Network perspective on childhood trauma and non-suicidal self-injury behaviors and functions in adolescents with depressive disorders

Abstract

BACKGROUND

Childhood trauma is a well-established risk factor for non-suicidal self-injury (NSSI) in adolescents with depressive disorder. However, whether trauma subtypes relate differently to specific NSSI behaviors and their underlying functional motivations are unclear. Understanding these patterns is crucial for identifying the pathways linking early adverse experiences to self-injury. We hypothesized that distinct trauma subtypes associate differently with specific NSSI behaviors and functional motivations.

AIM

To investigate the network linking childhood trauma, NSSI behaviors, and functional motivations in adolescents with depressive disorder.

METHODS

We conducted a cross-sectional study of 427 adolescents aged 12-18 years with depressive disorders in the psychiatric ward of a general hospital. Childhood maltreatment was assessed using the Childhood Trauma Questionnaire-Short Form, and NSSI behavior and functional motivation with the adolescent NSSI Assessment Questionnaire. A Gaussian graphical model with graphical least absolute shrinkage and selection operator regularization was applied to estimate the network structure and central and bridge indices.

RESULTS

The network contained 29 nonzero edges. Emotional abuse (EA) showed the highest centrality (expected influence = 1.42). The strongest trauma-NSSI association was between EA and non-damaging self-injury (NDSI; weight = 0.10). NDSI demonstrated the highest bridge expected influence of 1.75. Of the functional motivations, automatic negative reinforcement showed the strongest association with emotion expression (weight = 0.43). The network structure exhibited good accuracy, with adequate stability of the centrality estimates (correlation stability coefficient = 0.75).

CONCLUSION

Childhood trauma is distinctly associated with NSSI behaviors and functions. EA and NDSI are the key nodes that may inform symptom-level understanding of adolescents with depressive disorders.

Key Words: Depressive disorders; Adolescents; Network analysis; Childhood trauma; Non-suicidal self-injury; Functional motivations

Core Tip: This study used network analysis to reveal symptom-level associations between childhood trauma, non-suicidal self-injury behaviors, and functional motivations in adolescents with depressive disorders. Emotional abuse was identified as the most central trauma subtype and non-damaging self-injury as the most critical bridge node connecting childhood trauma with non-suicidal self-injury behavior-function system. The findings highlight specific trauma-behavior pathways and underscore the value of network-based models for identifying clinically meaningful intervention targets beyond traditional total-score approaches.

INTRODUCTION

Non-suicidal self-injury (NSSI) refers to deliberately and directly damaging one’s body tissue without suicidal intent[1]. Its increasing prevalence has made it a major global public health concern and attracted substantial societal attention[2]. Moreover, adolescents with depressive disorders are at high risk, with a prevalence of 54.60%[3]. Although NSSI is typically nonlethal, it strongly predicts suicidal ideation and suicide attempts[4] and is closely associated with diverse adverse outcomes, including antisocial behaviors and substance misuse[5]. Therefore, elucidating the underlying psychopathological mechanisms of NSSI is of considerable scientific and clinical importance for the development of targeted prevention and intervention strategies.

Childhood trauma refers to adverse experiences, such as emotional abuse (EA), emotional neglect (EN), sexual abuse (SA), physical abuse (PA), and physical neglect (PN), before the age of 18, which exert lasting negative effects on the individual’s physical and psychological development[6]. According to the stress-vulnerability model, early trauma experience may impair emotional regulation, cognitive processing, and behavioral coping strategies, thereby increasing susceptibility to maladaptive behaviors, such as NSSI and impulsive acts, when encountering later stressors[7]. Substantial evidence supports a robust association between childhood trauma and NSSI[8,9]. However, this relationship is not linear; instead, it may be driven by multiple intertwined mechanisms spanning behavioral manifestations, functional motivations, and emotional processing. For instance, different childhood trauma types may trigger distinct NSSI forms, whereas the motivations underlying NSSI (e.g., emotion regulation or social influence) may play a pivotal role in their development and maintenance.

Traditional statistical approaches, such as regression analysis, primarily focus on overall correlations at the variable level and often fail to identify the specific interactions between symptoms and behaviors. This limitation restricts the ability to uncover the underlying psychopathological mechanisms and clinical targets for intervention. By contrast, network analysis provides a novel methodological framework that overcomes these constraints by conceptualizing symptoms and behaviors as interconnected network nodes[10]. This approach enables the identification of central and bridge nodes across symptom clusters, thereby uncovering key psychopathological mechanisms and offering potential clinical targets for precise interventions[11]. However, studies have mainly constructed network structures of childhood-trauma dimensions and NSSI behaviors, with a lack of focus on systematically integrating childhood-trauma dimensions, NSSI behavior types, and functional motivations in adolescent samples with depressive disorders.

Therefore, in this study, we focused on adolescents with depressive disorders and employed network analysis to simultaneously examine the complex associations among childhood-trauma dimensions, NSSI behavior types, and functional motivations. The primary aims of this study were to: (1) Construct and visualize an integrated network structure of childhood trauma and NSSI behaviors and functions; (2) Identify hub nodes that exert central influence within the network; and (3) Detect bridge nodes that connect distinct symptoms or feature communities. This study sought to deepen the understanding of the mechanisms underlying NSSI in adolescents with depressive disorders and provide theoretical evidence and potential clinical targets for developing precise psychological interventions and prevention strategies.

MATERIALS AND METHODS

Participants

Convenience sampling was used to recruit adolescent patients with depressive disorders who were hospitalized in the Department of Psychiatry and Mental Health at the First Affiliated Hospital of Xi’an Jiaotong University between March 2023 and August 2025. Following were the inclusion criteria: (1) Met the diagnostic criteria for depressive episodes or recurrent depressive disorder according to the 11^th edition of the International Classification of Diseases; (2) Were aged 12-18 years; (3) Had no severe physical illness; and (4) Voluntarily agreed to participate. The exclusion criteria were as follows: (1) Inability to understand the questionnaire and instrument contents; and (2) Withdrawal of participation during questionnaire completion.

According to network analysis requirements, the sample size should exceed the total number of parameters (total parameters = threshold parameters + pairwise association parameters). The threshold parameters were equal to the total number of nodes and pairwise association parameters, nodes × (total nodes - 1)/2[12]. In this study, a 10-node network was constructed (five childhood-trauma dimensions, two NSSI behavior types, and three NSSI functions); therefore, the minimum required sample size was 55. A total of 440 questionnaires were distributed, and 427 valid responses were collected, yielding a valid response rate of 97.05%.

This study was reviewed and approved by the Medical Ethics Committee of the First Affiliated Hospital of Xi’an Jiaotong University (Approval No. XJTU1AF2023 LSK-132). All participants and their legal guardians provided informed consent and signed the relevant consent forms.

Measurement instruments

General information questionnaire: A self-developed questionnaire was used to collect demographic and background data, including age, sex, only child status, academic performance, current education stage, left-behind child status, and place of residence.

Adolescent NSSI Assessment Questionnaire: The adolescent NSSI Assessment Questionnaire, developed by Wan et al[13], comprises behavior and function components. The behavior subscale includes 12 items across two dimensions: (1) Non-damaging self-injury (NDSI); and (2) Damaging self-injury (DSI). Items are rated on a 5-point Likert scale (0 = “never”, 1 = “occasionally”, 2 = “sometimes”, 3 = “often”, and 4 = “always”). The total score of each dimension was divided by the number of items to obtain the mean dimensional score. Cronbach’s alpha for the behavior subscale was 0.921 in the original study and 0.932 in the present study. Following the criteria of the fifth edition of the Diagnostic and Statistical Manual of Mental Disorder, participants with a self-injury frequency ≥ 5 episodes in the past year were classified as having NSSI behaviors in this study.

The function subscale consists of 19 items assessing three dimensions: (1) Egoistic social interaction (ESI); (2) Automatic negative reinforcement (ANR); and (3) Emotional expression (EE). Items are rated on a 5-point Likert scale (0 = “completely inconsistent”, 1 = “inconsistent”, 2 = “uncertain”, 3 = “consistent”, and 4 = “completely consistent”). The total score of each dimension was divided by the number of items to obtain the mean score. Cronbach’s alpha for this subscale was 0.905 in the original study and 0.952 in this study.

Childhood Trauma Questionnaire-Short Form: The Childhood Trauma Questionnaire-Short was developed by Bernstein et al[14]. It comprises 28 items assessing five dimensions: (1) EA; (2) EN; (3) SA; (4) PA; and (5) PN. Items are rated on a 5-point Likert scale (1 = “never”, 2 = “rarely”, 3 = “sometimes”, 4 = “often”, and 5 = “always”). Items 2, 5, 7, 13, 19, 26, and 28 are reverse coded. The scores in each dimension were summed to generate the corresponding subscale scores. The Cronbach’s alpha values were 0.730 and 0.819 in the original and present samples, respectively.

Statistical analysis

Descriptive statistical analyses were performed using SPSS version 26.0. Categorical variables are reported as n (%) and continuous variables with normal distribution as mean ± SD. Network analysis was conducted using R version 4.4.3.

Network estimation: A partial correlation network model was constructed to examine the associations between childhood trauma, NSSI behaviors, and functional dimensions using the qgraph package in R. To reduce spurious associations, the least absolute shrinkage and selection operator algorithm was employed to estimate a regularized network and visualize its structure[15]. In the network, each childhood trauma and NSSI behavior/function was represented as a node, with edges indicating the strength of the associations between nodes. In the visualization, thicker edges represent stronger associations. Positive associations are indicated by solid red lines, whereas negative associations by solid blue lines.

Computation of centrality indices: Centrality indices - strength, closeness, betweenness, and expected influence (EI) - were calculated to assess the importance of each node[16]. Closeness and betweenness centrality yield unreliable results; therefore, they are not recommended as primary indicators of node importance in psychological networks[12]. Strength centrality does not differentiate positive and negative edges, which may limit the interpretability of networks with mixed associations. EI, defined as the sum of all edge weights connected to a given node, is particularly well-suited for evaluating nodes in networks that contain both positive and negative connections[17]. Accordingly, EI (standardized as Z-scores) was used as the primary centrality metric in this study.

The network tool package was used to compute the bridge centrality metrics: Bridge strength, closeness, betweenness, and bridge EI (BEI)[18]. BEI was adopted as the key indicator for identifying the nodes that connect communities within the network. Higher BEI values indicate a greater likelihood of a node activating cross-community interactions. The results were reported as standardized Z-scores[18].

Assessment of network accuracy and stability: The accuracy and stability of the estimated network were evaluated using the boot net package. First, the accuracy of the edge weights was assessed by computing 95% confidence intervals (CIs) through nonparametric bootstrap sampling (number of bootstraps = 1500), with a narrower CI indicating a higher precision of the edge weight estimates. Second, the stability of the centrality indices (i.e., EI and BEI) was evaluated using case-dropping subset bootstrapping (number of bootstraps = 1500) to calculate the correlation stability (CS) coefficient. A CS coefficient above 0.25 was considered acceptable; above 0.50 good; and above 0.70 optimal. Finally, bootstrapped difference tests for edge weights and paired comparisons of node centralities were conducted to examine statistically significant differences between specific edges and centrality metrics. In the CI plots, a wider shaded area indicates a greater magnitude of difference[12].

RESULTS

Descriptive statistics

A total of 427 adolescents with depression participated in the study (15.51 ± 1.75 years). Their detailed demographic information is presented in Table 1. Of them, 303 (70.96%) reported NSSI, highlighting a high prevalence in this clinical sample.

Table 1 Demographic characteristics of adolescent patients with depression.

Variable	Category	n (%)
Sex	Male	128 (29.98)
	Female	299 (70.02)
Only child	Yes	129 (30.21)
	No	298 (69.79)
Academic performance	Poor	98 (22.95)
	Average	243 (56.91)
	Good	86 (20.14)
Current education stage	Middle school	171 (40.05)
	High school	198 (46.37)
	Vocational secondary school	14 (3.28)
	Junior college	20 (4.68)
	Bachelor’s degree	7 (1.64)
	Dropped out	17 (3.98)
Left-behind children	Yes	32 (7.49)
	No	395 (92.51)
Place of residence	Urban	297 (69.56)
	Rural	130 (30.44)

The participants’ childhood trauma, NSSI behavior, and NSSI functional motivation scores are presented in Table 2. EN (17.22 ± 4.69) had the highest score, followed by EA (12.18 ± 4.86), PN and PA (moderate), and SA (the lowest; 5.74 ± 2.04). Of the NSSI behavior scores, the NDSI scores (1.25 ± 1.12) were higher than DSI scores (0.95 ± 0.98), indicating primarily mild forms of self-injury. Of the NSSI function scores, the ANR scores were the highest (1.91 ± 1.37), followed by EA (1.63 ± 1.26) and ESI (1.14 ± 0.91) scores. Overall, the sample exhibited a pronounced history of childhood emotional trauma, predominantly NDSI, and NSSI behaviors primarily driven by negative emotion regulation.

Table 2 Childhood trauma and non-suicidal self-injury behaviors and functional motivation scores.

Variables	mean ± SD
Childhood trauma
Emotional abuse	12.18 ± 4.86
Emotional neglect	17.22 ± 4.69
Sexual abuse	5.74 ± 2.04
Physical abuse	7.63 ± 3.70
Physical neglect	10.36 ± 3.27
NSSI behavior
Non-damaging self-injury	1.25 ± 1.12
Damaging self-injury	0.95 ± 0.98
NSSI function
Egoistic social interaction	1.14 ± 0.91
Automatic negative reinforcement	1.91 ± 1.37
Emotional expression	1.63 ± 1.26

NSSI: Non-suicidal self-injury.

Network structure

The estimated network comprised 45 potential edges, of which 29 (64.44%) exhibited nonzero weights, indicating a high level of internal connectivity in the system (Figure 1). The edge weight accuracy is shown in Supplementary Figure 1.

Figure 1 Network structure of childhood trauma and non-suicidal self-injury behaviors and functions in adolescents with depressive disorders. This network illustrates the relationships among childhood-trauma dimensions (emotional abuse, emotional neglect, physical abuse, physical neglect, and sexual abuse), non-suicidal self-injury behaviors (non-damaging and damaging self-injuries), and functional motivations (egoistic social interaction, automatic negative reinforcement, and emotional expression). Solid red lines represent positive associations between nodes. The thickness of the line represents the strength of the association between symptom nodes. SA: Sexual abuse; PA: Physical abuse; EA: Emotional abuse; EN: Emotional neglect; PN: Physical neglect; EE: Emotional expression; ANR: Automatic negative reinforcement; ESI: Egoistic social interaction; NDSI: Non-damaging self-injury; DSI: Damaging self-injury; NSSI: Non-suicidal self-injury.

In the childhood-trauma subnetwork, EA and PA had the strongest association (weight = 0.46), followed by EN and PN (weight = 0.35). These results indicate a structural co-occurrence pattern between the emotional and physical forms of trauma. In the NSSI behavior-function subnetwork, NDSI and DSI showed the strongest association (weight = 0.56), indicating a high co-occurrence of the two behavioral patterns. Of the functional nodes, EE and ANR (weight = 0.43) had the strongest association, followed by EE and ESI (weight = 0.41). Cross-construct connections showed that EA had the strongest association with NDSI (weight = 0.10). Additionally, DSI was associated with PN and SA with smaller edge weights (weight = 0.03 and weight = 0.06, respectively).

Centrality indices of the network

As shown in Figure 2A, EA exhibited the highest EI in the network (EI = 1.42), indicating its position as a central node and strong association with the other childhood-trauma dimensions, NSSI behaviors, and functional motivations.

Figure 2 Standardized value (Z-score) of expected influence and bridge expected influence for each node in the network. A: Each point represents a network node (childhood-trauma dimension, non-suicidal self-injury (NSSI) behavior, or NSSI functional motivation). The vertical axis shows the standardized expected influence (Z-score) of each node, indicating the strength of its connection to the other nodes in the network. Higher points correspond to more central nodes with a greater influence. The lines connecting the points are for visualization purposes only and do not imply causal relationships; B: Each point represents a network node (childhood-trauma dimension, NSSI behavior, or NSSI functional motivation). The vertical axis shows the standardized bridge expected influence (Z-score), indicating the strength of connection between the node and different symptom clusters (e.g., childhood trauma and NSSI behaviors/functions). Higher points indicate nodes that act as bridges between networks. The lines connecting the points are for visualization purposes only and do not imply causal relationships. EA: Emotional abuse; ESI: Egoistic social interaction; EE: Emotional expression; NDSI: Non-damaging self-injury; ANR: Automatic negative reinforcement; DSI: Damaging self-injury; PN: Physical neglect; EN: Emotional neglect; PA: Physical abuse; SA: Sexual abuse.

Bridge centrality analysis (Figure 2B) showed NDSI as the most prominent bridge node across communities (BEI = 1.75), playing a key structural role in connecting the childhood-trauma with NSSI behavior-function subnetwork.

Assessment of network accuracy and stability

Nonparametric bootstrap analysis indicated narrow 95%CIs of the edge weights, suggesting high reliability of the estimated edges (Supplementary Table 1). The CS coefficients for EI and BEI were both 0.75, indicating the robustness of the network centrality measures (Figure 3A and B). Significant differences were observed between the edge weights and centrality indices, indicating notable variability in the roles of the nodes and edges within the network (Supplementary Figures 2-4). These results, along with those of the bootstrap and centrality stability analyses, support the reliability of the network structure.

Figure 3 Stability of the expected influence and bridge expected influence values. A: The red bar indicates the average correlation between the expected influence values across bootstrap samples, reflecting the stability of the centrality estimates. Higher values suggest more reliable and stable measurements of symptom influence in the network; B: The red bar indicates the average correlation between the bridge’s expected influence values across the bootstrap samples, reflecting the stability of the centrality estimates. Higher values suggest more reliable and stable measurements of symptom influence in the network.

DISCUSSION

In this study, we employed network analysis to systematically examine the associations between childhood-trauma dimensions, NSSI types, and functional motivations in adolescents with depressive disorders. The participants’ scores showed that self-injurious behaviors are primarily used to alleviate negative emotions or internal distress while also partially serving as a means of EE.

The network model revealed the key connection patterns within this system. EA exhibited the highest EI, indicating its central position within the spectrum of childhood trauma. Meanwhile, NDSI showed the highest bridge centrality, serving as a critical bridge connecting the childhood-trauma and NSSI behavior-function subnetworks. These findings suggest that specific types of early adverse experiences, particularly EA, and mild and covert self-injurious behaviors, may play pivotal roles in structural associations across constructs. However, owing to the cross-sectional design, the observed network reflected associations rather than causal pathways. Future research should employ longitudinal or experimental designs to clarify the directionality and underlying mechanisms of these relationships.

The estimated network exhibited high overall connectivity, with 64.44% of edges having nonzero weights, suggesting that childhood-trauma dimensions, NSSI behavior types, and their functional motivations are not independent, but form a tightly coupled symptom-feature system. This finding is consistent with the core perspective of the psychopathological network model, which posits that mental disorders are not driven by a single latent factor but emerge from dynamic interactions among multiple interrelated nodes, such as traumatic experiences, behavioral manifestations, and functional motivations[19]. Therefore, to achieve systemic therapeutic effects, clinical interventions should not focus solely on individual symptoms, but also on target key nodes occupying central or bridge positions within the network.

Within the childhood-trauma subnetwork, EA and PA had the strongest connection, followed by EN and PN, indicating a stable co-occurrence pattern of emotional and physical trauma in adolescents with depressive disorders. This finding is consistent with previous results[20,21] and suggests that different types of trauma may not occur in isolation but often arise from similar family environmental factors. Caregiver dysfunction, emotion regulation difficulties, or hostile interactions frequently cause co-occurrence of emotional trauma with harsher disciplinary practices, such as corporal punishment or physical aggression[22]. Therefore, the strong EA-PA connection observed in this study may reflect the interactive effects of common risk factors, including family dysfunction and caregiver emotion regulation difficulties.

Within the NSSI behavior-function subnetwork, NDSI and DSI exhibited strong associations, suggesting a high co-occurrence of these two types of behaviors in adolescents with depressive disorders, potentially arising from shared behavioral triggers or underlying behavioral tendencies. Regarding functional nodes, the strong connection between EE and ANR indicates that internal emotion regulation needs are a core driving factor of NSSI maintenance. The stronger association between EE and ESI further suggests that self-injurious behaviors may play a dual role in emotion regulation and social communication, a point consistent with previous findings[23]. In this context, adolescents may engage in NSSI both to alleviate emotional distress and to convey suffering, seek attention, or elicit social support. Previous studies have similarly emphasized that NSSI functions as both an emergency emotion regulation strategy and as a nonverbal mode of EE[24]. This “emotion-social dual-function” pattern highlights the complexity of NSSI, aligning with emotion regulation theories[25] and social communication theories[26]. These theories suggest that self-injury serves as a dual tool for emotional release and social need expression. Therefore, clinical interventions should aim to enhance adolescents’ emotional recognition and regulation skills. In addition, they should help them develop adaptive social communication strategies, which would enable them to express emotional needs more effectively and seek social support in healthier ways, thereby reducing their reliance on NSSI for emotion regulation and social communication.

Cross-construct connections revealed differentiated associations between childhood-trauma dimensions and NSSI types. In particular, the strong association between EA and NDSI suggests that emotional trauma is more likely to drive covert, mild, and repetitive self-injurious behaviors. This may be related to its specific psychological impact on individual development, such as the formation of persistent negative self-concepts, shame, feelings of worthlessness, and emotional regulation difficulties[27]. This may prompt adolescents to adopt self-injurious strategies that are less detectable and can be self-regulated to alleviate internal negative emotions[28]. By contrast, the associations of DSI with PN and SA were weaker but identifiable, indicating that more intrusive or physically characterized traumas may elicit stronger psychological distress, impulsive behaviors, and expressive needs, leading individuals to engage in more severe and visible forms of self-injury. This pattern suggests that different childhood trauma types influence adolescents’ selection of specific forms of NSSI through distinct psychological pathways and motivations, thus providing a theoretical basis for understanding individual differences in NSSI and developing targeted intervention strategies.

Regarding centrality indices, EA exhibited the highest EI, indicating its close relationship with other childhood-trauma dimensions and higher impact on NSSI behaviors and functions. As a core pathological node, EA may serve as a key driver of overall network activation, consistent with the psychopathological network theory that posits that “central symptoms” have strong propagation effects[29]. Therefore, adolescents who have experienced EA may be at higher risk of NSSI, and identifying them is critical for clinical screening, preventive interventions, and case management. In clinical practice, interventions for adolescents who have experienced EA should focus on emotional regulation and self-concept reconstruction, helping them identify and address negative emotions and reduce self-criticism. However, several challenges must be addressed when designing interventions based on EA as a core pathological node. Early identification of EA is difficult due to its covert nature, which leads to underreporting. Additionally, individual variability in trauma responses complicates the design of personalized interventions. Given that EA frequently co-occurs with other trauma types, interactions between these cores’ nodes must be considered to ensure that interventions target the full network of symptoms. Despite these challenges, the network-based approach offers a holistic pathway, focusing on key nodes and their interactions, thus enabling tailored and effective treatments.

Bridge centrality analysis revealed NDSI as the most critical bridge node connecting childhood trauma with NSSI behavior-function system. As a primary pathway for cross-module propagation, NDSI may represent the initial behavioral response within post-trauma coping strategies and serve as the “entry point” through which emotional trauma influences NSSI functional activation. Clinically, this finding is particularly important: NDSI is often covert and difficult to detect, yet it may act as a hub node that accelerates the negative emotion-self-injury cycle within the network. Early identification of NDSI and intervention development may effectively disrupt cross-module pathological pathways, thereby reducing the risk of escalating to more severe forms of self-injury. In clinical practice with adolescents who rely on NDSI, which tends to be more covert, early intervention should emphasize the identification of these hidden self-injurious behaviors and provide emotional regulation skills along with effective coping strategies. These strategies aim to reduce the reliance on NDSI and promote healthier emotional regulation and adaptive social communication. However, the main challenge in designing interventions based on NDSI is the difficulty of early clinical identification, as NDSI often lacks obvious tissue damage and is frequently overlooked. This makes timely identification and intervention more complex. To address this, future research could explore the potential of intelligent systems, such as AI-based detection tools or digital monitoring platforms, to enhance early identification and improve the accuracy of interventions.

Despite providing novel insights, this study had several limitations: (1) Study design: The cross-sectional design could only identify associations between variables, not causal directions. Future research should use longitudinal follow-ups, experimental manipulations, or time-series network analyses to clarify the directionality and developmental trajectories of these structural connections; (2) Measurement bias: All variables were self-reported, which may have been influenced by recall bias, current emotional states, and social desirability, leading to potential common method bias. This limitation is particularly relevant in psychological assessments in which participants potentially provide responses that align with social expectations or their emotional states at the time of the survey. Future studies should incorporate multiple sources of information (e.g., reports from family members, clinical interviews) and objective indicators (e.g., biomarkers, physiological monitoring) to enhance the accuracy and reduce the potential for common method bias; (3) Sample representativeness: The sample was drawn from a single medical institution, which limits the generalizability of the findings to other regions and broader populations. This sample may not fully represent all adolescents with depression, particularly those who are outpatients or come from different backgrounds. For example, hospitalized patients may face more psychological and behavioral issues, whereas community or outpatient patients may exhibit different clinical characteristics. Moreover, a single-center sample may introduce biases related to geographical and cultural factors. Future studies should expand sample sources by including multicenter, community-based, or cross-regional samples to improve the external validity and broader applicability of the findings; (4) Unmeasured confounding: The network model did not include other variables that may influence self-injurious behaviors, such as psychiatric comorbidities, medication use, family functioning, biological factors, or personality traits. These unmeasured variables could potentially affect the network structure and confound the observed relationships. Future studies should construct more complex multilayer networks or include these variables as covariates to enhance the explanatory power; and (5) Cultural constraints: The specific cultural context and reporting of childhood trauma, expression of self-injurious behaviors, and social significance may have been influenced by cultural norms and social stigma. Future research should conduct cross-cultural comparisons to evaluate the stability and generalizability of these network features across cultural contexts.

CONCLUSION

Based on network analysis, this study identified structural interconnections among childhood trauma, NSSI behaviors, and functional motivations in adolescents with depressive disorders. The results indicated that EA had a central position in the network, while NDSI was the key bridge connecting childhood trauma and NSSI behavior-function subnetworks. Emotional regulation-related NSSI functional motivation also exhibited important network connectivity. The findings suggest that specific types of early adverse experiences and the associated psychological mechanisms may constitute critical pathways for cross-dimensional risk accumulation. Furthermore, the findings provide empirical evidence for early identification of NSSI and development of intervention targeting adolescents with depressive disorders. Mechanism-targeted interventions focusing on core and bridge nodes may be more effective at disrupting pathological processes than traditional symptom-centered approaches, potentially improving clinical outcomes.