Introduction

The World Health Organization (WHO) reports that there are over 700,000 annual deaths by suicide worldwide [1, 2]. Globally, suicide is the fourth leading cause of deaths for youth and young adults [1], and specifically it is the second in Canada and USA [3, 4], and the first in Australia [5]. As such, suicide prevention is a top global health priority [6]. Suicide is preventable with timely, evidence-based interventions [2]. There are evidence-based interventions for suicide prevention, such as risk assessment, safety planning interventions, and follow-up care [7, 8], all of which are delivered in clinical settings, such as emergency departments. We also recognize the importance of population-level approaches to suicide prevention, such as gatekeeper training programs in schools [8]. However, clinical attention for suicide prevention cannot be overlooked, and individuals who suffer from suicide ideation must receive clinical attention [9, 10].

A review published in 2002 investigated 40 studies from the United States (US), United Kingdom (UK), Canada, Finland, and Sweden and found that 33% of individuals who died of suicide had contact with mental health services in the year before death and 20% in the month before death [11]. Not much has changed since then, and we continue to observe missed opportunities. In Canada, a study in 2014 examined 8,851 suicide deaths and found 50% of these individuals had visited an emergency department in the year before death, and one third had died within the month of discharge [12]. This speaks to a critical opportunity for suicide prevention in clinical settings, which will be the focus of this review.

Information and communication technology (ICT) [13] or eHealth [14] includes a wide range of digital tools such as internet, telemedicine, and mobile technologies. In this review, we refer to ICTs collectively as technologies, advanced multimedia, software programmes and/or telecommunications that allows efficient communication, management, storage, dissemination and exchange of information [13], and eHealth refers to use of ICTs for health [14]. There is a large number of ICT-based interventions for mental health, including suicide prevention strategies [15, 16]. For example, Rassy et al. identified 115 ICT-based interventions for suicide prevention, and they include web-based tools, online programs, and mobile applications [16]. Given the widespread use of technologies in this modern world, including mobile phones, ICTs have the potential to improve suicide prevention by removing geographical barriers and increasing access and availability of evidence-based interventions [16]. Additionally, ICTs may not replace clinical encounters, but it can be augmented to expand existing suicide prevention care.

There is a growing body of evidence for the effectiveness of ICT-based interventions for suicide prevention [15, 17,18,19,20,21]. For example, Witt et al. identified 14 online programs and mobile apps for self-management of suicide ideation and concluded with some evidence of reductions in suicidal ideation associated with using these digital interventions [20]. Arshad et al. also identified 22 clinical trials of ICT-based interventions for suicide prevention, which included online support tools for coping skills often derived from a well-established cognitive behavioural therapy and concluded with a favourable effect on reducing suicide thoughts [19]. Despite the clinical potential and a large number of available ICTs for mental health, clinical integration remains limited, and clinicians, service users, and hospitals continue to face challenges to achieve sustainable adoption [22,23,24]. It has been repeatedly reported that implementation of ICTs rarely moves beyond feasibility trials or initial adoption, and sometimes ICTs are abandoned [25].

Healthcare is a complex adaptive system, which is shaped by multiple, constant interdependent interactions and relationships [26, 27]. When complexities exist related to care settings or implementation challenges, the less likely ICTs are to be adopted and sustained [25, 28]. As such, research teams are required to move beyond traditional cause-and-effect thinking, embrace complexity, and examine dynamic processes inherent within. Specifically for mental health apps, there was a recent call for attention to complex contexts in which apps are being implemented [22]. It is critical to prospectively assess determinants of implementation and then strategically develop implementation strategies to match the contextual needs.

Efforts are needed to support clinical integration of ICT-based interventions for suicide prevention as well as their spread and maintenance to ensure that useful ICTs are reaching people who are in need. Currently, the literature on ICT-based interventions for suicide prevention describes their characteristics and/or evaluates their effectiveness in reducing suicide behaviours and risks, but not much is known about clinical integration of ICTs [15,16,17,18,19,20]. For example, it remains unknown how many of 115 ICT-based interventions for suicide prevention identified by Rassy et al. have been implemented in clinical settings [16]. Research has not yet comprehensively explored evidence on ICTs implemented in clinical settings and their implementation characteristics, including barriers and facilitators. Given the lack of successful clinical integration of ICTs [22,23,24], this review was needed as a first step to inform implementation efforts for useful ICTs for suicide prevention in clinical settings. Scoping reviews are suggested when researchers need to identify the types of available evidence and key characteristics related to a concept rather than to perform a meta-analysis to make practice recommendations [29, 30]. Furthermore, this review was needed to determine the range of studies before carrying out our future multi-phase project to develop and evaluate implementation strategies for a mobile app-based intervention for suicide prevention in clinical settings. As such, the current scoping review aimed to systematically explore the breadth of evidence on ICT-based interventions for suicide prevention implemented in clinical settings and then to characterize barriers and facilitators to implementation, as well as measures and outcomes reported in this body literature.

Research questions

To achieve the research aim stated above, this scoping review addressed the following questions.

  • 1. What ICT-based interventions for suicide prevention have been implemented in clinical settings?

  • 1.1. What are the reported barriers and facilitators to implementing these ICT-based interventions?

  • 1.2. What are the reported measures and outcomes of these ICT-based interventions?

Methods

This review followed the Joanna Briggs Institute (JBI) methodology [31, 32] and this report was prepared following the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist [33]. Our a priori protocol has been previously published [34]. We also searched PROSPERO, the Cochrane Database of Systematic Reviews and JBI Evidence Synthesis and Open Science Framework in June-July 2021 and identified no ongoing systematic or scoping reviews on the same topic.

Inclusion/exclusion criteria

Population

All clinicians both licenced and regulated practitioners were considered for inclusion in this review. Various healthcare professionals, such as physicians, nurses and social workers, provide direct care, and they are often collectively referred to as ‘clinicians’ [35, 36]. Additionally, unregulated clinical support team members and peer support workers were considered for inclusion because these roles are increasingly integrated into mental health care settings [37, 38]. This population criterion was relatively less significant than the context criterion because who implemented ICT-based interventions was often part of the context.

Concept

This review considered all types of ICT-based interventions for suicide prevention. Routine care (i.e., treatment as usual) provided via virtual platforms or telephones were excluded unless an ICT-based intervention was delivered to patients. Therefore, following the WHO’s definition for intervention, ICT-based interventions needed to assess, improve or promote service users’ health outcomes [39]. Suicide-related thoughts and behaviours is an umbrella term that refers to a spectrum of suicide ideation, communication, behaviours, and attempts with having any frequencies of suicidal thoughts with actual, undetermined, or no suicidal intent [40]. To be included, ICT-based interventions must be related to any sub-category of suicide-related thoughts and behaviours including non-suicidal self-injury (NSSI). Although NSSI is a unique phenomenon from actual suicide attempt, we decided to include it because NSSI is one of the risk factors for future attempt and suicide [41, 42]. According to WHO, there are three levels of suicide prevention: 1) Universal, 2) Selective, 3) Indicated. Universal strategies for suicide prevention work at a population level [1]. Selective prevention strategies target individuals who may not be currently expressing suicidal behaviours but are at a greater risk of suicide based on their characteristics such as age, sex and/or medical history [1]. Indicated prevention strategies target individuals who are presenting active risk or early signs of suicide potential, such as suicide attempt [1]. ICT-based interventions for any levels of suicide prevention were considered for inclusion. Suicide prevention interventions in this review included, but were not limited to, suicide risk assessment, safety planning intervention and lethal means restriction [7, 8]. Lastly, this review considered all ICT-based interventions that targeted patients of any age.

Context

All hospitals or clinical settings were considered for inclusion. For this review, a clinical setting was defined as any context where clinician-patient interactions occurred in real-time. Therefore, who implemented the ICT-based intervention was part of the context. To be considered for inclusion, ICT-based interventions needed to be implemented and initiated in clinical settings. Therefore, this review excluded crisis services because they are first initiated by patients, often in a public context, which we assumed to have different implementation characteristics compared to ICTs initiated in clinical settings. Additionally, there has already been a systematic review investigating effectiveness of crisis lines [43]. Self-support tools that patients can freely download from app stores or tools that involved self-referrals after reading public advertisements were also excluded as these were being initiated in non-clinical settings. Further to this, studies focusing on the development of ICT(s) without implementation were excluded. See Table 1 for summary of eligibility criteria.

Table 1 Inclusion criteria
Full size table

Search strategy

We worked with a health sciences librarian to develop a comprehensive search strategy to find relevant scholarly literature in several bibliographic databases. This scoping review followed a three-step search strategy outlined in JBI methodology [31, 32]. First, a librarian developed and refined a draft strategy in Medline, then analyzed text words and index terms contained in titles and abstracts of relevant articles as well as the subject headings. Second, relevant text words and index terms from the selected articles were used to develop a full search strategy. Third, the search strategy comprised of all identified keywords and index terms was adapted for all included databases. This required iterative steps of revising and testing, and the final search strategies were peer-reviewed by a second research librarian using the Peer Review of Electronic Search Strategy (PRESS) guidelines [44]. A librarian ran the search in the following databases on August 17–20, 2021: MEDLINE (Ovid), Embase (Elsevier), CINAHL (EBSCO), PsycINFO (EBSCO), Web of Science, and Library, Information Science and Technology Abstracts (LISTA). The selection of the listed databases was informed by consultation with a librarian, and they provide full coverage of literature likely to provide information specific to ICTs in clinical settings. All final search strategies are presented in Additional File 1.

Godin’s targeted Google search method [45] was used to locate additional eligible sources. First, we conducted ten unique Google searches with combinations of keywords and then reviewed the first ten pages of each search results to identify international and national health services websites. Second, we hand-searched relevant websites identified in the first step to find eligible sources. These two steps were carried out in incognito’ mode, which limited the impact of previous search history on new results. This Google search was complementary to the database searches to identify additional sources of evidence that our search strategy might have missed.

Types of sources

All research study designs were included (e.g., quantitative, qualitative, mixed methods). Although study protocols did not have empirical data, we included them to capture relevant details. Protocols tend to include details on interventions and implementation, such as intervention components, implementation plans, implementation blueprints, and discrete implementation strategies. Such information is useful characteristics to identify. Furthermore, by including protocols, we can reflect upcoming trends, such as the most used research designs in the upcoming years. Reference lists of relevant literature reviews, commentaries, and opinion papers were reviewed to identify additional primary studies that met our eligibility criteria. We also considered grey literature for inclusion, such as conference papers and reports from relevant health organizations. Sources had to be available in English, and no date parameters were applied.

Study selection

All identified citations were uploaded into Covidence [46] and duplicates were automatically removed by Covidence. Two reviewers (HDS, LS) independently screened titles and abstracts against the eligibility criteria. Next, relevant full-text articles were retrieved into Covidence [46], and the primary (HDS) and secondary reviewers (KD, LS) independently assessed them in detail against the eligibility criteria. Reasons for exclusion were recorded at the full-text screening phase and reported in the PRISMA flow diagram. Any discrepancies between the reviewers (HDS and LS or HDS and KD) at each stage were resolved either through discussion or by a third reviewer (KD or LS). Scoping reviews generally do not require methodological assessment [32], thus critical appraisal was not conducted.

Data extraction

We developed an extraction tool in Covidence to capture characteristics of the paper, setting, participating clinicians, implementation strategies, descriptions of ICT-based intervention(s), patient population, barriers and facilitators to implementing ICTs, and reported measures and outcomes. Three reviewers (HDS, KD, LS) first pilot-tested the extraction tool on three studies to identify any discrepancies or inconsistencies prior to data extraction. Each person extracted the same three studies independently using the extraction tool. We initially proposed to pilot-test the extraction tool on five studies. However, after testing on three studies for calibration exercise, the team agreed that all relevant data were captured, so we decided to start independent extraction without testing on two more studies. Minor changes to the original extraction tool were made, such as extracting the reported use of theories, models, or frameworks. The primary (HDS) and secondary reviewers (KD, LS) independently extracted data using Covidence [46]. Any conflicts in data extraction were resolved either through discussion between the two reviewers (HDS and LS or HDS and KD) or by a third reviewer (KD or LS). Final version of the data extraction tool is included in Additional File 2.

Data analysis

Following data extraction, we characterized extracted data using frameworks, typology, and taxonomy. First, identified ICT-based interventions for suicide prevention were categorized using a typology for e-Mental Health created by the Mental Health Commission of Canada (MHCC) [47]. Intervention descriptions were then characterized based on the suicide prevention interventions category adapted from Wilson [7] and Zalsman [8], and the WHO’s three levels of suicide prevention [1].

Second, we performed directed content analysis [48] using the Behaviour Change Wheel (BCW) [49, 50] and the Theoretical Domains Framework (TDF) [51] to map clinician-reported barriers and facilitators to implementing ICT-based interventions. They are comprehensive, evidence-based behaviour frameworks that capture internal and external influences of behaviour change. The Capability, Opportunity and Motivation – Behaviour (COM-B) model within the BCW explains behaviours by describing interactions between one’s capability, opportunity and motivation [49]. TDF is a 14-domain behavioural framework that expands the COM-B [51], so when used together, TDF allows for granularity of behaviour analysis [52]. Furthermore, benefits of using BCW and TDF for assessing implementation barriers and facilitators have been previously documented across healthcare disciplines [53,54,55]. Narrative descriptions of reported barriers and facilitators were mapped onto the COM-B and TDF.

Third, this review categorized reported measures and outcomes of ICT-based interventions for suicide prevention. Outcomes were categorized as either implementation outcomes or impact outcomes of the ICTs. Implementation outcomes were further categorized using Proctor’s Implementation Outcomes Framework: (1) Acceptability, (2) Adoption, (3) Appropriateness, (4) Feasibility, (5) Fidelity, (6) Implementation cost, (7) Penetration, and (8) Sustainability [56]. Impact outcomes or intervention outcomes were categorized into three levels: (1) Patient, (2) Health care provider (i.e., clinician), (3) Health system. Patient level impact was further categorized into patient-reported outcomes (PRO) [57], patient-reported experience (PRE) [58], and patient health outcomes (e.g., mortality) [59]. PRO comes from patients and often records patients’ view of their health status and condition [57]. Patients’ views of their own health can provide insight into the impact of an intervention [58]. In contrast to PRO, PRE measures patients’ perceptions and experiences of receiving care, providing insight into the quality of care during the intervention and the process of care [58]. Health care provider level outcomes include conceptual knowledge use (i.e., proximal practice change), instrumental knowledge use (i.e., observable practice change) [60], and other provider-reported experiences. Examples of conceptual knowledge use include levels of knowledge, and examples of instrumental knowledge use include rates of completed assessments [60]. Lastly, system level outcomes include resource utilization and economic outcomes such as cost effectiveness, and readmission rates [59]. Additional File 3 provides the full coding strategy with operationalized definitions.

The coding strategy was pilot tested on three studies by the primary reviewer (HDS), who has experience in qualitative research. Then second reviewers (LS, KD) who also have qualitative research experience reviewed the coded data generated by HDS to identify discrepancies and ensure consistency in coding. LS reviewed the coded data for barriers and facilitators and KD reviewed the categorized outcomes. LS and KD each reviewed half of the coded data for the characteristics of ICT-based interventions for suicide prevention. No changes were made to the coding strategy after pilot testing, and the primary reviewer (HDS) coded the rest of the data. Then the second reviewers (KD, LS) reviewed all coded data to verify HDS’s work. Any disagreements between the reviewers were resolved through discussion.

Data summarizing and reporting results

We charted the data in a tabular form that aligns with the review questions and aim. We also produced descriptive numerical summaries of the quantitative data (i.e., frequency counts) and graphical figures. We then provided narrative summaries to accompany these presentations and addressed the review questions and aim.

Results

Our database searches resulted in 6,008 citations. After duplicate removal, 3,659 citations remained for assessment against the eligibility criteria. After screening titles and abstracts, 242 citations remained for full-text review, and we identified an additional 6 relevant papers through Google searches and reviewing references of relevant reviews. This review included a total of 75 citations, describing 70 studies and 66 ICT-based interventions. See Fig. 1 for the PRISMA flow chart which includes the reasons for excluding full-text articles.

Fig. 1
figure 1

PRISMA flow chart

Full size image

Characteristics of included studies

Of the 70 papers, there were 52 research studies and 18 study protocols. There were five protocols of completed studies (i.e., protocol-study dyads) [61,62,63,64,65]. Seventy studies were a mix of experimental design (n = 22) [66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87], observational design (n = 12) [88,89,90,91,92,93,94,95,96,97,98,99], qualitative design (n = 3) [100,101,102], case study (n = 1) [103], quality improvement report (n = 1) [104], and feasibility/pilot trial (n = 31) [105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135] that served as a precursor to a larger study. These 70 studies originated from USA (n = 32) [69, 71, 73, 75, 76, 78, 81, 83, 88,89,90, 94,95,96, 98, 103, 104, 106, 109, 112,113,114,115,116, 118, 120,121,122, 124, 125, 132, 134], France (n = 8) [68, 72, 77, 86, 87, 91, 92, 105], UK (n = 8) [84, 97, 101, 107, 108, 131, 133, 135], Australia (n = 5) [67, 85, 117, 126, 128], Denmark (n = 5) [66, 82, 100, 127, 130], Canada (n = 4) [74, 93, 102, 111], South Korea (n = 2) [99, 119], Netherlands (n = 1) [129], Iran (n = 1) [80], Sri Lanka (n = 1) [79], Japan (n = 1) [123], Spain (n = 1) [70], and Portugal (n = 1) [110]. Studies took place most commonly in out-patient clinical settings (n = 43) [66, 67, 69,70,71, 73, 75, 76, 82,83,84,85,86,87,88, 90,91,92, 94, 96, 99,100,101,102,103, 105,106,107, 109,110,111, 114, 115, 117, 118, 123, 125, 127, 129,130,131,132, 134], such as emergency departments and clinics, then in-patient clinical settings (n = 14) [77,78,79,80, 93, 97, 98, 108, 113, 120,121,122, 128, 135], such as in-patient psychiatric units, and a mixture of both (n = 11) [68, 72, 74, 81, 89, 95, 104, 116, 119, 124, 133]. The remaining two studies were conducted in mental health hospitals but did not report specific clinical setting characteristics [112, 126]. Examples of involved clinicians included psychiatrists, nurses, physicians, social workers, and behaviour health clinicians, such as psychologists. Lastly, there was a lack of reported theories, models, or frameworks (TMFs) guiding research. Seven studies explicitly reported TMFs guiding their research [116, 118, 119, 124, 129, 131, 135], including the User-Centered Design Principles, Proctor’s Implementation Outcomes Framework, Theory of Planned Behaviour, Interpersonal Psychological Theory of Suicide, Integrated Motivational-Volitional model of suicidal behaviour, Medical Research Council, Process evaluation framework for analysis, and Normalisation Process Theory. Table 2 summarizes overall characteristics of included papers.

Table 2 Characteristics of included studies
Full size table

What ICT-based interventions for suicide prevention have been implemented in clinical settings?

This review identified a total of 66 ICT-based interventions for suicide prevention implemented in clinical settings. Based on the WHO levels of suicide prevention strategies, identified ICT-based interventions were used as universal (n = 4), selective (n = 10), or indicated (n = 53) strategies for suicide prevention. One ICT (i.e., Virtual Hope Box app) was used as both selective and indicated strategies in different studies [69, 88, 109]. While most ICTs targeted individuals who were at an imminent risk of suicide or were displaying early signs of suicide potential, fewer ICTs were used as selective strategies targeting at-risk populations, such as veterans, or patients living with human immunodeficiency virus, or cancer. A few ICT-based interventions were used as universal strategies aimed at population level, which may be explained by this review’s inclusion criteria being clinical context.

The 66 ICT-based interventions for suicide prevention served multiple functions; they were used for suicide screening and assessment (n = 22), safety planning (n = 20), lethal means restrictions and/or counselling (n = 3), discharge or post-discharge follow-up care (n = 27), therapy such as dialectical behavior therapy (n = 4), and additional resources such as wellness tips and journals (n = 18). Other (n = 12) functions of ICTs included reminders to appointments, care plans, and peer supports. Following the MHCC typology, most of the ICTs were categorized as computerized interventions (e.g., web-based tools), resources, and applications (n = 55), of which 11 were text messages, 10 were mobile applications (apps), and two were emails. Other types included telehealth and telemedicine (n = 16), wearable computing and monitoring (n = 1), virtual reality (n = 2), peer support through social media (n = 2), and a robot (n = 1) (i.e., chatbot). Table 3 summarizes above characteristics of ICT-based interventions for suicide prevention.

Table 3 Characteristics of included ICTs
Full size table

Implementation strategies

Overall, there was a lack of reporting on the implementation strategies for the included ICTs. Of the 75 included papers, 31 reported implementation strategies, but the level of detail varied. Training clinicians (n = 15) was the most commonly reported implementation strategy for the new ICT, focusing on building new skills [62, 63, 70, 71, 75, 95, 101, 110, 114, 115, 121, 127,128,129, 134]. A few studies specifically reported using demonstration [101] and simulation methods for training [115]. Educational meetings or communication (e.g., phone, email) (n = 12) was the next common implementation strategy which provided clinicians with new information and/or instructions required for the ICTs [73, 74, 84, 93, 101, 104, 108, 114, 126, 128, 131, 135]. Education or training were sometimes accompanied by educational materials (e.g., written handouts or supportive tools like a pocket guide) (n = 6) [64, 73, 101, 110, 114, 131]. Training and education were made distinct in this review; training focused on building practical skills, whereas education focused on providing new information or knowledge. Eight studies reported collaborative initiatives with clinicians, Information Technology (IT) consultants, ministry, institutions and/or managers [74, 91, 94, 95, 114, 123, 124, 128]. For example, collaboratives initiatives involved nominating site staff as co-principal investigators [74], or consulting key stakeholders before the start of the study [123]. Six reported providing ongoing supervision for using the ICT [63, 67, 71, 72, 104, 127], of which one study specifically conducted audits and provided daily reports to unit managers and nursing leaders [104]. Three studies provided opportunities for clinicians to participate in discussion for improvement in the implementation of the ICT, contributing to iterative changes in the implementation process during the study [67, 95, 114]. Two studies reported tailored approaches to implementation; one created a new clinical workflow to ensure that the implementation was seamless and minimized interruptions by leveraging existing staff roles and processes as much as possible [132], and the other provided site-specific training [62]. Lastly, one study provided onsite technical IT support [104].

What are the reported barriers and facilitators to implementing these ICT-based interventions?

Overall, there was a general lack of reporting on barriers and facilitators to implementation. Nineteen studies reported several barriers and/or facilitators with a varying level of detail. Barriers and facilitators that were most frequently reported by identified studies were associated with physical (n = 12) or social (n = 10) opportunity within the COM-B/TDF. TDF domains for physical (i.e., external) opportunities include environmental context and resources, whereas social opportunities include the social influences, such as norms and cultural factors [51]. Internet instability [134], limited telephone lines [103], lack of patients’ access to smart devices [107], time limited nature of clinical settings [76, 82, 102, 131], and no access to research teams to troubleshoot technological issues [108] were physical barriers described in the included studies. Other physical barriers included administrative challenges such as hospital policy that did not allow patients to use smartphones in the in-patient settings [93]. Therefore, even if patients had their own devices, hospital policy or the discharge norms limited patients’ access and did not allow enough opportunity for clinicians to deliver the ICTs until the moment of discharge. This not only speaks to physical barriers (i.e., hospital policy), but also reflects social barriers of limiting ICT related interactions with patients [93]. Other barriers to implementing ICTs related to social opportunity included lack of engagement with clinicians in the study and lack of buy-in and support from the clinicians [128, 131, 132]. Some of the facilitators were the direct opposite of barriers. In contrast to lack of engagement with clinicians, positive working collaborations between clinicians and the research team facilitated the implementation process [62, 104, 108, 131]. For example, one study had a hospital staff member in the role of principal investigator at each study site [62]. Furthermore, leadership engagement, such as manager approvals for implementation, facilitated ICT implementation, and some managers insisted on circulating implementation information to clinicians via e-mail [131].

Reflective (n = 14) and automatic (n = 3) motivations were the next commonly coded barriers and facilitators in this review. Motivation encompasses all brain processes that direct behaviour [49]. This includes not just reflective motivation, such as goals, analytical and conscious decision-making that leads to behaviour, but it also includes autonomic motivation like habits and emotional responses [49]. Reflective motivation includes TDF domains of professional roles and identities, beliefs about consequences, beliefs about capabilities, optimism, intentions and goals [51]. Defining roles and responsibility attributes [108], perceived burdens, and uncertainties associated with ICTs [76, 82, 131] were examples of barriers noted among the reflective motivation category. For example, clinicians were worried about ICT devices being stolen or broken [108] and perceived that that the ICT may have a better fit in other, non-clinical settings such as schools [131]. Clinicians also did not appreciate the perceived burdens of implementing ICTs because introducing new ICTs possibly created new tasks, taking extra time in their usual clinical flow [76, 82]. When clinical settings included multi-disciplinary teams, clinicians were concerned about who should be responsible for the ICT, but identifying appropriate professional roles and having designated staff for the new ICT were reported facilitators [78, 104, 108]. For example, one study implemented caring emails as post-discharge follow-up care for suicide prevention and reported that the new task associated with this ICT could be reasonably done by existing hospital staff rather than hiring new staff [78]. Additionally, they reported minimal requirements for clinicians to manage the new ICT, which facilitated implementation [78]. In contrast to uncertainties around ICTs, perceived benefits and usefulness of ICTs were facilitators [102, 109]. Automatic motivation refers to the TDF domain of emotion [51]. Negative (“technophobia”) or positive outlook about the ICTs [102, 116, 131] were identified as barriers or facilitators.

Implementation barriers and facilitators related to psychological (n = 14) capabilities were the least frequently coded category. Psychological capabilities include one’s knowledge, memory, and ability to make decisions and regulate behaviours [54]. Identified papers reported barriers and facilitators related to the knowledge and skills about ICTs, awareness of necessary resources, and clinicians’ cognitive load. For example, having no manual or guidelines to instruct clinicians on how ICTs should be introduced to patients and used for suicide prevention treatment was a barrier [82, 107]. In contrast, training resources and education sessions were facilitators that helped to build clinicians’ psychological capabilities [104, 109, 116, 134]. Additionally, a few ICTs helped to decrease clinicians’ cognitive burden [116, 131]. A summary of the COM-B/TDF analysis can be found in Table 4, and a full breakdown of extracted and analysed data can be found in Additional file 4.

Table 4 Barriers and facilitators to implementing ICTs
Full size table

What are the reported measures and outcomes of these ICT-based interventions?

As shown in Fig. 2, studies reported PRO (n = 55), PRE outcomes (n = 31), and patient health outcomes (e.g., mortality) (n = 10). Examples of PRO included assessing patients’ suicide ideation, suicide risk, coping ability, depressive symptoms, and health-related quality of life using validated tools such as Beck Scale for Suicide Ideation, Patient Health Questionnaires, Columbia Suicide Severity Rating Scale, and Beck Depression Inventory. Examples of PRE outcomes included assessing overall experiences and perceptions of ICTs, patient satisfaction, engagement with ICTs using open-ended survey questions, Likert-scale surveys, written feedback, or interviews. Patient health outcomes such as mortality and adverse events often came from health administrative data, electronic health records, or insurance claim data. At health care provider-level outcomes, studies reported clinician experiences (n = 7), clinicians’ instrumental knowledge use (n = 4), such as number of documented referrals, and conceptual knowledge use (n = 1), such as professional knowledge about suicide. Thirteen studies reported health system-level outcomes such as readmission rates and medical costs. Additionally, eight studies specified usage data as an outcome of interest.

Fig. 2
figure 2

Reported outcome types

Full size image

Following Proctor’s definitions for implementation outcomes [56], studies reported feasibility (n = 20), acceptability (n = 14), fidelity (n = 10), and penetration (n = 1) of the ICTs. Feasibility outcomes included perceived compatibility of ICTs in the clinical settings or practicality of ICTs assessed by surveys, open-ended questionnaires, interviews or measuring the time required to complete the ICT-related module. Acceptability of ICTs was evaluated by user experience, perception, agreeableness, or satisfaction using surveys, open-ended questionnaires, or interviews. Fidelity outcomes included the completion of follow-ups and/or adherence to treatments using chart reviews or self-reported data. Penetration was measured by the proportion of people who attempted suicide and were enrolled in an ICT-based intervention (i.e., VigilanS) relative to all included samples of people who attempted suicide regardless of their enrollment. None of the studies reported adoption, appropriateness, implementation cost, or sustainability outcomes of implementation. See Table 5 for summaries of the outcomes of interest, outcome measures, measurement tools, and key results of the 70 included studies.

Table 5 Summary of the outcomes, measures, and key results
Full size table

Discussion

Summary of evidence

This scoping review describes characteristics of ICT-based interventions for suicide prevention implemented in clinical settings. In this review, we identified 75 papers that described 70 studies and 66 ICTs. Overall, the review findings provide detailed characteristics of the existing ICTs for suicide prevention implemented in clinical settings. We also identified common strategies for implementing these ICTs, related barriers and facilitators, as well as reported measures and outcomes of the included ICTs. The findings offer insights into how to better support the implementation of ICTs and highlight the important role of collaborative initiatives in providing both technical and social support to facilitate implementation of ICTs in clinical settings.

Characteristics of included studies

Most of the included studies were experimental designs and feasibility trials, and there were 18 protocols, indicating that many studies are currently underway. Despite the growing evidence in this field, we found a lack of qualitative evidence. This is a gap in the current literature, and future research should consider qualitative study designs to evaluate implementation and/or impact of ICT-based interventions for suicide prevention on patients, health care providers, and health systems. This is because clinical practice within hospitals is an example of a complex adaptive system [26, 27]. Evaluating and understanding implementation of ICTs in complex systems will benefit from using qualitative or mixed-methods designs because quantitative methods alone cannot capture the complexity inherent within the phenomenon nor can it unpack interplay of contextual characteristics that influence implementation and impact of ICTs. Efforts are needed to move beyond traditional effectiveness trials and better understand how and why innovations bring change in what context [136]. Qualitative research designs can facilitate benefits of unpacking contextual factors (e.g., barriers and facilitators) at multiple levels (e.g., individual, system) and answering complex questions [137] that are integral to moving ICTs forward. Moreover, qualitative methods alone or in mixed-methods designs can confirm, complement, or extend quantitative evaluation of effectiveness, providing explanatory knowledge [138].

Based on the paucity of TMFs identified in the include studies, future research should consider using TMFs to guide their study. Despite the clinical potential of using mental health apps, integrating these apps into routine practice is limited, partly attributable to a lack of theoretical foundations and rigour in research for implementation [23]. Future research can benefit from leveraging TMFs and qualitative and/or mixed methods designs to unpack the complexity and contribute to building a rich evidence base. Benefits of using established TMFs in research have been well documented. For example, TMFs can help researchers consider comprehensive list implementation outcomes [139,140,141]. Furthermore, TMFs can help researchers consider a complete list of determinants for implementation during the planning phase to maximize implementation success [139,140,141]. Implementation is a known determinant of intervention effectiveness [56], and as we continue to face challenges in moving ICTs beyond pilot trials, it is necessary to leverage TMFs to guide careful and purposeful implementation that accounts for the complex contexts in which ICTs are implemented [22]. This will ensure that implementation strategies are systematically selected to address barriers in the local context. However, it is difficult to know whether authors of the included studies in this review did not use TMFs or did not report TMFs. If it is a reporting issue, then researchers need to improve reporting on TMFs so we can learn how TMFs have been applied, build knowledge base, and modify TMFs as necessary.

Implementation of ICTs in clinical settings

Thirty-one studies reported implementation strategies and 19 studies reported barriers and facilitator. Despite the general lack of reporting details, useful insights about implementation supports can be drawn. Of the reported studies, education and training were the most commonly reported implementation strategies for the ICTs. This is consistent with the current literature for implementation practice and knowledge translation [142, 143]. Educational meetings and training workshops are less costly and more accessible to support implementation than complex strategies requiring organizational-level change [144]. Therefore, educational meetings and training workshops could have been feasible options. However, barriers related to psychological capabilities were the least frequently coded category in the included studies. It is important to note that improving clinicians’ level of knowledge and skills does not always lead to observable practice changes leading to successful implementation of innovations [145]. Therefore, we recommend strategically considering diverse types of implementation strategies, other than education and training, to target both clinician- and external-level barriers for a given context. Secondly, collaborative initiatives were the next commonly reported strategy for implementation identified in this review. While partnership approaches such as co-design are common for innovation development, people often think that implementing what has been designed is the responsibility of others [146]. This is not true; researchers can co-create changes in the workflow to support implementation [147]. We encourage researchers to continue to leverage collaborative initiatives within their studies as they can foster important relationships between knowledge users and researchers. This will allow researchers to focus on real-world needs and facilitating implementation efforts [148, 149].

Researchers need to consider the complex contexts in which apps are being implemented [22]. As such, reporting details of implementation plans are strongly encouraged to advance our understanding of implementation processes and context. During implementation, the influence of context, such as barriers and facilitators, and interactions between them, are necessary to explain how or why certain outcomes are achieved, as well as variations in outcomes across studies [150, 151]. Furthermore, implementation is a known determinant of intervention effectiveness, and barriers can significantly reduce the effectiveness of an intervention [56]. Not knowing contextual influences may limit the generalizability of study findings to different settings. In response to the general lack of reporting details identified in this review, we encourage future studies to consider Proctor’s recommendations for specifying and reporting implementation strategies [152] and the Expert Recommendations for Implementing Change (ERIC) taxonomy for implementation strategies [153]. Furthermore, considering the iterative nature of the implementation process, any changes to original implementation plans are also encouraged to be reported. Future studies can consider the Framework for Reporting Adaptations and Modifications–Enhanced (FRAME) to guide the reporting of adaptations and modifications to the design or delivery of an intervention [154].

It has been reported that researchers are faced with challenges of selecting implementation strategies [155]. Furthermore, implementation strategies have often been mismatched to existing barriers [156, 157]. For example, a review of 20 quality improvement studies found that many studies utilized clinician-oriented (individual-level) strategies, such as education, to address organizational-level barriers [156]. Similarly, the current review identified that the three most reported categories of barriers were related to physical opportunity, social opportunity and reflective motivation, and examples included poor internet connection, busy clinical settings, lack of buy-in from and engagement with clinicians, and perceived uncertainties around ICTs. However, the most reported implementation strategies were education and training support, all of which cannot address the barriers stated above. This is an example of missed opportunities and an area for future research efforts.

Guided by the BCW, we can identify intervention options that can address the barriers identified in this review. To overcome physical opportunity, Training, Enablement, Environment Restructuring, or Restriction are recommended [49]. To overcome social opportunity, Restriction, Environment Restructuring, Modelling, or Enablement are suggested [49]. The use of evidenced-based theories like the BCW can improve the selection of implementation strategies and subsequent integration of ICTs in clinical settings [139, 141]. Additionally, clinical practice within health systems as well as human behaviour are complex; it is not individual factors that facilitate implementation of a new innovation, but the dynamic interaction between them [28, 158]. Nonetheless, the BCW accounts for interactions between both internal (i.e., capability, motivation) and external (i.e., opportunity) factors that influence behaviour change [49]. Use of behaviour change theories will not downgrade the complexity, but rather it can help researchers organize complex data in a comprehensive way that is also accessible to work with. As such, we recommend future studies to use TMFs to guide the selection of implementation strategies to overcome existing barriers.

Consistent with the current review findings, other external barriers associated with implementing ICTs are related to limited access to ICTs and internet, and digital literacy skills [159]. Despite the widespread use of mobile phones, a phenomenon called the ‘digital divide’ can occur due to social equity factors such as education, income, age, and urban/rural residence [160,161,162]. Digital divide refers to inequities in accessing and using ICTs as well as associated outcomes of using ICTs [162]. To prevent digital divide amplification and to avoid unintended harm, implementation efforts for new innovations must account for digital equity considerations [163]. However, very few included studies considered equity concerns and provided patients with ICT devices [74, 101, 114, 134], free data plans [111], or options for alternative ICTs (e.g., email instead of texts) as per patients’ preferences [71, 133]. In contrast, several studies made ownership of ICT devices as one of the inclusion criteria [61, 69, 78, 82, 85, 88, 102, 105, 111, 113, 120, 128, 129, 131, 135], and one study excluded participants who reported difficulty using a computer [117]. This is a critical area of future efforts for minimizing the digital divide. Van Dijk [164, 165], and Selwyn [166] recommend addressing the digital divide through assessing patient ICT access, use, competence, and reasons for divided outcomes. As many ICTs are rapidly being adopted and implemented for suicide prevention, this review identified a lack of attention to equity-related considerations in the current literature. This highlights a critical direction for future research, as efforts are needed to prevent digital divide amplification and avoid unintended harm while advancing ICT use.

Reported measures and outcomes

We identified that studies of ICT-based interventions for suicide prevention reported implementation outcomes and/or interventions’ impact on patients, clinicians, and/or health systems. Most studies reported patient-level outcomes, such as suicide risk and behaviours, and implementation outcomes of feasibility. However, no studies reported long-term outcomes of implementation such as sustainability. This is a gap in the current literature, and future research should consider assessing long-term outcomes, or at least should consider sustainability potential beyond feasibility. The end goal of implementing new innovations in clinical settings is routinization, achieving seamless integration of ICT use in routine clinical flow [167]. Despite the promising clinical benefits of ICTs for suicide prevention, clinical integration remains limited [22,23,24]. This problem is consistent across ICTs in general. It has been repeatedly reported that ICTs are not fully implemented, not moving beyond pilot trials, or being abandoned [25, 168]. To move beyond initial adoption of useful ICTs, we encourage future research to consider sustainability outcomes early on. Proctor’s Implementation Outcomes Framework [56] and the Reach, Effectiveness, Adoption, Implementation Maintenance (RE-AIM) [169] are example tools to guide outcome selections related to implementation and sustainability of interventions. Several studies included in this review measured both intervention outcomes and implementation outcomes in one study [67, 76, 80, 84, 89, 91, 101, 106, 114, 115, 117, 118, 120, 122, 124, 126,127,128,129,130,131,132, 135]. Similarly, future research can benefit from leveraging effectiveness-implementation hybrid designs that have a dual focus of evaluating intervention effectiveness and implementation outcomes simultaneously [170, 171]. Hybrid designs are encouraged to move interventions to the real-world more rapidly because the traditional research approach of keeping efficacy, effectiveness, and implementation research separate and sequential slows down the process and overlooks complex contexts inherent within [170, 171].

Limitations

Several limitations may affect the interpretation and use of our review findings. Many papers lacked detail on the barriers and facilitator to implementation, which made challenging to categorize them into the three overarching domains of capability, opportunity, and motivation. We conducted directed content analysis of the barriers and facilitators, and we report the frequency counts of these barriers and facilitators. However, this may not be a complete list of barrier and facilitators to implementation. Additionally, the categories within the COM-B and TDF are not mutually exclusive; many barriers and facilitators interact with each other, and this is one of the underlining assumptions of the human behaviour [49].

Secondly, our search strategy was limited to papers published in English. This may partly explain our finding that most studies originated from North America and Europe. As shown in the Fig. 1, we excluded eight papers written in non-English languages. We also acknowledge that our search strategies may not have captured studies conducted in low and middle-income countries. As a result, this review does not reflect evidence of ICTs for suicide prevention written in non-English languages or low- and middle-income countries, possibly resulting in underrepresentation and/or underreporting of the authorship and the amount of literature.

Third, we did not include ICT-based interventions in non-clinical settings such as schools. There are many other ICT-based interventions for suicide prevention that exist beyond what is included in this review. Lastly, despite our comprehensive search strategy, which included varied terms to describe ICTs, it is possible that relevant literature was not captured. To mitigate this limitation, we used Google search as a complementary to locate additional studies that our search strategy might have missed. We believe that our final search strategies were sensitive enough to provide full coverage of relevant literature because many papers identified during the second step of Google search were already captured by our main database searches. It is also important to recognize the inherent limitation of Google searches related to reproducibility of results [172]. A researcher from a different country may receive different results with the same steps, which is why Google search was complementary to full search strategies and not used alone.

Conclusions

This scoping review provides a comprehensive overview of published literature on the ICTs for suicide prevention implemented in clinical settings. The findings revealed the most common types of ICTs for suicide prevention, including apps, text messages, and telemedicine. These ICTs were commonly used as a targeted strategy for suicide prevention and served multiple functions, including suicide screening and assessment, safety planning, and post-discharge follow-up care. Additionally, the findings revealed that the most common strategies for implementing these ICTs included education, training, and collaborative initiatives. However, barriers collectively influenced clinicians’ capability, opportunity, and motivation to implement ICTs for suicide prevention. Therefore, implementation strategies must be tailored and multi-faceted to target specific barriers in a given context in order to facilitate implementation efforts for ICTs in clinical settings. Along with the lack of qualitative evidence in this field, the lack of reporting of implementation strategies and related barriers and facilitators was an evident gap in this body of literature, highlighting the need for more explorative research and a call for better reporting. Additionally, the lack of theoretical frameworks identified in included studies encourages the use of established TMFs to guide future work. Lastly, the absence of sustainability outcomes and digital equity considerations identified in the current literature highlights a critical direction for future research.