Internationally there is a move toward using digital technologies in mental health care, including the development of an increasing number of Web-based therapy programs [1,2]. The reasons for this interest in electronic mental health commonly include accessibility, flexibility (in terms of when and where they can be accessed), lack of mental health care professionals to cover need, and cost considerations [3,4]. Randomized controlled trials of web-based therapies show promise, with efficacy reported to be similar to that of face-to-face interventions [5]. A systematic review of internet-based cognitive behavioral therapy [6], for example, found that this therapy has been tested for 25 different clinical disorders, with large effect sizes meeting the highest level of criteria for evidence for disorders such as depression, anxiety, and severe health anxiety.

Despite this interest, research on referral and uptake of web-based therapies in routine care is limited [7]. In a rapid review of literature on this topic, we found no studies set in hospital-based medical clinics and only 1 study in primary care, which reported extremely low rates of referral (14% of the expected number) and uptake (1% of those referred) [8]. A major implementation project in Europe, known as MasterMind, is studying web-based therapies in routine mental health care [9,10], but no data are published at this stage.

Nor is there a large body of evidence on effective processes for implementation of web-based therapies in routine clinical care [7]. A recent review of staff-reported barriers to and facilitators of hospital-based interventions of any kind (web-based or face-to-face) [11] identified 3 key domains that may affect implementation success—system (eg, workload and workflow, physical structure and resources, culture, communication, and external pressures), staff (eg, attitudes, understanding and awareness, role identity, skills, and confidence), and intervention (eg, ease of integration, validity and evidence base, safety and ethics, supporting resources)—with similar barriers likely in community settings [12]. Several conceptual frameworks have been developed to further explicate and guide health care implementation, including the Consolidated Framework for Implementation Research (CFIR) [13]. CFIR is composed of 5 major domains: intervention characteristics, the inner setting, the outer setting, characteristics of individuals involved in implementation, and the implementation process. As health professionals often act as gatekeepers to web-based programs, they are key stakeholders and may be very influential in determining whether web-based programs are disseminated widely. Thus, in this systematic review, we decided to focus on health professionals’ perspectives of web-based therapies, which falls within the fourth CFIR domain.

Several individual studies reported data on health professionals’ perceptions of web-based therapy programs, which revealed concerns about therapeutic alliance and the quality and effectiveness of programs [14,15], patient commitment and compliance [15,16], patient barriers such as internet literacy and access [16,17], and suicide risk [16]. However, we found no synthesis of these studies to provide an overall picture of health professionals’ perspectives to date. Thus, we sought to fill this gap by conducting a systematic review. Our research question was what are health professionals’ perceptions of the barriers to and facilitators of the implementation of web-based psychological treatments in routine care in health settings?

We registered the review with the International Prospective Register of Systematic Reviews (PROSPERO CRD42018100869).

We searched 5 bibliographic databases: MEDLINE, EMBASE, PsycINFO, CINAHL, and the Cochrane Library (Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Cochrane Methodology Register). The search strategy included terms relating to web-based approaches, psychological therapies, and views of barriers to and facilitators of dissemination.

Qualitative, quantitative, or mixed-methods studies exploring health professionals’ views on computer- or internet-based psychological treatment programs were eligible if the treatment contained a component designed to change psychological symptoms or behaviors either associated with a diagnosable mental health condition (using International Classification of Diseases, Tenth Revision, Diagnostic and Statistical Manual of Mental Disorders [Fifth Edition], or equivalent) or secondary to a physical health condition such as cancer. This could include anxiety, depression, substance misuse, suicidality, insomnia, complex pain, fatigue (eg, chronic fatigue syndrome or cancer-related fatigue), or medication adherence. Studies of views on all models of web-based therapy, from entirely self-directed to blended with face-to-face care, were eligible. Additionally, the patients targeted (and the health professionals) could be situated either in a hospital or clinic setting or in the general community.

We included only studies investigating views on web-based programs for adults (>18 years of age); we excluded programs targeting adults who were completing the program on behalf of someone else, for example, parents completing a program to help their children with anxiety or behavioral problems.

We included studies of the views of all health professionals involved in referring, triaging, or providing psychosocial care to patients with the above conditions (eg, doctors, nurses, psychosocial health professionals, allied health practitioners). We excluded nonclinical professional groups who do the above (eg, schoolteachers, religious advisers).

We limited included studies to peer-reviewed articles, excluding conference abstracts. We searched studies published from 1986 (when the first known digital mental health site, Ask Uncle Ezra, at Cornell University was launched) to the end of October 2019.

Titles or abstracts, or both, of articles retrieved during the search strategy or from reference lists were screened by 1 reviewer to identify studies that met the inclusion criteria described above. As an initial step, 10% of the titles or abstracts and full-text articles were independently reviewed by a second reviewer to confirm inclusion and exclusion decisions. Following this, any titles or abstracts or full-text articles where the reviewer had any uncertainty about the decision was put aside and discussed with a second reviewer. Any continuing uncertainty was resolved through retrieving the full text of the study or discussion with other members of the authorship team. We hand searched reference lists of included articles for relevant articles.

We extracted information from articles meeting the inclusion criteria, with 2 team members completing extraction for each article. Any uncertainty was resolved through discussion with additional members of the review team. Data extracted were author, country, study design, sample characteristics, web-based therapy characteristics, measures, results, and limitations.

We assessed the selected articles for risk of bias using checklists appropriate to quantitative and qualitative studies from Kmet et al [18] and to mixed-methods studies from Pluye et al [19]. All articles were graded by 2 team members and discussed with the author team if there was any ambiguity in grading.

We analyzed the articles using a 3-stage synthesis process that combined qualitative, quantitative, and mixed-methods studies comprehensively and rigorously to answer the review question, following the methods of Thomas et al [20]. First, we summarized quantitative data using descriptive statistics. Second, we synthesized qualitative data by themes [21]. All authors coded 2 qualitative studies line by line to develop a draft coding tree and themes, which we then refined after coding 2 additional studies. We then applied this coding structure to the remaining qualitative studies, with emerging themes created and discussed by all authors. Similarities and differences between codes and their content were examined to develop higher-order themes. (See Multimedia Appendix 1 for the final coding tree.) Third, we used a matrix to compare the qualitative, quantitative, and mixed-methods findings from the first 2 stages to answer the review research question.

We initially identified 4210 potentially eligible articles from the search, with 2545 remaining after deduplication. Figure 1 shows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram, with 29 articles meeting full inclusion criteria: 14 qualitative, 11 quantitative, and 4 mixed methods. Most articles were published since 2009 and reported results from the United Kingdom (n=5), the United States (n=6), Australia (n=4), Canada (n=4), Europe (n=10), and Israel (n=1). A total of 16 studies specified the professional background of the included health professionals (psychologists, n=12 studies; social workers, n=9 studies; general practitioners, n=4 studies; nurses, n=3 studies; and psychiatrists, n=2 studies), and the remaining 13 studies used general terms (eg, “psychotherapist” or “counsellor” or “practitioner”). Where studies included both health professionals and patients, we included only data from the health professionals in the review. See Multimedia Appendix 2 for study characteristics. All included articles were of acceptable quality, with reflexivity and use of verification procedures to establish credibility the most poorly done (see Multimedia Appendix 3 for quantitative studies, Multimedia Appendix 4 for qualitative studies, and Multimedia Appendix 5 for mixed-methods studies).

We identified 6 themes, which mapped well to the CFIR: therapy factors, organizational (or inner setting) and system (or outer setting) factors, patient factors, health professional factors, the therapeutic relationship, and models of care.

Despite good evidence of efficacy in randomized controlled trials [5], the potential of web-based psychological therapies to increase access to mental health care has not been well realized in nontrial settings to date [49]. This is, to our knowledge, the first synthesis of quantitative and qualitative data on health professionals’ perceptions of web-based therapy for mental health conditions (both primary and secondary to physical disorders). We found that health professionals perceived many barriers to routine use of web-based therapy and had strong views about appropriate models of care. While health professionals’ concerns may not all be evidence based, health professionals are in many cases the gatekeepers to referral. Therefore, addressing their concerns is a key strategy to improve implementation, alongside ongoing education of health professionals so that they are familiar with the content, processes, and evidence base of web-based therapies and are comfortable with incorporating them into their model of care. The importance of thinking carefully about models of care specifically in the context of web-based therapies has not been addressed adequately in the literature to date.

Health professionals’ concerns corresponded closely with the 5 domains of the CFIR [13], with issues raised pertaining to intervention characteristics (eg, flexibility), the inner setting (eg, managerial support), the outer setting (eg, data security and privacy), characteristics of individuals (eg, their health and internet literacy), and the implementation process (eg, health professional education). Furthermore, they also accord with the much broader findings of a systematic review of reviews concerning factors that influence implementation of electronic health [50]. We found, as did the review of reviews, that the factors influencing implementation are multilevel and complex, with no single factor acting as a key barrier or facilitator.

This review suggests that for web-based treatments to maximize their acceptability to health professionals, they should incorporate the following features.

This review identified only 29 articles examining health professionals’ perspectives on web-based therapy. These articles varied considerably in terms of the population of health professionals included as participants, as well as the model of care that was being examined (eg, blended, self-directed), and are unlikely to give a complete picture. The main limitation in the literature so far is the inclusion of a very small proportion of health professionals who are experienced in web-based therapy, with most health professionals responding to questions in the absence of direct experience.

This is, to our knowledge, the first systematic review of health professionals’ perspectives on web-based therapy and provides a synthesis of the research to date.

The findings of this review echo the factors found in a recent article documenting the experiences of 2 successful online clinics [49], which identified the importance of considering the patient, therapists and their training, comprehensive organizational and systemic support, and the place of web-based therapies in models of care. These results have important implications for the implementation of web-based therapy, emphasizing that important preparatory work is required if implementation is to be successful, in the way the intervention is designed, the training and support health professionals receive, and the engagement of senior management in supporting the transition to a different model of care.

Further research is needed to examine the acceptability of specific models of care [2], which may vary in level of therapist support, timing of the intervention (waitlist, posttherapy, etc), positioning in stepped care, or the way it is incorporated in routine care. As health professionals gain experience with web-based therapies, it will be important for their perspectives to be sought again, to elicit perspectives grounded in real interactions and patient feedback. Interventions to educate and support health professionals in using web-based methods will need to be developed and evaluated. Finally, large-scale implementation studies that document the implementation strategies employed, uptake and retention rates, and effectiveness outcomes will be needed to provide a solid evidence base for this therapeutic approach moving forward.