to extrapolate. Among the 15 studies with a control group design, 4 concluded that bPT interventions were effective compared with control groups. However, 2 of these studies used the same study population, with significantly varied control groups. Additionally, 3 studies investigated patients with affective disorders.

Our feasibility results are in line with earlier systematic evidence on the feasibility of eHealth applications for SMI [66,67], as well as with 2 recent reviews on mHealth for patients with SMI, with a focus on monitoring [18] and disorder management applications [68]. These reviews also established the general acceptance and feasibility of digital interventions in patients with SMI. Our efficacy conclusion is more reflective of the quality of the reviewed research than of the actual subject matter. It primarily focuses on 2 noticeable aspects: First, bPT interventions under scrutiny in the analyzed studies varied greatly in their therapeutic approaches, objectives, digital and face-to-face component features, target groups, and treatment settings, making it challenging to generalize across studies. Second, the studies were heterogeneous regarding research designs, control types, and outcome measures. While these aspects are an outcome of the review’s overall broad scope, they also mirror the current heterogeneity of this field of research.

Despite these conclusions, we remain optimistic about the potential of bPT for the treatment of SMI based on several considerations. On the one side, substantial and increasing evidence supports the effectiveness of face-to-face psychotherapy and psychosocial care for SMI [69-72], although the effects are low to moderate [73,74], and access is often limited due to systemic [75,76] and individual factors [77]. On the other side, the utilization of digital technology and applications is becoming increasingly common among individuals with SMI, and there is ongoing development of digital health interventions [78], including recent advancements in artificial intelligence for SMI treatment [18,79]. However, research also indicates a negative correlation between adherence and the effectiveness of stand-alone TBIs and disorder severity [23,65,80], indicating a fundamental need for human facilitation of digital interventions for SMI [66]. Therefore, bPT, with its concept of enhancing empirically supported psychological treatment with digital content, appears to hold promise for effective, potentially cost-effective, and more accessible treatment for this complex and severely affected patient group. We conclude our review by emphasizing the need for further research in this field.

We propose several considerations for further research and clinical directions. First, the design of bPT interventions should take into account various contexts of bPT for SMI and consider the advantages and opportunities of different bPT concepts (Multimedia Appendix 4). Using participatory research approaches, as demonstrated by Austin et al [41], can help ensure the creation of appropriate and sustainable treatment solutions based on patients’ needs. Second, it would be advantageous for bPT research to investigate the effects of bPT, giving equal attention to both digital and face-to-face components. This should include the exploration of how these components interact with each other and with additional contexts, such as treatment settings, therapists, and mental health care systems. This would necessitate careful deliberation of suitable control groups. Third, adherence is a critical factor in the effectiveness of TBIs in general [23,65,80], and particularly in the treatment of SMI [12,31,66,81]. Therefore, more research on this topic, including the development of analytical concepts, is required. Fourth, given the multidisciplinary and often long-term nature of SMI treatment, implementation research on bPT for SMI could foster the development of protocols for integrating face-to-face and digital treatment components across varied treatment settings. Lastly, considering the chronic nature of SMI and the increased potential for relapse or decompensation, research on bPT for SMI should encompass the investigation of short- and long-term treatment effects, as well as aspects of treatment safety.

In line with our study design, our literature review was broad in scope, encompassing a wide range of inclusion criteria in terms of study populations, bPT treatment approaches, and study designs. Given the nascent state of the field, this broad approach seemed fitting and allowed for a preliminary overview of the literature. However, the resulting heterogeneity in the findings may reflect this broad approach. Future systematic investigations should consider a more narrow focus to validate our conclusions and build upon the systematic evidence concerning subgroups and categories of bPT treatments for patients with SMI. Additionally, it would be beneficial to include stakeholder and patient perspectives directly in future research endeavors.

Similarly, the inclusion of both controlled and uncontrolled studies may limit the findings of this review. Moreover, due to the absence of quality assessment standards for digital mental health interventions [82], we did not exclude studies based on the potentially inadequate quality of the digital applications utilized. In our review, we only identified 5 studies using mHealth interventions for bPT, which was surprising given the significant advancements and research dynamic in this field. This might suggest that our findings may not generalize to bPT with mHealth.

The treatment of SMI, which is both prevalent and burdensome for individuals and society, remains challenging [69,75]. In recent decades, psychotherapeutic interventions have emerged as evidence-based first-line treatments, even for patients within this group, yielding mild to moderate effects. Concurrently, the past 25 years have witnessed the development of a broad array of digital mental health interventions. These digital mental health interventions are designed to address treatment gaps, offering affordable, accessible, and scalable alternatives to traditional mental health treatments, and contributing to the overall improvement of mental health care [20,23,80]. In this review, we have summarized the literature on the feasibility, adherence, and effectiveness or efficacy of bPT for patients with SMI. Our findings confirm the feasibility of such treatments for SMI and underline the need for further research to establish the effectiveness of this treatment delivery mode.

Considering that SMI treatment often entails a long history of patient suffering and cost-intensive multidisciplinary care [29,83,84], the further development and implementation of bPT for SMI treatment presents a significant opportunity. This opportunity not only has the potential to further improve the effectiveness and cost-effectiveness of mental health care but is also particularly relevant for this population.