Relevant articles were selected through a comprehensive search of electronic databases, covering the entire period from the inception of each database until August 23, 2024. The search was updated on December 21, 2024, to ensure the inclusion of the most recent studies. The databases used included the Web of Science Core Collection, PubMed, the Cochrane Library, PsycINFO, CNKI, WANGFANG, and Weipu. The search strategy in the title, abstract, and keywords included the following terms: For intervention, “Acceptance and Commitment Therapy”, “acceptance-based”, or “ACT”; For population, “depression”, “depressive disorder”; For age group, “adolescent”, “teen”, “teenager”, or “adolescence”; For study design, “randomized controlled trial”, “randomised controlled trial”, or “RCT”.

A comprehensive search was conducted for studies published in peer-reviewed publications in both English and Chinese. Additionally, a manual search was undertaken by examining the reference lists of the identified articles and utilizing the related article attributes in the databases.

The PICOS framework emphasizes the elements of Population (P), Intervention (I), Comparison (C), Outcome (O), and Study design (S). For Population (P), adolescents aged 10–18 diagnosed with depression through a DSM-based clinical interview or other standardized diagnostic criteria were classified as clinical samples, whereas those assessed using standardized questionnaires with established reliability and validity were classified as non-clinical samples. For Intervention (I), the intervention groups comprised individuals receiving ACT or Internet-based Acceptance and Commitment Therapy (IACT). For Comparison (C), control groups comprised individuals on a wait-list (WL), those receiving treatment as usual (TAU), those under medical therapy management (MTM), or those undergoing other forms of treatment. For outcome (O), studies were required to report baseline depression scores, with depression levels serving as the primary outcome measure. Studies were required to report mean values, standard deviations, and sample sizes. For Study design (S), this Meta only includes RCTs to ensure the rigor of the intervention effect assessment. The exclusion criteria for this review were as follows: Duplicate publications, studies with designs that do not meet the RCT requirements, and experiments lacking ethical review. Additionally, redundant studies and studies with missing data were excluded to ensure the reliability and accuracy of the review.

The studies were selected based on the PRISMA guidance (38). Two assessors independently reviewed the literature. In case of any disagreement, the assessors engaged in a discussion to resolve the issue. If a consensus was not reached, an additional assessor was consulted. The process of data extraction adhered to the Cochrane principles for systematic evaluation, and the following information was documented: This includes (1) bibliographic details (2); participant count, average age, and standard deviation (3); gender distribution of the sample (female%) (4); details of the interventions in ACT group (number of ACT interventions, duration of a single session, and overall duration) (5); characteristics of the control group (6); outcome measures and duration of follow-up (see Table 1 for details).

Two independent researchers assessed the methodological quality of the included RCTs using the Cochrane Collaboration Risk of Bias Tool 2.0 (ROB 2) (39), which evaluates aspects such as random sequence generation, allocation concealment, blinding of participants and investigators, blinding of outcome assessment, completeness of outcome data, and selective reporting. The risk of bias in each domain was categorized as low, high, or unclear, based on the criteria in the Cochrane Collaboration Handbook. The screening process was conducted in two phases: initial screening of title and abstracts and subsequent screening of full texts. Two researchers (XY and BZ) independently performed both screenings, recording the reasons for exclusion at each stage. In the initial screening phase, the researchers assessed 295 articles, achieving good consistency with Cohen’s kappa = 1. In the subsequent phase, the remaining 34 full-text articles were independently reviewed, with similarly high consistency (Cohen’s kappa = 1). In addition, the risk of publication bias was assessed by visual inspection of the funnel plots, the Trim and Fill test, and the Egger test.

The statistical analysis was conducted through a meta-analysis of the included studies using RevMan 5.4 and Stata 15.0 software. In this study, we clearly defined the primary outcome as adolescent depression levels both post-intervention and at follow-up, with PF as the secondary outcome measure.

For each study, variance and standard error were calculated based on its sample size, effect size, and reported standard error. To ensure rigorous statistical analysis, we used the standardized mean difference (SMD) to represent effect sizes when the same outcome was assessed using different scales. According to Cohen’s rule of thumb for interpretation, an effect size of 0.2 or smaller is considered small, between 0.2 and 0.5 is considered medium, and an effect size of 0.8 or larger is considered large (40, 41).

Heterogeneity was assessed using both the I² statistic and the Q statistic. The I² statistic reflects the proportion of variability in the overall effect size that is attributable to heterogeneity, ranging from 0% to 100%. Specifically, heterogeneity is categorized as low (0% < I² < 25%), moderate (25% ≤ I² < 50%), high (50% ≤ I² < 75%), and significant (75% ≤ I² < 100%). When heterogeneity is low (I² < 25%), a fixed-effects model is applied; when heterogeneity is high (I² ≥ 50%) and multiple potential factors are present, a random-effects model is used. The Q test further supports the model choice: if the Q test is not significant (p > 0.05), a fixed-effects model is appropriate for low heterogeneity; if the Q test is significant (p ≤ 0.05), a random-effects model is warranted. Subgroup analyses were conducted when significant heterogeneity was present. The combined effect size (ES) was calculated and evaluated statistically, and the outcome measures in this study are expressed as weighted mean differences (MD) along with their 95% confidence intervals (95% CI).

To assess whether PF can predict the impact on depression levels, we conducted a meta-regression analysis, using PF as the predictor and depression levels as the dependent variable. Meta-regression was also employed to evaluate the potential impact of continuous variables, such as average age and the percentage of females in the sample. Additionally, subgroup analyses were performed based on the following variables: intervention type (e.g., offline ACT vs. IACT), control group type (WL vs. TAU vs. MTM), sample type (clinical vs. non-clinical), and intervention setting (group vs. individual).

Figure 1 presents a flow diagram illustrating the article selection process—from the initial search to final inclusion—in accordance with PRISMA guidelines. The initial database search yielded 354 articles, with an additional 3 articles identified through reference tracking. After removing 86 duplicate records, 271 articles remained. Following a screening of titles and abstracts, 18 studies were selected for full-text review. Furthermore, an additional search conducted on December 21, 2024, yielded 13 more studies following full-text review. Notably, one trial—a PhD dissertation (42) —included an ACT intervention section that was consistent with its earlier published, in-print article (43). Consequently, the final selection comprised 24 papers that met the inclusion and exclusion criteria.

Of the 24 publications considered, one article by Livheim et al. (43) reported two separate randomized controlled trials conducted in different countries. Consequently, this article was treated as two distinct studies, bringing the total number of studies to 25. (see Figure 1 )

Twenty-five studies, published between 2009 and 2024, were included in the meta-analysis; all of them were RCTs. These studies comprised 5 Chinese-language papers (44–48) and 20 English-language papers (23, 43, 49–65), involving a total of 2352 participants. Regarding sample type, 6 studies were conducted on clinical samples (45, 46, 48, 51, 58, 64), while 19 studies targeted non-clinical samples (23, 43, 44, 48–50, 52–57, 59–65). Among these, 4 studies focused on IACT (50, 52, 53, 59), while 21 studies focused on offline ACT (23, 43–49, 51, 54–58, 60–65). The control groups varied across studies: Some studies used TAU (23, 43, 47, 51, 56, 57), while other control conditions included WT (44, 48–50, 52–55, 59, 61–65) and MTM (45, 46, 58, 60). In terms of intervention setting, 3 studies delivered interventions on an individual basis (52, 53, 59), whereas the remaining studies were conducted in group settings. The duration of a single session varied between 25 and 120 minutes; notably, 6 studies did not provide information on the duration of a single session (43, 48, 49, 51, 62–64), and 6 studies did not disclose the overall intervention duration (48, 55, 58, 62). Regarding follow-up assessments, 10 studies included a follow-up period (47, 50, 51, 54, 55, 58–61, 65), while 2 studies mentioned a follow-up period but did not provide detailed and comprehensive follow-up data (43, 64) (see Table 1 ).

Summary assessments of the risk of bias, including results from 25 studies, are presented in Figure 2 . All of these studies were RCTs, with no baseline differences between the experimental and control groups. The studies reported the methods and procedures for correctly assigning participants to groups. Six of the 25 studies employed blinded outcome measures (50–54, 56). A double-blind approach was not feasible due to the nature of the ACT intervention, leading to an overall high risk of bias.

Additionally, a funnel plot was used to assess the potential publication bias of each outcome measure. The funnel plots for follow-up depression levels and the secondary outcome of PF showed symmetric scatter distributions, suggesting no publication bias in the analysis. Further Egger tests for each outcome revealed no publication bias (p > 0.05). However, the funnel plot for the immediate post-intervention depression levels showed an asymmetric scatter distribution. Further Egger regression tests indicated statistical significance (p < 0.05). The Trim-and-Fill method was used to impute missing studies for the post-intervention depression data, and the results were found to be relatively stable, indicating that publication bias had little impact on the study’s findings. The results and the funnel plot are presented in Table 2 , Figure 3 .

This meta-analysis includes 25 randomized controlled trials that investigated the impact of ACT on depression in adolescents. A total of 2,352 subjects participated in these studies. The results show that ACT is effective in reducing depression levels in adolescents, and PF plays a moderating role in this effect. Through practices such as acceptance, cognitive defusion, and other related techniques, individuals can align their behaviors with their personal values, thereby effectively reducing depressive symptoms (66, 67). A meta analysis by López-Pinar et al. found that improvements in PF significantly predicted reductions in depressive symptoms and accounted for a substantial proportion of the variance (24). This finding provides strong empirical support for PF as a core mechanism in ACT interventions. Previous research has shown that the therapeutic effects of ACT are facilitated by enhanced PF (68). Recent studies have also emphasized that ACT not only improves depressive symptoms but also enhances PF (26, 69). Several studies have demonstrated that PF predicts improvements in depressive symptoms.

ACT has been evaluated in 25 randomized controlled trials assessing its efficacy against various control conditions, including wait-list groups, conventional treatments, and medication. These trials included a variety of study designs and participant profiles, offering a comprehensive understanding of ACT’s performance across diverse therapeutic contexts. The analysis of these studies demonstrated that ACT significantly reduces depression levels in adolescents compared to wait-list controls, supporting the findings reported by López-Pinar (24). His study also highlighted ACT’s unique advantages in emotional regulation and behavioral change, suggesting that ACT offers clear benefits over no treatment. ACT works by increasing PF, enabling adolescents to better accept negative emotions and engage in valued actions despite emotional discomfort (70). This is especially beneficial for adolescents who experience intense negative emotions and cognitive distortions, as it helps them separate their sense of self from their emotional states. Similarly, Fang highlighted the efficacy of ACT in children, and our study further supports its effectiveness in treating adolescent depression, suggesting that ACT is broadly applicable across age groups (25). However, when compared to conventional treatments or medication, ACT did not demonstrate statistically significant superiority, indicating that its effects are comparable to those of other active treatments. This conclusion is further corroborated by subsequent studies examining IACT (21, 71), which similarly found no significant differences in outcomes compared to other established therapeutic interventions. Given the variable effectiveness of ACT across different control conditions, a one-size-fits-all approach is unlikely to be optimal for treating adolescent depression. Developing individualized treatment plans that reflect each adolescent’s clinical profile, circumstances, and needs is therefore essential. Tailored care can better address the nuanced challenges adolescents encounter in real-world settings, ultimately improving outcomes and ensuring treatment that is both effective and contextually appropriate.

IACT can mitigate certain drawbacks of traditional manualized interventions, notably their high cost and time-intensive nature. However, this study found that IACT did not outperform offline ACT in reducing depression levels. These findings are consistent with the outcomes of prior meta-analyses (72, 73). First, the therapeutic relationship in online interventions tends to be less robust than in offline, face-to-face interventions (74). Participant persistence is a key factor affecting the effectiveness of both ACT and IACT interventions. In offline ACT interventions, participants typically show higher engagement, with persistence rates exceeding 70%. In contrast, lower persistence in IACT interventions results in less effective outcomes. Lappalainen et al. (53) reported that IACT interventions were effective in reducing depressive symptoms among adolescents who completed at least half of the program. This success in alleviating depressive symptoms is observed in adolescents who adhere to the IACT intervention schedule. However, for the entire randomized sample, there were no significant differences in changes in depressive symptoms between the IACT intervention group and the control group. Therefore, it is essential to improve persistence and motivation among adolescents in IACT interventions to effectively alleviate adolescent depression. Finally, peer support and interaction significantly influence the outcomes of interventions. Peer support is vital for improving the well-being of adolescents with depression (5). Adolescence is characterized by a significant behavioral shift toward increased social interaction, a trend less prevalent in adults (75). Consequently, adolescents tend to seek more peer support and social engagement, and are greatly influenced by peer behaviors (76). Offline ACT primarily consists of group sessions, facilitating substantial peer support and interaction. Therapy participation with peers fosters mutual recognition, shared experiences, and a sense of belonging, thereby enhancing experiential empathy among participants (77, 78). Simultaneously, group therapy participation can reduce feelings of loneliness and provide opportunities to learn new coping strategies and behaviors from peers (4, 7, 37, 79, 80). The studies by Lappalainen et al. on individual interventions (52, 53)indicated no significant improvement in depressive symptoms in their whole randomized samples. Notably, the 2023 study was conducted during the COVID-19 pandemic (52). Adolescents have a strong need for peer contact and social support (78), and the constraints imposed by COVID-19, requiring them to confront significant external challenges, can exacerbate depressive symptoms (37). During the lockdown, with schools closed and a shift to online learning, adolescents experienced diminished social relationships and isolation from their peers (79). Consequently, the application of individual IACT interventions during this period proved ineffective in treating depressive symptoms.

Although no significant difference in ACT’s intervention effects was found between clinical and non-clinical samples, ACT demonstrated a significant effect on adolescent inpatients with depressive disorders, effectively reducing depressive symptoms. Additionally, in non-clinical samples of healthy individuals, ACT has been shown to reduce depression levels. This finding is consistent with previous research, which suggests that individuals with more severe depression show greater potential for improvement following intervention, leading to stronger treatment effects (80). This suggests that ACT may be an effective intervention not only for clinical populations but also for individuals at risk of developing depressive symptoms. The lack of a significant difference between clinical and non-clinical samples may be due to common underlying factors contributing to depression, such as emotional regulation difficulties and low PF, which ACT directly targets (24). However, it is important to consider that the severity of depression in clinical samples may require more intensive or tailored interventions, which could explain the variation in effects between clinical and non-clinical groups. Moreover, the finding that ACT reduces depression in healthy, non-clinical populations supports the growing body of evidence suggesting that even subclinical levels of depressive symptoms can benefit from ACT (25, 26). This highlights the importance of early intervention in preventing the onset of full-blown depressive disorders.

This meta-analysis evaluated the use of ACT in treating adolescent depression and its impact on PF, considering the psychological differences between adolescents and adults. The study also examined the varying effects of interventions across control groups, differences between offline ACT and IACT, and intervention variations among different populations. This study is the latest meta-analysis to analyze the efficacy of ACT in treating adolescent depression. It includes comprehensive subgroup analyses and wide-ranging coverage, thereby enhancing its informativeness.

This study has several significant limitations that impact the interpretation of its findings. Primarily, the meta-analysis included only four IACT studies that met the specified inclusion and exclusion criteria. The small sample size significantly limits the generalizability of the findings. To mitigate this limitation, it is essential to increase the number of studies included in future meta-analyses, thereby enhancing both the generalizability and reliability of the findings. Additionally, the capacity to perform comprehensive subgroup analyses in this study was restricted by the limited number of studies and inconsistent reporting of key variables (participant numbers, intervention durations, symptom severity, and measurement tools). These inconsistencies significantly hinder the effective evaluation of ACT’s efficacy across various subgroups. To systematically and accurately evaluate the efficacy of ACT in treating adolescent depression, researchers must ensure consistent and comprehensive reporting of critical variables. Standardizing these variables can substantially enhance the comparability and generalizability of research findings, thereby yielding more definitive conclusions regarding the effectiveness of ACT.

Adolescents experience rapid physical and mental development, marked by intense, fluctuating emotions and often displaying contradictory and rebellious behavior. Individuals with depression disorders at varying ages of onset exhibit distinct variations in the manifestation of clinical symptoms. Comparing depression in adulthood to depression in adolescents, it is crucial to consider that physiological and environmental changes throughout adolescence can affect symptoms and progression in this age group. Therefore, analyzing the effectiveness of ACT on adolescent depressive symptoms is essential. This meta-analysis demonstrates ACT’s capability in addressing these symptoms, offering valuable insights for psychologists, therapists, and psychiatrists involved in adolescent depression interventions. It emphasizes the importance of incorporating interactive group settings to boost motivation among adolescents, particularly in IACT programs, to enhance their engagement and persistence in the therapy. Moreover, the findings of this study broaden the applicability of ACT, showing its effectiveness not only in alleviating depressive symptoms among the general adolescent population but also in providing significant benefits for those already experiencing depression. By highlighting the advantages of group interactions and the extended impact of ACT, this research contributes to the field by suggesting practical approaches for improving mental health interventions for young people.

For future research aimed at optimizing ACT effectively, comprehensive reporting of treatment details is crucial. This will enable researchers to draw reliable conclusions on how to maximize the impact of ACT. Additionally, more extensive randomized controlled trials are needed to investigate the effectiveness of both traditional ACT and IACT. Such studies are essential for determining the efficacy of these interventions across diverse settings and populations. Intervention strategies must be thoughtfully tailored to account for the unique physiological and psychological characteristics of adolescents, as well as the environments in which they live. This targeted approach not only enhances the relevance of the interventions but also improves their overall effectiveness. Future research should further explore the relationship between the intervention mechanisms of ACT and depressive symptoms. By focusing on these areas, future research and clinical practice can significantly improve the outcomes of ACT interventions for adolescents, ultimately leading to more sustainable and impactful treatment strategies.