Suicide is the fourth highest cause of death among older adolescents, globally (World Health Organisation, 2021a). Adolescence is a unique developmental stage where risk-taking behaviours and mental health difficulties increase (Patton et al., 2016; Spear, 2000). Incidences of suicidal thoughts and behaviours (STBs) occur more frequently during adolescence (van Vuuren et al., 2020), including suicidal ideation (SI), which are thoughts, wishes and preoccupations with suicide (Harmer et al., 2022), and suicide attempts (SA), which are non-fatal acts with inferred or explicit intent to die (Goldsmith, 2002). As previously defined by Walsh et al. (2022), post-primary school-based suicide prevention (PSSP) consists of the following preventive strategies: universal (i.e. targeting the wider school population); selective (i.e. identifying at-risk groups); and indicated (i.e. targeting STBs in high risk adolescents) (Goldsmith, 2002); and also includes interventions targeting STBs as both primary intervention outcomes and in addition to other well-being outcomes (Walsh et al., 2022), in line with the emerging focus on upstream approaches which prevent STBs by targeting related factors (Wyman, 2014).

PSSP is a key strategy for preventing STBs among adolescents (Walsh et al., 2022; Robinson et al., 2018a). However, many review studies of PSSP interventions have identified studies of randomised and non-randomised designs containing independent control groups (Cusimano & Sameem, 2011; Pistone et al., 2019). Recognising the challenges to evaluating PSSP interventions using randomised controlled trial (RCT) study designs, Robinson et al. (2018b) included all study designs in their review of suicide prevention interventions in educational settings targeting STBs, similar to Balaguru et al. (2012). As such, there is need for a synthesis of PSSP interventions targeting STBs as both primary intervention outcomes and in addition to other well-being outcomes, in studies which employ wide-ranging study designs.

Additionally, there are long-standing challenges to translating PSSP effectiveness to practice (Kalafat & Ryerson, 1999; Singer et al., 2018), with intervention uptake inhibited in school settings (Reifels et al., 2022; WHO, 2018). Interventions which have worked in one time and place too often do not work across multiple settings (Damschroder et al., 2009; Fletcher et al., 2016), particularly in schooling contexts (Joyce & Cartwright, 2020). Research to practice gaps in PSSP can manifest as (1) lack of referral pathways for adolescents identified as at-risk through PSSP (Breet et al., 2021), (2) unsuitability of suicide screening instruments to school settings (Hallfors et al., 2006), (3) challenges engaging teachers and adolescents in school-based interventions (Stallard et al., 2013), and (4) difficulties accessing external facilitators and mechanisms to implement PSSP (Lindow et al., 2019). It is speculated that the ongoing challenges to implementing prevention interventions in schools are due to the fact that young people engage with these interventions in classrooms, which are nested in schools, which are further situated in wider communities and districts, resulting in complex implementation needs (Chiodo & Kolpin, 2018). The field of implementation science endeavours to provide an arsenal of knowledge and tools to address knowledge-practice gaps (Westerlund et al., 2019). Drawing on implementation science can identify how PSSP interventions are effective in particular contexts and how to promote PSSP uptake in practice (Breet et al., 2021).

A key tenet of effective school-based prevention is the consideration to contextual factors that influence intervention implementation (Chiodo & Kolpin, 2018). Realist approaches in intervention research consider context as a multi-faceted concept reflecting ‘what works, for whom, in what circumstances, in what respects, and how?’(Pawson & Tilley, 1997) and proposes that intervention effectiveness is likely impacted by contextual mechanisms (Fletcher et al., 2016). Key intervention and school-related contextual factors at individual-, school-, district-, state- and federal-levels (Domitrovich et al., 2008) have been seldom considered in understanding effectiveness and implementation requirements of school-based mental health interventions (Clarke et al., 2021; Lendrum & Humphrey, 2012; Walker, 2004). Recent reviews of school-based identification of at-risk youth demonstrate a lack of understanding of (1) intervention effectiveness across schooling contexts, (2) the impact of school-related factors on intervention effectiveness and feasibility, and (3) school personnel perspectives on intervention acceptability and feasibility (Pierret et al., 2020; Soneson et al., 2020).

Understanding the context in which interventions are evaluated and implemented in is fundamental to ensuring that evidence-based interventions are effective in practice (Damschroder et al., 2009; Singal et al., 2014). Decision-making surrounding the implementation of PSSP is rarely informed only by the evidence-base; contextual factors such as resources, school needs, policy and funding allocations are key drivers on the ground which impact the uptake of PSSP (Reifels et al., 2022). As such, highlighting intervention and contextual factors with respect to PSSP is important both theoretically (i.e. for understanding how and when intervention components achieve intervention goals) and practically (i.e. knowledge of the settings in which interventions are evaluated informs future implementation considerations) (Singal et al., 2014).

To our knowledge, no synthesis of intervention and contextual factors with respect to PSSP interventions exists, with most known PSSP reviews describing interventions (Breet et al., 2021; Cusimano & Sameem, 2011; Hofstra et al., 2020; Katz et al., 2013; Pistone et al., 2019; Zalsman et al., 2016), with the addition of who the intervention was delivered by (Balaguru et al., 2012; Robinson et al., 2018a; Surgenor et al., 2016) or school setting (Miller et al., 2009). Understanding important intervention and contextual factors in the PSSP literature will contribute to bridging the gap between research and practice in PSSP. In this present systematic review, the multi-level conceptual framework based on theoretically and empirically informed factors proposed to influence school-based preventative intervention implementation (Domitrovich et al., 2008) was used as a guiding framework to organise intervention and contextual factors in the PSSP literature, with the latter categorised on macro-, school- and individual-levels.


This review had three objectives:

  1. 1.

    To estimate the effectiveness of PSSP interventions targeting STBs as both primary intervention outcomes and in addition to other well-being outcomes, on STBs among adolescents, in studies which employ randomised and non-randomised study designs.

  2. 2.

    To highlight important intervention and contextual factors in the PSSP literature.

  3. 3.

    To investigate PSSP effectiveness with respect to intervention and contextual factors, in line with realist perspectives.


Study Design

The protocol for a larger systematic review and meta-analysis was registered with PROSPERO (CRD42020168883). A separate meta-analysis synthesised the effectiveness of twelve eligible cluster randomised trial (CRT) studies identified in this review, in line with best practice (Borenstein et al., 2011; Donner & Klar, 2002) and is reported upon elsewhere (Walsh et al., 2022). Appendix Table 1 outlines amendments to the protocol relevant to the present systematic review.

This review was conducted in accordance with PRISMA guidelines (Page et al., 2021) (see Appendix Table 2 for PRISMA checklist). Realist perspectives informed the guiding conceptual framework for this review (Rycroft-Malone et al., 2012), which prioritised important intervention and contextual factors in the PSSP intervention literature. The multi-level conceptual framework guided the organisation of contextual factors into three subcategories in our review, which Domitrovich and colleagues define as follows: Macro-level factors include broader community factors which can influence schools’ intervention implementation, school-level factors include factors pertaining to the school environment which can influence intervention implementation, and individual-level factors refer to characteristics of implementers of PSSP interventions. As end-user acceptability is a key factor in intervention effectiveness (Kazdin, 1980), individual-level factors were operationalised to include adolescents engaging with PSSP. Intervention characteristics were organised into core elements and delivery of the intervention proposed as part of the intervention model (Domitrovich et al., 2008). In line with narrative synthesis guidance by Popay et al. (2006), vote-counting numerated study patterns and a textual approach to synthesis outlined and combined study findings, to ultimately summarise PSSP effectiveness and enhance understanding of how intervention and contextual factors impact intervention effectiveness and implementation.

Search Strategy

The search strategy has been published elsewhere (Walsh et al., 2022). Searches of PsycINFO, Medline, Education Source, ERIC, Web of Science, and the Cochrane Central Register of Controlled Trials in February 2020 and again in January 2021 identified studies published from database inception. Screened studies’ reference lists were searched. The Participants, Interventions, Comparators, Outcomes and study design (PICOs) tool was used to appraise studies (Methley et al., 2014); ‘Participants’ included adolescents 11–19 years of age in post-primary school, which aligns with key definitions of adolescence (WHO, 2021b); ‘Interventions’ included PSSP interventions targeting STBs as both primary intervention outcomes and in addition to other well-being outcomes; ‘Comparators’ included no intervention, other intervention, wait-list control, status before, at different time periods during and postintervention; ‘Outcomes’ included STBs such as SA, SI and planning; and ‘study design’ included experimental study designs. Excluded studies were (1) unavailable in English, (2) non-peer-review publications, and (3) non-experimental design.

EW initially assessed study eligibility based on title and abstract, and eligible studies were then screened in full by both EW and JMM using the screening web-tool Rayyan (Ouzzani et al., 2016). EW and JMM reached consensus on full-text screening disagreements and Cohen’s kappa coefficient indicated almost perfect agreement between EW and JMM’s original decisions on full-text screening; κ = .953, p < .001. One of the three trials evaluated in one study (Poppelaars et al., 2016) contained a non-school-based intervention, which was excluded from the synthesis.

Data Extraction and Analysis

EW extracted the data fully, with study, intervention and school data extracted in duplicate by CB. Data items presented in Table 1 and Appendix Table 3 were organised by PICOs and Domitrovich and colleagues’ framework; macro-level factors included ‘Region’, ‘Policies and funding’, ‘Community–university partnerships’ and ‘Leadership and human capital’; school-level factors included ‘Mission–policy alignment’, ‘Decision structure’, ‘School characteristics’, ‘Classroom climate’, ‘Resources’, ‘Personnel expertise’, ‘Administrative leadership’, ‘School culture’ and ‘School climate and organizational health’; and individual-level factors included ‘Professional characteristics’, ‘Perceptions of and attitudes to the intervention’ and ‘Psychological characteristics’. Statistics were presented for STBs where there were significant differences across control and intervention comparators. Effectiveness was reported separately for studies including an independent control comparator and studies containing the same participants for control and intervention conditions. Additional publications were located when information relevant to the interventions was published elsewhere.

Table 1 Summary of study characteristics

Risk of Bias

Risk of bias was rated using the Cochrane Collaboration Risk of Bias tools for CRT, RCT and non-randomised design studies, in line with the tools’ guidance (Eldridge et al., 2016; Higgins et al., 2011; Sterne et al., 2016, 2019). The risk of bias tools included domains of bias related to missing data, results reporting, deviations from interventions and randomisation/experimental design. Amendments to all tools included omitting the ‘Risk of bias in measurement of the outcome’ Domain and question ‘Were participants aware that they were in a trial?’, as assessing the awareness of participants in a trial is unsuitable to PSSP research, and similar omissions to risk of bias assessments have been made previously in reviews of PSSP (Robinson et al., 2018a). Risks of bias of eight studies were randomly assessed by two authors. Disagreements were resolved with consensus and there was adequate agreement between EW, and JMM (κ = .666, p = .008), and MH (κ = .625, p = .030).


Table 1 and Appendix Table 3 outline data from the included 28 studies evaluating 36 trials, published between 1991 and 2020. Trials include 46,979 participants aged 11–19 years. As outlined in Fig. 1, study designs include CRTs (15/28), RCTs (5/28), quasi-experimental (3/28) and within-subjects (5/28). All studies included mixed gender samples, aside from one female sample (Poppelaars et al., 2016). Twenty-one out of twenty-eight studies contained largely White samples or sampled participants from predominantly White countries.

Fig. 1
figure 1

PRISMA 2020 flow diagram for systematic reviews: identification of studies via databases and registers

Suicide Outcomes

Trial outcomes were conceptualised as SI (23/36); SA (18/36); behaviours (1/36); tendencies (1/36); threats (1/36); thoughts (1/36); suicide probability (2/36); planning (2/36); and suicide risk (1/36). STBs were measured only by single items in over half the studies (15/28), with four studies using single items from the Youth Risk Behavior Survey (YRBS). The Suicide Probability Scale (SPS) measuring self-reported hopelessness, hostility, negative self-evaluation and SI, was administered in two studies and was the most commonly validated questionnaire used. There was considerable variability in both categorical and continuous statistics reported, limiting the amenability of findings.

Among the 29 trials comparing an intervention condition with an independent control condition (i.e. randomised and quasi-experimental design studies), 13 trials reported statistically significant differences in STBs, with all but one trial (Shinde et al., 2020) demonstrating STB reductions in comparison to controls at postintervention. Reported follow-up time periods for the 12 interventions demonstrating STB reductions ranged 2/3 weeks–2.5 years (median = 6 months). Four out of twelve of the trials demonstrated that significant STB reductions were sustained at a minimum of 12 months postintervention. An odds reduction in the intervention condition of 64% (p < .05) in SA at 3 months postintervention (Schilling et al., 2016) was among the largest reduction in STBs observed postintervention. In terms of longer term effects, there were 55% and 50% lower odds of SA and SI, respectively, at the 12-month follow-up of the Youth Aware of Mental Health (YAM) intervention among + 4200 adolescents (Wasserman et al., 2015). The proportion of effective PSSP interventions varied across study design: CRT trials yielded the lowest proportion (6/20), followed by RCTs (4/8), quasi-experimental designs (2/3) and within-subject designs (5/7).


Core Elements

Trials contained the following intervention components: universal (18/36), selective (10/36), indicated (15/36). Trials with universal components demonstrated significantly lower STBs in comparison to independent control comparators (6/13), with a further 4/5 trials employing within-subjects designs effective in reducing STBs. Trials with selective components demonstrated significantly lower STBs in comparison to independent control comparators (4/8), with a further 2/2 trials employing within-subjects designs effective in reducing STBs. Trials with indicated components demonstrated significantly lower STBs in comparison to independent control comparators (4/13), with a further 2/2 trials employing within-subjects designs effective in reducing STBs. The Saving and Empowering Young Lives in Europe (SEYLE) study including three interventions (YAM, Profscreen and Question, Persuade and Refer [QPR]) was conducted across 10 European countries (Wasserman et al., 2015). The Signs of Suicide (SOS) intervention was the most common intervention evaluated across the included studies (n = 3) (Aseltine et al., 2007; Schilling et al., 2014, 2016), with all trials reporting significantly reduced STBs postintervention (p < .05).

Five studies detailed stakeholder intervention input, including adolescents and teacher/counsellors selecting activities (Shinde et al., 2020); adolescents identifying cultural adaptations (Robinson et al., 2016a); and adolescents sharing their perspectives on issues to inform the intervention (Eggert et al., 1995; Kim et al., 2020) and identifying case managers (Eggert et al., 1995); and adolescents, teachers and parents co-designing the intervention (Gravesteijn et al., 2011). Two other studies described community input in interventions (LaFromboise & Howard-Pitney, 1995; Silverstone et al., 2017). All but two of these studies (Eggert et al., 1995; Shinde et al., 2020) demonstrated significantly lower STBs postintervention.


Reported intervention duration ranged 1.5 hours–2 years (median = 7 weeks). Teachers were mostly involved in delivering intervention trials (14/36), and school counsellors and psychologists delivered 9/36 intervention trials. One intervention involved adolescents as peer leaders (Wyman et al., 2010). Parents were involved in 9/36 interventions. One study which evaluated the Strengthening Evidence base on scHool-based intErventions for pRomoting adolescent health (SEHER) intervention demonstrated statistically significant lesser odds of SA in the lay counsellor delivered-intervention, compared to the teacher-delivered intervention; odds ratio (OR) = .59 (.38–.93), p = .02.

Contextual Characteristics

Macro-level Factors


Studies were located in North America (14/28), Europe (5/28), Australia (4/28), Asia (4/28) and one study in South America. Three studies described rising regional suicide rates as a prompt for intervention implementation. It was hypothesised that regional differences in study response rates and participation in the SEYLE study were linked to attitudinal differences of principals and an environmental emergency occurring in one study site (Carli et al., 2013).

Policies and Funding

National and state policies relevant to the intervention were cited in 3/28 studies. One study described funding provision for participating schools (Robinson et al., 2016a).

Community–University Partnerships

Four studies described university and community partnerships facilitating interventions, with Silverstone et al. (2017) describing a multi-sectoral community partnership approach.

Leadership and Human Capital

School stakeholders co-ordinated with mental health services throughout two interventions (King et al., 2011; Silverstone et al., 2017), with the intervention evaluated by Silverstone and colleagues additionally involving the school district, education services, primary health care, social services, local services for family practices, and police services. In addition to school personnel, hospital representatives informed the intensive interpersonal psychotherapy for depressed adolescents at suicide risk (IPT-A-IN) intervention by sharing insights on adolescent’s mental health and co-ordinating follow-up mental health services (Tang et al., 2009). Three studies described assistance with intervention design and delivery to ensure cultural appropriateness of interventions and three further studies detailed intervention technical support.

School-Level Factors

Studies did not report characteristics referring to ‘Mission–policy alignment’, ‘Personnel expertise’, ‘Decision structure’, ‘Administrative leadership’, ‘School culture’ and ‘School climate and Organizational health’.

School Characteristics

Reporting of school characteristics including location, size and gender profile varied considerably (see Appendix Table 3). Only two studies investigated variation in effectiveness across schools. A non-significant intervention impact on peer leader referrals was observed in smaller, rural schools, but not in metropolitan schools (Wyman et al., 2010). Orbach and Bar-Joseph (1993) reported that a special education class found the intervention to be too exposing, while mainstream classes wanted to extend the intervention. There were reductions in STBs across both mainstream and special education classes.

Classroom Climate

The study evaluating the IPT-A-IN described participant stress and worry related to attention from teachers due to participating in the study and their parents’ receiving information about their participation (Tang et al., 2009).


One study described technical issues related to the lack of capacity of the school information technology system to host the online-based SPARX-R intervention, which limited students engagement with the intervention.

Individual-Level Factors

Studies did not report on ‘Psychological characteristics’ individual-level factors.

Professional Characteristics

Seven studies referred to implementer professional characteristics such as expertise, previous qualifications, and willingness to be involved in interventions. Having experience relevant to delivering the intervention was a criteria described in four studies.

Perceptions of and Attitudes to the Intervention

Six studies measured intervention acceptability using qualitative (3/6) and quantitative (3/6) methods. Anecdotal participant feedback on interventions was described in two further studies (Silverstone et al., 2017; Tang et al., 2009). All of the studies measuring intervention acceptability assessed participant treatment acceptability, with Shinde and colleagues additionally assessing acceptability of principals and teachers. Overall, studies reported satisfaction, acceptability and perceived usefulness of interventions, with interventions mostly perceived to be worthwhile in preparing adolescents to help themselves and their friends. Studies reporting on the SEHER, YAM and Reframe IT trials demonstrated adolescents’ satisfaction with elements of PSSP interventions which actively involved youth in intervention activities, including roleplay, discussions and online activities related to the intervention. Furthermore, adolescents appreciated elements of the SEHER and YAM interventions which facilitated youth input, such as speak out boxes and ability to direct the content of the programme and share opinions on the intervention (Shinde et al., 2017; Wasserman et al., 2018). Three studies investigated adverse effects associated with engaging in PSSP and supported PSSP to not be distressing for the most part. Bailey et al. (2017) demonstrated that suicidal thoughts were not associated with acceptability, but weak evidence (OR = .42, p = .047) indicated that preintervention suicidal thoughts were associated with perceiving the intervention as upsetting.

Information related to engagement with interventions was provided by study authors in 3/28 studies, including failure of several schools to return participant questionnaires on time (Perry et al., 2014), low levels of intervention implementation (Fekkes et al., 2016), and refusal to participate in data collection phases in the intervention (Vieland et al., 1991). Only one study measured and reported on the perspectives of those delivering the intervention as part of an multistakeholder investigation of intervention acceptability and feasibility (Shinde et al., 2020). Shinde and colleagues reported that teachers felt overwhelmed delivering the intervention while balancing other role duties (Shinde et al., 2020), with teachers and lay counsellors from lower-performing schools highlighting a lack of support from fellow teachers who did not approve of the intervention or consider it important. Key facilitators and barriers reported highlight the importance of considerations to teacher engagement, participatory nature of activities, and support from school management.

Risk of Bias Within Studies

Appendix Table 3 outlines overall and domain summaries of risk of bias for each study.

Non-randomised Studies

Overall summaries of bias were mostly rated as ‘Moderate’ (2/5) and ‘Critical’ (2/5) for within-subjects studies. Quasi-experimental studies were mostly rated as ‘Critical’ (2/3). ‘Moderate’ overall summaries indicate that the evidence emerging from the study is sound, and ‘critical’ overall summaries indicate that the emerging evidence is problematic (Sterne et al., 2016).

Randomised Control Trial Studies

Overall summaries of bias were mostly rated as ‘Some concerns’ (3/5), which indicates that there are some issues regarding the quality of evidence emerging from these trials (Higgins et al., 2016).

Cluster Randomised Control Trial Studies

Overall summaries of bias were mostly rated as ‘Low’ (7/15), which indicates that the evidence emerging from these trials is mostly at low risk of bias (Higgins et al., 2016).


Our systematic review of 28 randomised and non-randomised studies summarised the effectiveness of PSSP interventions for reducing STBs among 46,979 adolescents and highlighted key intervention and contextual factors with respect to PSSP interventions. Forty-one percent (i.e. 12) of the 29 trials comparing intervention and independent control group comparators reported statistically significant reductions in postintervention STBs, and 5/7 trials evaluating effectiveness of interventions using preintervention and postintervention participant scores reported a significant decrease in STBs over time.

Meaningful reductions in STBs are evident in this review, given that 4/12 trials demonstrated significant reductions in STBs in intervention groups at a minimum of 12 months postintervention. These findings are clinically relevant for researchers and practitioners alike given the (1) typically low base incidence rates of STBs, which makes it more difficult to detect statistical effects of interventions (Goldsmith, 2002), and (2) capacity to deliver PSSP interventions to groups of adolescents in classroom settings at one given time. Risk of bias was assessed using tools tailored for study design, and there was adequate agreement across authors in risk of bias assessment, strengthening the reliability of our conclusions. The proportion of trials demonstrating effectiveness of PSSP was lower in rigorous study designs, with CRT trials demonstrating the lowest proportion (30%), followed by RCT designs (50%), further followed by quasi-experimental designs (67%) and within-subject designs (71%). Risk of bias was considerably greater in non-randomised studies, with most overall bias assessments rated as ‘Critical’ for these studies, with CRT studies demonstrating the lowest risk of bias assessment. The inverse associations between both study quality and study design with PSSP effectiveness in our review highlights the importance of considering both study design and quality in conclusions of PSSP effectiveness. While we recognise that CRT design studies are difficult to implement in school settings, and ethical and methodological challenges may hinder the use of randomised designs in suicide prevention research (Robinson et al., 2018a), we nonetheless recommend that CRT designs are employed to produce high-quality evaluations of PSSP effectiveness.

Our findings differ to previous reviews which concluded that PSSP was effective in reducing STBs in 70% (Robinson et al., 2018b) and 7.6% (Miller et al., 2009) of 17 and 13 randomised and non-randomised studies, respectively. Our findings extend these review findings by an additional 11–15 studies, evaluating interventions with universal, indicated and selective intervention components, located in post-primary school-based settings, targeting STBs as both primary intervention outcomes and in addition to other well-being outcomes. This review is among the first to narratively synthesise this expanded focus of PSSP interventions. Interventions containing universal components were most commonly implemented across the included studies and mostly demonstrated effectiveness in reducing STBs. These strategies enable early detection and prevention, which overcomes challenges in identifying students at risk (Wasserman et al., 2012, 2021), and therefore, should be strongly considered as a key PSSP strategy.

Importantly, this is the first known review to comprehensively highlight and investigate intervention and contextual factors with respect to PSSP intervention effectiveness, using narrative synthesis. Overall, there was incomplete reporting of intervention and contextual factors. Our findings are consistent with a previous review identifying the lack of clear reporting of school characteristics across school-based mental health research (Cohen & Barron, 2021). The implications of the lack of reporting of contextual and intervention factors in our review are both theoretical; there is little understanding of the factors which underlie intervention effectiveness, which limits our understanding on whether intervention effects on STBs are dependent on the settings in which they are evaluated in, and practical; there is lack of guidance for practitioners regarding which PSSP interventions will suit their unique schooling contexts, which is likely to hinder uptake of PSSP in practice.

Few studies analysed intervention effectiveness with respect to intervention and contextual factors. Based on the findings that students from a special education class found the intervention too exposing (Orbach & Bar-Joseph, 1993), future research should prioritise investigating the suitability of PSSP interventions for special education contexts, to ensure that schools can identify appropriate and effective PSSP. The finding that significant intervention effects for peer leader referrals as part of the Signs of Suicide programme were observed only in metropolitan schools supports that there is a need to investigate PSSP intervention effectiveness across types of school. School- and classroom-level factors (e.g. resources and culture) can influence intervention effectiveness and implementation quality (Domitrovich et al., 2008), and there is evidence of association between school factors (e.g. religious ethos of school and school connectedness) and STBs (Ayyash-Abdo, 2002; Young et al., 2011). As such, we recommend that school-level factors, particularly pertaining to school type (i.e. location, gender, religious ethos, mainstream and special educational needs schools), should be described in PSSP research, to enable investigation of PSSP effectiveness with respect to types of schools. In practice, this will aid school personnel’s assessment of suitability of potential PSSP interventions to their own schools.

Given that individual-level factors can impact both PSSP effectiveness and implementation (Kodish et al., 2020; Miller et al., 2009), investigation of these factors should be prioritised in future PSSP research. A lack of knowledge and confidence in PSSP can lead to school personnel’s non-response and inappropriate response to suicide among young people (Pierret et al., 2020). Additionally, school personnel’s attitudes toward suicide prevention are associated with their behaviours around PSSP (Smith-Millman & Flaspohler, 2019), and cultural and moral beliefs about suicide prevention were identified as barriers to both client self-disclosure and clinicians’ inquiries about suicidal intent (Turecki & Brent, 2016). A recent review of school-based mental health interventions found that the most important factors for the sustainability of these interventions pertained to school staff perspectives, including availability of resources and capacity to support interventions (Moore et al., 2022).

Individual-level factors pertaining to end-users of PSSP interventions were rigorously investigated in few studies. User perspectives are particularly important for both intervention effectiveness (Taylor et al., 2011) and implementation (Proctor et al., 2011), and their investigation should be prioritised, particularly when adolescents engage with PSSP in groups as part of classrooms and are subsequently more susceptible to peer social influence (Prinstein & Dodge, 2008). Furthermore, student engagement with and input on school mental health interventions were identified as key facilitators to intervention engagement by school personnel, with school personnel more likely to engage with interventions when they perceived that students were engaging meaningfully with interventions (Moore et al., 2022).

Our findings contribute to the limited evidence on potential adverse effects of PSSP, showing mostly no distress experienced by adolescents engaging with PSSP. These findings align with meta-analytic evidence supporting the appropriateness of universal screening for STBs (DeCou & Schumann, 2018). As only three studies investigated adverse effects of PSSP in our review, further investigation of adverse effects in PSSP studies is needed, considering that concern of adverse effects remains a key barrier for progressing research on youth suicide prevention (Ballard et al., 2013; Thorn et al., 2020).


The lack of reporting and analysis of important intervention and contextual factors in PSSP has implications for ascertaining whether intervention effectiveness is attributed to PSSP interventions’ key ingredients, or mechanisms involving contextual factors and intervention components. If evidence-based PSSP interventions are to be useful to policymakers and practitioners interested in PSSP implementation, the consideration of intervention components as separate entities to context needs to be replaced by investigation of intervention effectiveness with respect to context (Moore & Evans, 2017), in line with realist perspectives (Pawson & Tilley, 1997).

Our review demonstrates that more than half of the included studies used single-item measures of STBs. Single-item measures of adolescent SI are commonly used due to the barriers to assessment, including insufficient time, training and measures (Gratch et al., 2022). Other reviews evaluating the effectiveness of alcohol prevention (Witt et al., 2021) and E-Health (Christensen et al., 2014) interventions similarly identified the use of both single- and multiple-item measures of STBs in studies. Although type of outcome measurement did not meaningfully moderate STB effect measures among 365 studies conducted over the last +50 years (Franklin et al., 2017), use of single-item measurements of STBs has been cautioned against, based on the risk of misclassification (Millner et al., 2015).

Lower income countries and disadvantaged schools were underrepresented in the included studies, yet adolescent suicide behaviours are higher in lower income countries (McKinnon et al., 2016). As such, there is a particular need for evidence of PSSP effectiveness in these contexts. Furthermore, three-quarters of samples in the included studies were largely White, but suicide risk peaks in adolescence and young adulthood in ethno-racially minoritised youth, in comparison to White youth (Alvarez et al., 2022), which highlights the urgent need to prioritise the evaluation of PSSP targeted at STBs among minority samples.

Our review identifies a gap in replicated effects of PSSP interventions across diverse contexts, which contributes to the lack of cross-cultural transferability of mental health promotion and prevention interventions for adolescents (Kuosmanen et al., 2019). Furthermore, there was little reference to national and state policies relevant to interventions in the included studies. It has been long recommended that researchers and programme developers understand the mission of those who are at the grassroots of school-based prevention at local, state and federal levels, to achieve mutual interests (Kazdin, 1980).

Few studies included community and university partnerships. A recent review revealed that multi-sectorial collaborations are lacking in suicide prevention more broadly, despite the encouragement of such collaborations by government and research funding entities (Pearce et al., 2022). Few of the included studies described the co-ordination between intervention facilitators and mental health services for at-risk students, which aligns with the common, but sometimes incorrect assumption that external referral trajectories are in-place in PSSP (Breet et al., 2021). Inter-sectorial involvement is a key component of successful adolescent health interventions (Patton et al., 2016), and it is recommended that schools implementing PSSP partner with community health providers, adolescent health clinics, or mental health agencies to ensure appropriate aftercare services (Hallfors et al., 2006). Consideration to these factors by researchers during the PSSP intervention design phase may entice more schools and districts to invest time and resources in PSSP research and practice.

It was not within the scope of our review to assess the support system and the standardisation of interventions (see Domitrovich and colleagues for further details). Finally, we had challenges locating English language versions of non-English screened studies (n = 6). Future syntheses should endeavour to synthesise non-English language PSSP effectiveness studies, particularly in the context of our review findings that the included studies contained studies with largely White samples or sampled participants from predominantly White countries.

Recommendations for Future Research

Our findings contribute to greater understanding of the PSSP research-practice gap, which aligns with key actions proposed by Reifels and colleagues to advance implementation research and practice in suicide prevention. Full reporting of key intervention and contextual factors with respect to PSSP will allow for integrative analysis of group- and individual-level data, which would strengthen evidence pertaining to causal impacts of PSSP interventions on STBs. PSSP should undergo realist evaluation using frameworks, such as The Context-Mechanism-Outcome-Configuration framework, which evaluates intervention effectiveness as a function of key mechanisms in particular contexts (Pawson, 2013). Multi-site evaluation of PSSP across diverse contexts will enable investigation of how PSSP effectiveness varies (Breet et al., 2021) and will result in knowledge of how, what, where and why PSSP interventions work, which is needed for greater understanding of implementation effectiveness across settings (Damschroder et al., 2009). Furthermore, employment of CRT designs in PSSP research is needed to explore school-level contextual factors with respect to intervention effectiveness. We also recommend that future PSSP studies use validated, multi-item measures of STBs which are youth-friendly and as concise as possible.

Who delivered the intervention, and perspectives of school personnel and adolescents on PSSP interventions were intervention and contextual factors that analytically and anecdotally varied intervention effectiveness. Where possible, researchers should harness school environments and important stakeholders (e.g. school personnel and adolescents) in their research design, similar to the procedure by Silverstone et al. (2017), which involved the school district in intervention design and adopted a multi-sectoral community approach to the implementation of PSSP. Schools offer a range of unique educational and interpersonal opportunities encompassing families, young people and the broader community, and involving important stakeholders can ensure that important voices are heard and reflected in services (Fazel & Hoagwood, 2021).

Our review demonstrates the importance of and critical need for greater investigation of individual-level factors related to stakeholders engaging with PSSP interventions. Understanding stakeholders’ perspectives is key to the success of school-based mental health interventions (Movsisyan et al., 2021; Stephan et al., 2007). More research focusing on understanding of stakeholder’s perspectives will contribute to ensuring that PSSP interventions are suitable, acceptable and compatible with school and community contexts, which will enhance the translation of PSSP research findings to practice.

Furthermore, our review identifies that adolescents had little involvement in designing, inputting on and delivering PSSP interventions. Despite youth voice inclusion being a hallmark of ecologically oriented public health strategies for school mental health, youth input has been absent in understanding wider school mental health approaches (Fazel & Hoagwood, 2021). Global suicide prevention policy advises the involvement of young people in the design of PSSP interventions (WHO, 2018), and therefore youth input in PSSP must be prioritised. Only one study included adolescents in the delivery of interventions. Peer-led PSSP interventions should be given greater consideration, as there are wide-ranging benefits to peers, recipients and communities in peer-led mental health interventions (King & Fazel, 2021; Patton et al., 2016; Wyman et al., 2010).

Preregistration of intervention outcomes is recommended to strengthen reliability of syntheses of PSSP interventions which primarily target non-STB outcomes, given that non-reporting of non-significant STB outcomes could result in non-inclusion of relevant studies in reviews, particularly when interventions primarily targeting non-STB outcomes may measure and report on various non-STB intervention outcomes.

Recommendations for Future Practice

Our review supports PSSP to be safe and effective for reducing STBs in adolescents. Furthermore, we recommend that schools prioritise PSSP with universal components. The YAM and SOS interventions demonstrated effectiveness in reducing STBs across European and North American contexts, respectively, and therefore should be considered by schools situated in similar cultural contexts. Additionally, the evaluation of the YAM intervention in North America has preliminarily demonstrated significant increases in help-seeking behaviours and improved mental health literacy, decreased mental health–related stigma (Lindow et al., 2020), and acceptability of the YAM by students, parents and school staff (Lindow et al., 2019), which supports suitability and effectiveness of the YAM intervention in North American contexts. PSSP approaches which align with educational and well-being goals and address suicide along with other health and well-being goals merit particular consideration by practitioners.

Notably, our review preliminarily demonstrates that PSSP effectiveness varies across school contextual factors. As such, practitioners should consider the aspect of the contexts in which PSSP interventions of interest are evaluated in, prior to selecting interventions for their own contexts. PSSP effectiveness differed across school type in our review, and therefore, practitioners should ensure that the PSSP intervention they plan to implement has shown effectiveness in similar types of schools to their own particular school setting.

Our findings identify teachers and school counsellors/psychologists to most commonly deliver PSSP interventions, which aligns with conclusions that school personnel are typically tasked with PSSP (Smith-Millman & Flaspohler, 2019), and that teachers are key PSSP gatekeepers (Nadeem et al., 2011). Teachers need to be supported in their role in PSSP to prevent burnout from balancing their competing demands in their profession (Nadeem et al., 2011). In our review, parental involvement in interventions mostly consisted of receiving updates on their child’s mental health and referrals, yet, key barriers to youth suicide prevention service use include parents’ and youth’s perceptions that problems (1) did not warrant services and (2) would be best solved privately (Gould et al., 2009). In addition, parents’ engagement in and communication about school-based mental health services were identified as key challenges reported by administrators at mental health clinics partnering with schools (Cummings et al., 2022). As such, greater parental and youth involvement in interventions should be prioritised in PSSP.


Our review makes an important contribution to the field of adolescent suicide prevention by supporting effectiveness of PSSP interventions evaluated using randomised and non-randomised designs, which target STBs as primary outcomes and in addition to other well-being outcomes. However, our conclusions of PSSP effectiveness in 17/36 randomised and non-randomised trials warrant caution, given the high-risk of bias identified in our review, particularly with respect to non-randomised studies. Our findings are among the first to demonstrate the need for greater reporting and analysis of intervention and contextual factors with respect to PSSP in randomised and non-randomised studies, to clarify how PSSP is effective and understand how interventions should be delivered, where and to whom (Fletcher et al., 2016; Singal et al., 2014; Taylor et al., 2011), which is critical for enhancing the translation of PSSP research to practice. Our findings demonstrate that PSSP effectiveness and acceptability vary across school type, but more research is needed to clarify these relationships. Notably, the perspectives of school personnel and adolescents on PSSP are particularly scant, despite their integral role in PSSP, and greater involvement of adolescents in designing, inputting on and delivering PSSP interventions is needed. Our findings have wide-ranging research, policy and programmatic implications pertinent to educational and clinical practice.