The systematic review protocol was registered and published with the International Prospective Register of Systematic Reviews (PROSPERO) (registration number: CRD42023384321). The review utilized the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for reporting the findings (See Supplementary file 1).

Population (P): Studies involving participants aged 11 to 19 were included, covering a broad adolescent range from pre-adolescents (11–12 years) to adolescents (13–19 years). There were no geographical restrictions, encompassing studies worldwide to ensure a comprehensive understanding of preventive mental health interventions across diverse cultural contexts.

Interventions (I): The interventions assessed in the reviewed studies targeted preventive mental health interventions to address depression and suicidal tendencies among adolescents. These interventions encompassed a range of approaches, such as psychotherapy, CBT, mindfulness-based techniques, group therapy, educational programs, and other structured interventions, aimed at enhancing mental well-being and alleviating depressive symptoms and suicidal tendencies among adolescents.

Comparator (C): The review examined randomized controlled trials (RCTs) that compared intervention groups to control groups. Control groups may consist of no treatment, standard care, or an alternative intervention.

Outcomes (O): The primary outcomes considered included measures of depression and suicidal tendencies. The secondary outcomes included changes in mental well-being, reductions in depressive symptoms, and the mitigation of suicidal behavior.

Exclusions: Publications in languages other than English, pilot studies, study protocols, reviews, commentaries, and letters to the editor. Additionally, studies specifically addressing post-COVID-19 depression, psychotic depression outcomes, and hospital-based treatments were omitted.

The exclusion of non-English publications stemmed from the review team’s linguistic limitations and the costs associated with translation services, leading to a potential language bias. This decision aimed to maintain the review’s feasibility within available resources while acknowledging the bias limitation. To address this limitation, a comprehensive search of English-language databases known for their extensive international research coverage was conducted.

Although valuable in the early intervention stages, pilot and feasibility RCTs were excluded because they focus more on procedural feasibility than on efficacy outcomes and often lack the statistical power needed for reliable efficacy data. This review focused on studies that were sufficiently robust to provide conclusive data on the efficacy of preventive mental health interventions for adolescents.

Studies on post-COVID-19 and psychotic depression were excluded due to their distinct etiological mechanisms and specialized treatment requirements. Including them could compromise the generalizability of our findings, as these conditions demand tailored interventions rather than the broad preventive strategies analyzed for general depressive populations. The focus was on non-hospital settings to emphasize interventions accessible in community or school environments, aligning with the review’s aim to explore preventive strategies rather than treatments for severe conditions necessitating acute care (Supplementary file 2).

The review employed a systematic literature search strategy to identify relevant studies on mental health interventions aimed at reducing depression and suicidal tendencies in adolescents. The search encompassed databases including PubMed, CINAHL, Embase, and APA PsycINFO from 1 January 2011 to 15 December 2023 (n = 1,093). To update the review with recent literature, a revised search was conducted from 16 December 2023 to 23 February 2024, yielding 117 citations (Supplementary file 3). We selected 1 January 2011 as the starting point for our literature search to focus on the most recent evidence reflecting advancements. This date corresponds with a notable shift in the availability of high-quality RCTs identified during a preliminary review. Spanning a decade, this timeline facilitates a comprehensive analysis of contemporary evidence. Previous seminal reviews (Das et al., 2016; Fazel et al., 2014a; Fazel et al., 2014b) have addressed research prior to 2011. To prevent overlap and build on this existing knowledge base, we focused on newer studies while also reviewing the references of earlier works to include any crucial prior research. Figure 1 illustrates the updated search process, from database querying to the final inclusion of citations, as indicated by an asterisk in the PRISMA Flow diagram.

By utilizing Colandr, a web-based open-access reference manager, we systematically stored, screened, and shortlisted studies based on predefined inclusion and exclusion criteria. From the identified 1,210 records, we removed 279 duplicates, leading to the review of 931 citations. During the title and abstract screening, we excluded 860 studies that did not meet the preset criteria. Two independent reviewers assessed the titles and abstracts, resulting in the shortlisting of 71 articles for full-text screening. Following this comprehensive process, 13 studies were ultimately included in the review. The PRISMA flow diagram (See Figure 1) details the process through which the studies were identified, screened, and included in this review. The full texts of the included citations were retrieved and saved in Colandr for further data extraction, while the excluded studies were coded to indicate the reason for their exclusion. Any disagreements regarding the selection of citations between the two primary reviewers were resolved by a third reviewer and research advisors.

Upon the identification of eligible studies, the first and second authors collaborated to prepare a literature summary table that facilitated the systematic data extraction for review. Two reviewers then independently utilized this summary table to meticulously extract relevant information from each study, ensuring a solid foundation for our analysis. This comprehensive data extraction process included authors, country of origin, year of publication, quality appraisal score, sample size, gender, age, setting, specifics of the intervention, follow-up intervals, employed tools/instruments/measures, primary outcomes, and data regarding effect estimates and the intervention’s effectiveness compared to a control group, as well as statistical significance.

In addition to these critical elements, special attention was given to the extraction of effect size data, including reported effect sizes (e.g., Cohen’s d, correlation coefficients), means, standard deviations, and confidence intervals. This approach allowed for a uniform assessment of the interventions’ effectiveness across studies, accommodating the diverse reporting formats found in the literature. In cases where effect sizes were not explicitly reported, such as in studies providing only means and standard deviations, Cohen’s d was calculated to standardize the effect size measurement across studies. This approach was exemplified by our treatment of specific studies such as Bernal et al. (2019), which reported results using correlation coefficients, thereby facilitating effect size calculation. This detailed and methodical extraction process fostered a nuanced understanding of intervention effectiveness, ensuring a comprehensive and insightful analysis (Bernal et al., 2019).

For this systematic review, the assessment of RCTs was rigorously conducted using the Joanna Briggs Institute (JBI) tools, which are widely recognized in academic circles for their robustness and validity in evaluating research quality (Ma et al., 2020). Two independent reviewers meticulously carried out the evaluation process, analyzing each study based on its objectives, sampling strategy, measurement accuracy, and analytical robustness.

The assessment framework included a comprehensive risk of bias analysis, detailed across 13 specific criteria (Supplementary file 4). These criteria covered five critical domains of RCT methodology: the randomization process, adherence to intended interventions, management of missing outcome data, accuracy of outcome measurement, and selection of reported results. Each criterion was assigned a point, with the total score providing a quantifiable measure of the study’s methodological soundness. A higher score indicated superior research quality (Porritt et al., 2014; Ma et al., 2020). In instances of disagreement, reviewers reached consensus through thorough discussion, ensuring a balanced and objective quality appraisal.

In this systematic review, the primary goal was to synthesize evidence regarding the effectiveness of preventive interventions, as reported in the studies. Among the 13 studies examined, only three directly reported the effect size of their intervention. Of the remaining studies, five reported the significance of results using means, standard deviations, and confidence intervals, necessitating the computation of Cohen’s d to standardize the measure of intervention effectiveness. Specifically, Bernal et al. (2019) utilized a correlation coefficient for effect size calculation, highlighting the varied methodologies employed across the research landscape. Given the significant variation in follow-up time points for outcome measurement across studies and the prevalence of selective data reporting, a strategic decision was made to compute effect sizes at four critical time points: baseline, post-intervention, mid-point, and end-point. This approach facilitated a more granular analysis of intervention effectiveness over time. For instance, Bernal et al. (2019) conducted a 12-week intervention with six follow-up time points at 3, 6, 9, 12, and 15 months, enabling the computation of effect size at 3 months (post-intervention), 9 months (mid-point), and 15 months (end-point) (Bernal et al., 2019), as detailed in Supplementary file 5.

The significant heterogeneity among the studies—regarding outcome measures, intervention types, and follow-up periods—prevented the feasibility of conducting a meta-analysis. Instead, a narrative synthesis approach in systematic reviews, was employed. This methodology allowed for the systematic evaluation and integration of findings from the diverse range of studies included in the review. In doing so, it provided substantive insights into the effectiveness of various preventive interventions. The narrative synthesis approach proved particularly advantageous given the complexity of the review, allowing for a thorough exploration of the data while carefully navigating the inherent heterogeneity in study designs, active controls, outcome measures, and reporting practices encountered.

Thus, the strategic computation of Cohen’s d and, in cases such as that of Bernal et al. (2019), the use of correlation coefficients for effect size calculation were integral to the analytical strategy. These calculations enriched the synthesis by providing a quantifiable measure of intervention effectiveness where possible, demonstrating a commitment to a rigorous, yet flexible, examination of the evidence. This nuanced approach, aimed at identifying effective strategies for reducing depression and suicidal tendencies among adolescents, elevates the review beyond mere descriptive summarization, offering a comprehensive and insightful analysis of preventive mental health interventions.

The PRISMA flow diagram (Refer to Figure 1) illustrates the study selection process. All selected studies were RCTs. These studies examined the effectiveness of at least one preventive intervention compared to a control group, with a focus on reducing depression and suicidal tendencies among adolescents as outcome measures.

The JBI tool for RCTs, as outlined in the Supplementary material, was utilized to assess the quality of each study included in this review. The majority of these studies achieved a score of 85% (10/13). Notably, 10 of the 13 studies encountered challenges related to the blinding of participants and treatment providers, which resulted in lower scores. This is a frequent issue in studies involving psycho-educational interventions, highlighting the inherent difficulty of achieving blinding in behavioral or psycho-educational research (See Supplementary file 4).

This systematic review included 13 RCTs targeting depression and suicidal tendencies among adolescents (Refer to Supplementary Table 1). These studies were conducted across various developed countries, with four studies conducted in the USA (Bernal et al., 2019; Diamond et al., 2019; Saulsberry et al., 2013; Schwartz et al., 2023), two in New Zealand (Merry et al., 2012; Whittaker et al., 2017), two in Norway (Waraan et al., 2021a; Waraan et al., 2021b), and one each in Australia (Hetrick et al., 2017), Canada (Silverstone et al., 2017), the UK (Stallard et al., 2013), the Netherlands (de Jonge-Heesen et al., 2020), and Japan (Nagamitsu et al., 2022). The total sample comprised 10,707 adolescents aged 11 to 19 years, with a significant proportion of women (40%). The studies vary widely in sample size, ranging from 50 (Hetrick et al., 2017) to 5,030 (Stallard et al., 2013), indicating diverse research scopes and settings (Refer to Table 1 for key findings).

Interventions were predominantly conducted in school settings (de Jonge-Heesen et al., 2020; Hetrick et al., 2017; Nagamitsu et al., 2022; Silverstone et al., 2017; Stallard et al., 2013; Whittaker et al., 2017), underscoring their accessibility and the potential for integrated mental health support in educational environments. Community clinics were used in five studies (Bernal et al., 2019; Waraan et al., 2021a; Waraan et al., 2021b; Saulsberry et al., 2013; Schwartz et al., 2023), highlighting their role in targeted intervention delivery. Additionally, two studies utilized multiple settings, including schools, community clinics, and emergency rooms, to broaden demographic reach (Diamond et al., 2019; Merry et al., 2012).

Each study employed different intervention strategies, with a common focus on structured, manualized approaches. Predominantly, face-to-face delivery was employed in six studies (Bernal et al., 2019; de Jonge-Heesen et al., 2020; Diamond et al., 2019; Schwartz et al., 2023; Stallard et al., 2013; Waraan et al., 2021a; Waraan et al., 2021b), allowing therapists and participants to engage directly, which is crucial for effective mental health interventions. Additionally, four studies integrated digital platforms with traditional face-to-face methods (Hetrick et al., 2017; Nagamitsu et al., 2022; Saulsberry et al., 2013; Silverstone et al., 2017), enhancing accessibility and engagement through modern technology. Three studies adopted a fully digital approach, utilizing web-based platforms, mobile apps, and gaming to deliver interventions (Merry et al., 2012; Whittaker et al., 2017), demonstrating innovative strategies to address teen mental health in the digital era. These diverse methods reflect the evolving landscape of mental health interventions, adapting to both traditional and contemporary modalities to better meet the needs of adolescents. Intervention lengths ranged from 2 weeks (Nagamitsu et al., 2022) to 16 weeks (Diamond et al., 2019), with session durations adjusted for target populations, varying between 40 and 120 min. Specialists, including psychologists, physicians, and social workers, primarily implemented these interventions to ensure professional oversight and consistency.

Nine studies utilized active controls to benchmark the interventions (Bernal et al., 2019; de Jonge-Heesen et al., 2020; Diamond et al., 2019; Saulsberry et al., 2013; Schwartz et al., 2023; Stallard et al., 2013; Waraan et al., 2021a; Waraan et al., 2021b; Whittaker et al., 2017), with comparisons ranging from standard care practices to more structured psychotherapeutic approaches such as CBT. Two studies used treatment-as-usual as a control (Hetrick et al., 2017; Merry et al., 2012), providing a baseline for evaluating the added value of the interventions. This diverse array of control conditions facilitates a robust assessment of intervention effectiveness across different therapeutic environments and methodologies.

The primary outcomes targeted by the interventions varied across the 13 studies, focusing on the effectiveness of preventive interventions against depression and suicidal tendencies among adolescents. Seven studies specifically reported depression as their primary outcome (de Jonge-Heesen et al., 2020; Merry et al., 2012; Nagamitsu et al., 2022; Saulsberry et al., 2013; Stallard et al., 2013; Waraan et al., 2021a; Waraan et al., 2021b; Whittaker et al., 2017), three studies examined the impact on suicidal tendencies (Hetrick et al., 2017; Silverstone et al., 2017; Waraan et al., 2021a; Waraan et al., 2021b), and three others addressed both depression and suicidal tendencies within their analyses (Bernal et al., 2019; Diamond et al., 2019; Schwartz et al., 2023).

There was also significant variability in the frequency and duration of outcome assessments across these studies, reflecting the complexity of measuring the long-term effects of mental health interventions. The follow-up periods ranged from two to five time points (up to 15 months), with some studies, such as those by Bernal et al. (2019) and Silverstone et al. (2017), providing more extensive longitudinal data to better understand the sustainability of intervention effects. Notably, Diamond et al. (2019) focused exclusively on post-intervention outcomes, underscoring the immediate impact of their intervention. The variations in follow-up assessments across these studies highlight the challenges in synthesizing the effectiveness of interventions and demonstrate the need for consistent measurement approaches to reliably gauge their long-term benefits.

In evaluating the outcomes of 13 selected studies, it was revealed that seven reported reductions in depression and suicidal ideation among adolescents following the interventions, lasting up to 3 months. Despite these positive findings, a direct comparison did not consistently demonstrate that these interventions outperformed usual care or control groups. Meanwhile, two studies, Nagamitsu et al. (2022) and Silverstone et al. (2017), highlighted the effectiveness of their interventions compared to control conditions, while Merry et al. (2012) reported outcomes similar to those of usual care.

Only three of the 13 studies provided effect size data (de Jonge-Heesen et al., 2020; Diamond et al., 2019; Schwartz et al., 2023). For the remaining 10 studies, effect size was computed (Please refer to the “Methods” section and Supplementary file 5). The analysis revealed varied outcomes: While the majority of interventions did not show a significant advantage over control or treatment-as-usual (TAU) groups, exceptions were noted. Specifically, three studies (Nagamitsu et al., 2022; Schwartz et al., 2023; Silverstone et al., 2017) highlighted the efficacy of their interventions beyond that of their respective controls. In contrast, one study (Merry et al., 2012) reported that its intervention achieved results comparable to TAU, underscoring the potential for certain approaches to match the effectiveness of existing care protocols. In addition, the majority of evaluated interventions yielded minimal to small effect sizes, indicating a modest immediate impact on depression and suicidal ideation.

On the other hand, the impact of interventions on suicidal ideation and tendencies presents a complex picture, with studies demonstrating a range of outcomes. For example, Silverstone et al. (2017) and Diamond et al. (2019) explored interventions that included significant parental involvement and reported a notable reduction in suicidal tendencies among participants. Silverstone et al. (2017) implemented the EMPATHY program, which combined face-to-face and online CBT elements, showing a moderate effect on reducing suicidal ideation. Diamond et al. (2019)‘s approach, integrating Attachment-Based Family Therapy (ABFT) with parental coaching, yielded one of the highest effect sizes for reducing both depression and suicidal tendencies, highlighting the added value of family involvement in treatment.

Many of the examined interventions were found to have effects equal to or less than those of their control counterparts in reducing depression and suicidal tendencies among adolescents. This variability in outcomes can be attributed to several factors: small sample sizes with attrition rates exceeding 20% and significant missing data (Asarnow and Miranda, 2014), interventions not being sufficiently tailored to meet the unique developmental and emotional needs of adolescents (Ragelienė, 2016; Weersing and Brent, 2006), and the influence of the provider’s competence, available resources, and the context of delivery (Cuijpers et al., 2016; Purgato et al., 2020). While many interventions target both depression and suicidal tendencies, it is important to recognize that these two conditions may require different intervention approaches. Depression-focused interventions often prioritize mood regulation, cognitive restructuring, and coping strategies (Cuijpers et al., 2014), whereas interventions targeting suicidal tendencies may emphasize crisis management, safety planning, and immediate risk reduction (Zalsman et al., 2016). Although depression is a significant risk factor for suicidal behavior, the pathways leading to suicidal actions can be distinct, necessitating more specialized intervention strategies. Future research should explore these differences in greater detail to better tailor preventive interventions to each condition and improve their effectiveness.

Moreover, using active controls in studies may further obscure the true efficacy of interventions. According to equipoise theory, ethical trial design requires that treatments provided to control groups be as effective as possible, potentially reducing observed differences (Freedman, 1987). Additionally, response expectancy theory suggests that participants’ expectations of receiving beneficial treatment, regardless of group assignment, can enhance outcomes in both the intervention and control groups (Kirsch, 1985). This psychological effect, along with the ethical obligation to provide effective treatment to control groups, may lead to smaller observed differences between intervention outcomes and those of active controls, complicating the assessment of intervention superiority.

Effective intervention requires meticulous implementation and evaluation, tailored to the specific needs of adolescents (Fazel et al., 2014a; Fazel et al., 2014b; Weersing and Brent, 2006). Future research should focus on enhancing adherence strategies and integrating qualitative components to understand the motivations behind participant non-adherence. Additionally, adopting hybrid designs could provide insights into how different implementation strategies affect the overall effectiveness of interventions (Trutschel et al., 2023; Ullman et al., 2022).