Adverse childhood experiences, bullying victimization, and depressive symptoms among Chinese adolescents: the moderating role of perceived social support Authors: Qingyuan Liu, Yuan Li, Cong Wang, Yunyun Zhao, Wanqi Ren, Puyu Su, Gengfu Wang, Qunshan Shen Published: 2026 ABSTRACT AbstractBackgroundAdverse childhood experiences (ACEs) and bullying victimization are prevalent psychosocial stressors associated with adolescent depressive symptoms. However, less is understood about their co-occurrence, whether distinct exposure profiles confer differential risk, and the role of perceived social support in moderating the impact of ACEs and bullying on depression. This study addresses these gaps within a large cohort of Chinese adolescents.MethodsA cross-sectional survey was conducted with 5,724 middle school students (mean age = 13.5 years) in Anhui Province, China. Participants completed self-report measures on ACEs, peer victimization, depressive symptoms, and perceived social support. Linear regression analyses explored the relationships between ACEs, bullying victimization, and depressive symptoms, as well as the moderating effect of perceived social support. Latent class analysis (LCA) identified patterns of co-occurring ACEs and victimization, with subsequent regression models examining associations with depressive symptoms and the moderating effect of social support.ResultsAmong the participants, 65.5% reported experiencing at least one ACE, and 30.0% reported bullying victimization. Linear regression indicated that both ACEs and bullying victimization significantly elevated the risk of depressive symptoms, with emotional abuse and relational bullying showing the strongest associations. LCA revealed four distinct exposure profiles: a low-risk group with minimal exposure to ACEs and victimization; a family adversity group with high household adversity but low victimization; a victimization group with high peer victimization but relatively low ACE exposure; and a high-risk group with elevated probabilities of both ACEs and bullying victimization. Linear regression further showed that the combined ACEs and bullying victimization group had the highest depressive symptom scores, and LCA identified a high-risk subgroup with both ACEs and victimization, which also exhibited high depressive symptoms. Perceived social support significantly moderated the relationship between ACEs and depressive symptoms, but did not moderate the relationship between bullying victimization and depressive symptoms.ConclusionsBoth ACEs and bullying victimization independently and cumulatively contribute to adolescent depressive symptoms. Perceived social support may serve as a protective factor against ACE-related depressive symptoms, though its role in buffering the effects of bullying victimization is less pronounced. Interventions should be tailored to specific risk profiles, addressing both familial and peer-related adversities to mitigate adolescent depression.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-27896-y.Keywords: Adverse childhood experiences, Bullying victimization, Depressive symptoms, Adolescents, Social support Background Depressive symptoms include persistent low mood, fatigue, and social withdrawal. While these may not always meet the criteria for a clinical diagnosis, they can still significantly impair academic performance, peer relationships, and emotional well-being [1]. Even subclinical depressive symptoms in adolescence are linked to future mental health challenges and reduced life functioning [2, 3]. Adolescence, therefore, represents a critical period for identifying psychosocial risk and protective factors associated with depression [1]. Among these factors, adverse childhood experiences (ACEs) and bullying victimization have received considerable attention as key family- and peer-related stressors during development [4, 5]. ACEs refer to harmful events occurring before the age of 18, typically classified into three categories: abuse (physical, emotional, or sexual), neglect (physical or emotional), and family dysfunction [6]. It is estimated that 57.7% of children and adolescents experience at least one ACE, and 14.8% encounter four or more types [7]. Empirical studies have demonstrated a strong dose–response relationship between ACEs and depression, showing that increased ACE exposure correlates with more severe depressive symptoms [8]. Bullying is characterized by repetitive aggressive behavior aimed at harming or controlling the victim through physical violence, verbal humiliation, relational exclusion, or cyber aggression [4]. Bullying victimization is a significant source of psychosocial stress and is closely associated with a range of mental health outcomes, including depression, anxiety, and trauma-related stress [9]. Individuals exposed to ACEs are more likely to experience bullying victimization [10]. One possible explanation is that ACEs may contribute to psychosocial vulnerabilities, such as lower self-esteem, difficulties in peer relationships, and heightened emotional sensitivity, which may increase susceptibility to bullying [10, 11]. A longitudinal study of Chinese adolescents found that those who experienced three or more ACEs were significantly more likely to be bullied during middle school, with a risk ratio up to 2.3 times higher compared to peers without ACE exposure [12]. Consequently, family and peer adversities often co-occur within the adolescent developmental context [10]. Although ACEs and bullying victimization are both linked to depressive symptoms, few studies have explored their combined impact on adolescent mental health. Specifically, it remains unclear whether their co-occurrence exacerbates depressive symptoms or whether distinct exposure profiles confer varying levels of risk. Existing research typically employs variable-centered approaches, examining ACEs or bullying victimization in isolation. However, person-centered methods, such as LCA, may offer a more nuanced understanding by identifying subgroups defined by the co-occurrence of family and peer adversities [13]. These questions are particularly pertinent in the Chinese context, where adolescents often grow up in family environments marked by strong academic pressures and relatively limited emotional expression [14]. Furthermore, peer victimization in this setting may predominantly take relational forms, such as exclusion and social isolation, within highly competitive school environments [15]. These contextual dynamics highlight the importance of exploring the intersection of ACEs and bullying victimization, as well as the potential moderating role of social support, among Chinese adolescents. In addition to identifying risk factors, understanding protective correlates is essential for the prevention and intervention of adolescent depression. Social support refers to the assistance and understanding individuals receive from their social networks, including family, friends, educational institutions, or the broader community, encompassing emotional, informational, and material dimensions [16]. One prominent theory explaining the protective role of social support in psychological distress is the stress-buffering model proposed by Cohen and Wills [17, 18]. This model suggests that social support serves a regulatory function in response to stressful events [18], with higher levels of support buffering the psychological strain caused by adversity, thus reducing psychological distress. Previous studies have consistently shown that higher perceived social support is linked to fewer depressive symptoms among adolescents facing adversity [19–21]. While the stress-buffering model offers a valuable framework for understanding the protective effects of social support, its applicability may vary depending on the nature of the stressor [17, 22]. ACEs and bullying victimization differ in several key dimensions, including context (family vs. peer), temporal characteristics (chronic vs. situational), and perceived controllability. ACEs typically stem from family environments and are chronic, enduring stressors that influence developmental trajectories over time [5, 23]. In such contexts, external support—whether from peers, teachers, or extended family—can provide compensatory emotional resources and alternative relational experiences, potentially reducing ACE-related psychological distress [24]. In contrast, bullying victimization occurs within the peer context, where the very social environment that could offer support may also be the source of stress [25, 26]. According to interpersonal stress theory and Lazarus and Folkman’s cognitive appraisal model, the effectiveness of coping resources depends on how individuals appraise the stressor and the perceived availability of support within that context [22]. When the stressor arises from peer relationships, adolescents may view available support as less accessible or trustworthy, particularly due to concerns over social exclusion, stigma, or retaliation. Furthermore, bullying often involves repeated, immediate, and publicly visible interactions, which may limit the protective capacity of general social support [4, 27]. Moreover, bullying victimization frequently involves stigma-related and relational dynamics that inhibit help-seeking behavior. Experiences of relational aggression, for example, may be internalized as social rejection, leading to shame, self-blame, and a fear of negative judgment. Additionally, relational bullying often manifests through covert actions such as exclusion, rumor-spreading, and manipulation of peer relationships, making it less visible to adults and harder to detect or intervene upon [28, 29]. Emerging biological evidence indicates that the neurobiological correlates of various adversities may differ, influenced by the context and chronicity of the stressor [30, 31]. Family-related adversities (e.g., abuse or neglect) are typically chronic, developmentally entrenched, and closely linked to attachment systems, often leading to long-term dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis and altered baseline stress responsiveness [32–34]. In contrast, peer victimization, especially in school environments, tends to involve more acute, socially evaluative stressors, activating neural circuits associated with social threat processing, such as increased amygdala reactivity and sensitivity to social rejection [25, 27, 35]. These context-specific neurobiological patterns have significant implications for intervention strategies. For instance, individuals exposed to ACEs may be more responsive to external emotional resources aimed at stress regulation [17, 24], whereas victims of peer bullying—whose neural responses are more attuned to ongoing social threats and peer evaluation—may show limited responsiveness to generalized social support that does not directly address the peer context. This distinction may help explain why social support appears to weaken the ACE–depression link while showing a less pronounced effect in the bullying victimization–depression pathway. The primary objectives of this study were: (1) to examine the independent associations of ACEs and bullying victimization with depressive symptoms among Chinese adolescents; (2) to investigate whether the co-occurrence of ACEs and bullying victimization was linked to heightened depressive symptoms using cumulative exposure and latent class approaches; and (3) to assess whether perceived social support differentially moderated the associations of ACEs and bullying victimization with depressive symptoms. Methods Participants This cross-sectional study was conducted as part of the project “Health and Risky Behaviors among Middle School Students in Anhui Province.” Participants were selected from three regions in Anhui Province (Huaibei, Hefei, Xuancheng) using a multistage random cluster sampling method. Detailed information can be found in a previous publication [36]. A total of 5,724 valid questionnaires were retained for analysis, yielding an effective response rate of 98.15% (5,724/5,832). The mean age of the sample was 13.5 years (SD = 1.0), with 3,006 males (52.5%) and 2,718 females (47.5%). Demographic Characteristics of the Participants were shown in Table 1. Table 1Demographic Characteristics of Adolescent ParticipantsSample CharacteristicsM(SD) / Frequency (%)Gender Female2718(47.5) Male3006(52.5)Age13.54(0.96)Only child status Yes1942(33.9) No3782(66.1)Residential area Rural2955(51.6) Urban2769(48.4)Family type Nuclear family3221(56.3) Single-parent family833(14.6) large family1596(27.9) Others74(1.3)Family economic status Poor499(8.7) Medium4277(74.7) Good948(16.6)Mother’s education level Primary and below1245(21.8) Junior middle2640(46.1) Senior middle1839(32.1)Relationship with mother Poor to general1726(30.2) Good2573(45.0) Very good1425(24.9)Father’s education level Primary and below746(13.0) Junior middle2674(46.7) Senior middle2304(40.3)Relationship with father Poor to general2432(42.5) Good2234(39.0) Very good1058(18.5)Academic performance Poor1208(21.1) General2723(47.6) Good1793(31.3)Friend number ≤ 21669(29.2) 3–52336(40.8) ≥ 61719(30.0)Continuous variables are presented as mean (standard deviation), and categorical variables are presented as frequency (percentage). Academic performance, parent–child relationship, family economic status, and number of friends were self-reported by participantsN = 5724 Demographic Characteristics of Adolescent Participants Continuous variables are presented as mean (standard deviation), and categorical variables are presented as frequency (percentage). Academic performance, parent–child relationship, family economic status, and number of friends were self-reported by participants N = 5724 Measures Adverse Childhood Experiences (ACEs) Data on 12 categories of ACEs were collected within the context of Chinese culture. Five types of childhood maltreatment were assessed using the Chinese version of the Childhood Trauma Questionnaire-Short Form (CTQ-SF), while seven types of family adversity were evaluated based on our previous research [37]. Each category of childhood adversity was dichotomized as either present or absent. Detailed information regarding the ACE measures is provided in the Supplementary Method. Any ACE exposure was defined as present if at least one category of adversity was endorsed, and absent otherwise. The Chinese version of the CTQ-SF has demonstrated acceptable reliability and validity among Chinese adolescents [38]. Bullying victimization Questions regarding school bullying victimization were adapted from previous research [39]. Responses were assessed on a 5-point scale: 1 indicating “none,” 2 for “only once or twice,” 3 for “two to three times a month,” 4 for “approximately once a week,” and 5 for “several times a week.” Participants were first provided with a standardized definition of bullying (qifu) in Chinese, then asked two parallel questions regarding the frequency of either bullying others or being bullied at school within the past two months [39]. The details of this scale, which has shown high reliability and validity, are documented in our previous study [11], which reported a test‑retest Kappa of 0.85–0.96 and internal consistency coefficients of 0.74–0.80 in a sample of Chinese adolescents. Following the Olweus criterion for bullying as repetitive victimization, students who reported being bullied at least two to three times a month were classified as bullying victims. Participants were considered victims of peer bullying if they endorsed at least one subtype of bullying at this frequency threshold [40]. Depressive symptoms Depressive symptoms were assessed using the 20-item Chinese version of the Depression Scale (CES-DC) [41]. This scale has been validated, with higher scores indicating more severe depressive symptoms. The total score ranges from 0 to 60. It has been widely used among Chinese adolescents and has demonstrated both reliability and validity [42]. The Cronbach’s alpha for the CES-DC in this study was 0.81. Perceived social support Perceived social support was assessed using items from the Global School-Based Student Health Survey [43]. Peer support was measured with a single item: “When you encounter difficulties, can you receive support and assistance from your peers?” Family support was evaluated similarly with the item: “When you encounter difficulties, can you receive support and assistance from your family?” Teacher support was assessed using the item: “When you encounter difficulties, can you receive support and assistance from your teacher?” Response options ranged from 1 = “Almost impossible to obtain,” 2 = “Possible to obtain in a few cases,” 3 = “Achievable in most cases,” to 4 = “Always achievable.” Higher summed scores indicated greater perceived availability of support across peer, family, and teacher contexts. The composite score ranged from 0 to 12. This composite score reflected overall perceived support across major interpersonal sources rather than a strictly unidimensional construct. The Cronbach’s alpha for the three-item composite was 0.82, and the factor loading of three items in the confirmatory factor analysis were all above 0.72. Covariates Several potential confounders were included in the analysis, based on prior literature and theoretical considerations [44–46], as they may be related to both exposure to ACEs/bullying victimization and depressive symptoms. These covariates included gender (male or female), age, only child status (yes or no), family structure (nuclear, single-parent, large, or other), self-perceived family socioeconomic status (high, middle, or low), parents’ education level, relationship with mother and father (poor, general, or good), self-rated academic performance (poor, general, or good), and number of friends (≤ 2, 3–5, ≥ 6). Statistical analysis All statistical analyses were conducted using SPSS for Windows (version 23.0; SPSS Inc., Chicago, IL, USA) for descriptive statistics, linear regression models, and moderation analyses. Latent class analysis (LCA) was performed using Mplus (Version 8.3; Muthén & Muthén, Los Angeles, CA, USA) to identify patterns of co-occurring ACEs and bullying victimization. Given the low proportion of missing data (< 1%), missing values were handled through multiple imputation (n = 10). Descriptive statistics were calculated to summarize demographic characteristics and key study variables. Linear regression models were used to examine the associations between ACEs, bullying victimization, and depressive symptoms. All models controlled for a set of covariates based on prior literature. We assessed multicollinearity by computing Variance Inflation Factors (VIFs) for all predictors in the regression models. All VIF values were below 2.0 (range: 1.07–1.52), well under the conventional threshold of 10, indicating that multicollinearity does not unduly influence our estimates. A stringent significance threshold of p ≤ 0.001 was applied to account for the large sample size and minimize the risk of false positives. However, this conservative threshold could increase the risk of Type II error, especially for interaction effects of modest magnitude; thus, non-significant interaction terms were interpreted with caution, alongside effect sizes and confidence intervals. Standardized β coefficients were also reported to aid interpretation of the magnitude and practical relevance of statistically significant associations. Additionally, a person-centered approach, latent class analysis (LCA), was used to capture heterogeneity in co-occurring exposure patterns of ACEs and bullying victimization. LCA examines patterns of exposure to ACEs and bullying victimization, capturing population heterogeneity based on behavioral or characteristic patterns rather than single-variable variation [13]. To determine the optimal number of classes, model fit was assessed using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), adjusted BIC (aBIC), Lo–Mendell–Rubin (LMR) test, Bootstrapped Likelihood Ratio Test (BLRT), and entropy [47]. Lower values of AIC, BIC, and aBIC indicate better model fit [48]. A significant p-value (p < 0.05) for LMR and BLRT suggests that the current model fits better than a model with one fewer class [49]. Entropy values range from 0 to 1, with values ≥ 0.6 indicating clearer class separation [50]. The final model was selected based on fit indices, interpretability, parsimony, and class prevalence, in accordance with recommendations [51]. Classes comprising less than 5% of the sample were considered insufficient for predictive utility [52]. To examine the moderating role of perceived social support, three complementary linear regression models were performed. Model 1 (Table 6) included binary indicators of ACE exposure (presence/absence), bullying victimization (presence/absence), perceived social support, all two-way interaction terms, and the three-way interaction term. This model assessed the independent and cumulative associations of ACEs and bullying victimization with depressive symptoms and whether these associations varied by social support. Model 2 (Table 7) employed a four-category combined exposure variable (neither exposure, ACEs only, bullying victimization only, both exposures), alongside interaction terms with perceived social support, to explore cumulative exposure patterns and moderation effects. Model 3 (Table 8) used LCA-derived class membership as a categorical predictor, with interaction terms between latent class membership and perceived social support added to examine whether associations with depressive symptoms differed across levels of perceived social support. Table 6Interaction Between ACE Exposure and Bullying Victimization on Depressive Symptoms: Results from linear regressionPredictorDepressive Symptoms (CES-DC Total Score)β (95%CI)p-valueStandardized βACEs Exposure5.361(4.654, 6.069)<0.0010.159Bullying Victimization5.965(5.430, 6.501)<0.0010.236Social support-2.764(-3.023, -2.506)<0.001-0.239ACEs × Bullying Victimization1.463(-0.740, 3.667)0.1930.020ACEs × Social support-2.001(-2.670, -1.333)<0.001-0.027Bullying Victimization × Social support-0.038(-0.574, 0.500)0.891-0.001ACEs × Bullying Victimization × Social support-3.911(-4.413, -3.409)<0.001-0.180β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends. Higher CES-DC scores indicate more severe depressive symptoms Interaction Between ACE Exposure and Bullying Victimization on Depressive Symptoms: Results from linear regression β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends. Higher CES-DC scores indicate more severe depressive symptoms Table 7Moderating Effect of Perceived Social Support on Associations Between ACEs, Bullying Victimization, and Depressive SymptomsPredictorDepressive Symptoms (CES-DC Total Score)β (95%CI)p-valueStandardized βSocial Support-0.811(-0.997, -0.626)<0.001-0.172Bullying victimization7.760(3.804,11.716)<0.0010.154ACEs9.412(7.200, 11.623)<0.0010.400Bullying victimization and ACEs (combined exposure)14.182(11.797, 16.567)<0.0010.526Bullying victimization × Social Support-0.248(-0.696, 0.199)0.276-0.043ACEs × Social Support-0.641(-0.882, -0.399)<0.001-0.243(Bullying victimization and ACEs) × Social Support-0.518(-0.794, -0.243)0.076-0.172β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. Interaction terms were included to test the moderating effect of perceived social support. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends Moderating Effect of Perceived Social Support on Associations Between ACEs, Bullying Victimization, and Depressive Symptoms β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. Interaction terms were included to test the moderating effect of perceived social support. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends Table 8Moderating Effect of Perceived Social Support on ACE Latent Class Associations with Depressive SymptomsPredictorDepressive Symptoms (CES-DC Total Score)β (95%CI)p-valueStandardized βSocial Support-2.274(-2.597, -1.952)< 0.001-0.197High ACEs and victimization12.498(11.250, 13.747)< 0.0010.267Victimization5.453(4.865, 6.041)< 0.0010.204Family adversity8.518(7.629, 9.406)< 0.0010.208(High ACEs and victimization) × Social Support0.111(-0.922, 1.144)0.8330.003Family adversity × Social Support-0.811(-1.404, -0.209)0.007-0.033Victimization × Social Support0.205 (-0.660, 1.069)0.6430.005β represents unstandardized regression coefficients with 95% confidence intervals (CIs), and standardized β values are also reported. Interaction terms were included to test the moderating effects of perceived social support. Latent class indicators were entered as categorical variables, with the low ACEs and victimization class serving as the reference group. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends Moderating Effect of Perceived Social Support on ACE Latent Class Associations with Depressive Symptoms β represents unstandardized regression coefficients with 95% confidence intervals (CIs), and standardized β values are also reported. Interaction terms were included to test the moderating effects of perceived social support. Latent class indicators were entered as categorical variables, with the low ACEs and victimization class serving as the reference group. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends Ethical considerations The study procedures received approval from the Biomedicine Ethics Committee of Anhui Medical University (NO. 20180083). Written informed consent was obtained from participants, their parents, and teachers. All procedures were conducted in accordance with the ethical standards outlined in the Declaration of Helsinki. Results Prevalence of ACEs and bullying victimization As shown in Table 2, a total of 65.5% of participants reported experiencing at least one ACE, with the most frequently reported experiences including family financial difficulties (26.3%), parental conflict or domestic violence (30.7%), the death of a close family member (24.2%), and emotional abuse (19.3%). Additionally, 30.0% of participants reported experiencing at least one form of bullying victimization. Among these, verbal bullying was the most prevalent (20.1%), followed by relational bullying (17.9%), physical bullying (7.6%), and cyberbullying (4.9%) (Table 2). Table 5Model fit indices and class prevalence for latent class analysis of adverse childhood experiences and bullying victimizationLLAICBICaBICEntropyBLRTVLMRCategory probability1c-34255.53168543.06168649.50068598.6572c-31161.86262389.72362609.25362504.3890.794<0.001<0.00125.63/74.373c-30650.64261401.28561733.90661575.0210.770<0.001<0.00121.58/11.44/66.984c-30273.23360680.46561126.17860913.2720.761<0.001<0.0016.53/59.70/25.02/8.755c-30096.43160360.86260919.66660652.7390.7680.0030.0033.84/6.59/21.73/9.15/58.686c-30026.48360254.96760926.86260605.9130.7790.3020.3043.88/1.47/5.75/22.40/8.47/58.04BLRT and VLMR p values indicate the statistical significance of improvement in model fit for the k-class model compared with the (k–1)-class model. Entropy reflects classification accuracy, with values closer to 1 indicating better classification quality. Category probability represents the estimated proportion (%) of participants assigned to each latent class. Lower values of AIC, BIC, and aBIC indicate better model fitLL log-likelihood, AIC Akaike Information Criterion, BIC Bayesian Information Criterion, aBIC sample-size adjusted BIC, BLRT Bootstrap Likelihood Ratio Test, VLMR Vuong–Lo–Mendell–Rubin likelihood ratio test Model fit indices and class prevalence for latent class analysis of adverse childhood experiences and bullying victimization BLRT and VLMR p values indicate the statistical significance of improvement in model fit for the k-class model compared with the (k–1)-class model. Entropy reflects classification accuracy, with values closer to 1 indicating better classification quality. Category probability represents the estimated proportion (%) of participants assigned to each latent class. Lower values of AIC, BIC, and aBIC indicate better model fit LL log-likelihood, AIC Akaike Information Criterion, BIC Bayesian Information Criterion, aBIC sample-size adjusted BIC, BLRT Bootstrap Likelihood Ratio Test, VLMR Vuong–Lo–Mendell–Rubin likelihood ratio test ACEs and depressive symptoms Emotional abuse had the strongest association with depression (standardized β = 0.300, p < 0.001; Table 3), followed by emotional neglect (standardized β = 0.225, p < 0.001). Frequent parental conflict (standardized β = 0.209, p < 0.001) and parental absence (standardized β = 0.129, p < 0.001) were also linked to higher levels of depressive symptoms. Table 3Associations Between ACEs and Depressive Symptoms: Results From Linear RegressionType of ACEsDepressive Symptoms (CES-DC Total Score)β (95%CI)p-valueStandardized βFamily financial difficulties3.728(3.095,4.360)< 0.0010.142Parents often fight or quarrel5.254(4.677,5.831)< 0.0010.209Parental adverse events3.871(3.067,4.675)< 0.0010.111Family members with serious illness (mental illness, disability, etc.)2.464(1.231,3.698)< 0.0010.046Parental absence5.183(4.250,6.115)< 0.0010.129Parental divorce1.794(0.842,2.745)< 0.0010.048Death of family member1.816(1.200,2.431)< 0.0010.067Emotional abuse (EA)8.812(8.149,9.475)< 0.0010.300Physical abuse (PA)6.460(5.558,7.362)< 0.0010.165Sexual abuse (SA)5.073(4.052,6.095)< 0.0010.112Emotional neglect (EN)10.081(9.050,11.112)< 0.0010.225Physical neglect (PN)7.297(6.286,8.308)< 0.0010.164Any ACEs Exposure (Yes vs. No)5.396(4.833,5.959)< 0.0010.222β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends. Higher CES-DC scores indicate more severe depressive symptoms Associations Between ACEs and Depressive Symptoms: Results From Linear Regression β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends. Higher CES-DC scores indicate more severe depressive symptoms Bullying victimization and depressive symptoms Linear regression analysis indicated a significant correlation between bullying victimization and depressive severity, with relational bullying exerting the strongest influence (standardized β = 0.269, p < 0.001; Table 4), followed by physical bullying (standardized β = 0.158, p < 0.001) and cyberbullying (standardized β = 0.142, p < 0.001). LCA of ACEs and bullying victimization The LCA models were compared for fit statistics across two to six classes (Table 5). The 4-class model was selected due to its lower BIC and aBIC values, and an entropy value close to 0.8 (0.761). All identified classes accounted for more than 5% of the sample: (1) a low probability of exposure to both ACEs and peer victimization, labeled low ACEs and victimization (59.70%); (2) a high probability of exposure to family-related adversities and maltreatment, labeled family adversity (25.02%); (3) a high probability of exposure to multiple forms of peer victimization with relatively low exposure to ACEs, labeled victimization (8.75%); and (4) a high probability of exposure to both ACEs and peer victimization, labeled high ACEs and victimization (6.53%) (Fig. 1). Table 2Prevalence of Adverse Childhood Experiences and Bullying Victimization in the Study SampleType of ACEs or Bullying victimizationPrevalence (%)ACEs Family financial difficulties26.3 Parents often fight or quarrel30.7 Parental adverse events12.5 Family members with serious illness (mental illness, disability, etc.)4.8 Parental absence9.1 Parental divorce11.0 Death of family member24.2 Emotional abuse (EA)19.3 Physical abuse (PA)9.6 Sexual abuse (SA)7.1 Emotional neglect (EN)7.2 Physical neglect (PN)7.3 Any ACE exposure65.5Bullying Victimization Verbal bullying20.1 Relational bullying17.9 Physical bullying7.6 Cyber bullying4.9 Any type of bullying victimization30.0Values represent prevalence (%). ACEs and bullying victimization were assessed as binary variables (yes/no). Any ACE exposure and any type of bullying victimization indicate exposure to at least one corresponding typeAbbreviations ACEs adverse childhood experiences, EA emotional abuse, PA physical abuse, SA sexual abuse, EN emotional neglect, PN physical neglect Prevalence of Adverse Childhood Experiences and Bullying Victimization in the Study Sample Values represent prevalence (%). ACEs and bullying victimization were assessed as binary variables (yes/no). Any ACE exposure and any type of bullying victimization indicate exposure to at least one corresponding type Abbreviations ACEs adverse childhood experiences, EA emotional abuse, PA physical abuse, SA sexual abuse, EN emotional neglect, PN physical neglect Latent class profiles of adverse childhood experiences and bullying victimization among adolescents. Note: The figure displays the estimated probabilities of each ACE and bullying victimization indicator for the four latent classes identified by latent class analysis (LCA). The y-axis represents the probability of endorsement for each item, while the x-axis lists individual ACE and bullying indicators. The four classes include: High ACEs and victimization (6.53%), Low ACEs and victimization (59.70%), family adversity (25.02%), and Victimization (8.75%), with percentages indicating the prevalence of each latent class in the sample. Higher values indicate greater likelihood of exposure to the corresponding adversity or victimization type within each class Relationship between ACEs, bullying victimization, and depressive symptoms: the moderating role of social support As shown in Table 6, both ACEs and bullying victimization were positively associated with depressive symptoms, whereas social support demonstrated a negative relationship with depressive symptoms, supporting the independent effects model. Furthermore, social support significantly moderated the relationship between ACEs and depressive symptoms (standardized β = − 0.027, p < 0.001). In contrast, the interaction between bullying victimization and social support was not statistically significant (standardized β = − 0.001, p = 0.891). The interaction term between ACEs and bullying victimization on depressive symptoms was also not significant, indicating that their associations with depressive symptoms were cumulative rather than multiplicative in this regression model. Additionally, the three-way interaction between ACEs, bullying victimization, and social support was statistically significant (standardized β = − 0.180, p < 0.001). Table 7 presents linear regression results showing that both ACEs (standardized β = 0.400, p < 0.001) and bullying victimization (standardized β = 0.154, p < 0.001), as well as the combination of both, were significantly associated with higher depressive symptom levels. Adolescents exposed to both ACEs and bullying victimization had the highest depressive symptom scores, with their CES-DC scores being 14.18 points higher than those of adolescents with no exposure (β = 14.182, 95% CI: 11.797–16.567, p < 0.001), representing the strongest association among all exposure groups. The interaction term ACEs × social support (standardized β = − 0.243, p < 0.001) was statistically significant. In contrast, the interaction between bullying victimization and social support was not statistically significant (standardized β = − 0.043, p = 0.276), and the interaction between the combined ACEs-and-bullying exposure group and social support was not significant (standardized β = − 0.172, p = 0.076). Table 4Associations Between Bullying Victimization and Depressive Symptoms: Results from linear regressionType of Bullying VictimizationDepressive Symptoms (CES-DC Total Score)β (95%CI)p-valueStandardized βVerbal bullying6.555(5.911,7.199)< 0.0010.227Relational bullying8.109(7.444,8.775)< 0.0010.269Physical bullying6.916 (5.921,7.912)< 0.0010.158Cyber bullying7.596(6.385,8.808)< 0.0010.142Any Bullying Victimization (Yes vs. No)7.075(6.518,7.633)< 0.0010.280β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends. Depressive symptoms were assessed using the CES-DC, with higher scores indicating greater symptom severity Associations Between Bullying Victimization and Depressive Symptoms: Results from linear regression β represents unstandardized regression coefficients with 95% confidence intervals (CIs); standardized β indicates standardized coefficients. All models were adjusted for age, sex, only-child status, family structure, self-perceived family socioeconomic status, parental education, parent–child relationship, academic performance, and number of friends. Depressive symptoms were assessed using the CES-DC, with higher scores indicating greater symptom severity To further examine whether the relationship between latent class membership and depressive symptoms varied by levels of perceived social support, interaction terms between social support and LCA-derived classes were included in linear regression models (Table 8). Compared to the low ACEs and victimization class, adolescents in the high ACEs and victimization class (standardized β = 0.267, p < 0.001), the family adversity class (standardized β = 0.208, p < 0.001), and the victimization class (standardized β = 0.204, p < 0.001) exhibited significantly higher depressive symptom levels. Perceived social support was independently associated with lower depressive symptom scores (standardized β = − 0.197, p < 0.001). A significant interaction was found between social support and the family adversity class (standardized β = − 0.033, p = 0.007). In contrast, the interaction terms between social support and both the high ACEs and victimization class and the victimization class were not statistically significant (p > 0.05). Discussion This study, based on a sample of 5,724 adolescents, investigated the associations between ACEs, bullying victimization, perceived social support, and depressive symptoms. The findings revealed that both ACEs and bullying victimization were prevalent and independently associated with higher levels of depressive symptoms. Adolescents exposed to both forms of adversity exhibited the highest depressive symptom levels, and LCA identified a high ACEs and victimization subgroup with elevated depressive symptoms. Additionally, higher perceived social support was associated with a weaker relationship between ACEs and depressive symptoms, though this pattern was not observed for bullying victimization. These results contribute to existing literature by demonstrating that family-based and peer-based adversities may cluster into distinct exposure profiles, and that perceived social support does not operate uniformly across different adversity contexts [10, 12]. Both ACEs and bullying victimization were linked to higher depressive symptoms, although the associations varied by subtype. Emotional abuse and emotional neglect showed the strongest associations with ACEs, aligning with previous research indicating that emotionally adverse caregiving may particularly influence adolescent internalizing symptoms through negative self-evaluation, interpersonal insecurity, and impaired emotion regulation [8, 53]. From a neurobiological perspective, chronic emotional maltreatment has been associated with persistent dysregulation of the HPA axis, a core stress–response system, potentially increasing vulnerability to depressive symptoms [30]. Among bullying forms, relational bullying exhibited the strongest association, likely because peer exclusion, rumor spreading, and social manipulation directly threaten belonging and social status during adolescence, and may be less detectable by adults [25, 28]. Notably, relational victimization also affects HPA axis functioning, particularly in response to social-evaluative threats, suggesting a unique neurobiological pathway linking peer-based adversity to depression [27]. Overall, family-based and peer-based adversities likely contribute to depressive symptoms through distinct emotional-security and peer-social pathways. The co-occurrence of ACEs and bullying victimization was linked to the highest depressive symptom burden. Rather than simply indicating two separate risk factors, this finding suggests that family-based adversity and peer-based victimization often cluster within the same adolescents, corresponding to a more severe depressive symptom profile [10, 12]. Adolescents facing both types of adversity may experience stress across multiple interpersonal contexts, leaving fewer safe relational spaces for emotional recovery. This pattern supports cumulative associations between family and peer adversity, suggesting that co-occurring stressors may coincide with a greater psychological burden than either exposure alone [54]. However, the interaction term between ACEs and bullying victimization was not statistically significant in the regression model. Therefore, the present findings should be interpreted as evidence of cumulative associations rather than a formal statistical interaction. It is crucial to distinguish variable-centered interaction analyses from person-centered latent class analyses [13, 51]. The non-significant interaction term suggests that this study did not identify a formal statistical interaction between ACEs and bullying victimization. In contrast, the LCA identifies subgroups characterized by co-occurring adversities. Hence, the elevated depressive symptoms observed in the high ACEs and victimization class should be interpreted as reflecting the co-occurrence of adversities across family and peer contexts, rather than suggesting that one exposure causally intensified the association of the other with depressive symptoms. From an intervention standpoint, this cumulative pattern implies that ACEs and bullying victimization should be treated as distinct, yet often co-occurring, risk factors. Higher perceived social support was associated with lower depressive symptoms overall, but its moderating effect was selective rather than uniform. In line with the stress-buffering model, social support may be particularly effective when it provides coping resources, emotional reassurance, or compensatory relational experiences in the face of adversity [16, 17]. In variable-centered analyses, perceived social support significantly moderated the association between ACEs and depressive symptoms, whereas it did not significantly moderate the relationship between bullying victimization and depressive symptoms. In person-centered analyses, perceived social support was independently associated with lower depressive symptoms, but the significant moderation was observed only for the family adversity class, not for the victimization class or the high ACEs and victimization class. These findings suggest that the relevance of perceived social support may vary across adversity profiles, rather than uniformly across all exposure types. This pattern can be explained by differences in adversity context. ACEs and bullying victimization differ in several theoretically important ways, including their context, chronicity, and perceived controllability, which may lead to differential moderation by social support. When adversity is rooted in the family domain, support from teachers, peers, or other family members may provide alternative relational experiences, emotional validation, and a sense of security [16, 24]. In contrast, when victimization occurs in peer settings, the same social environment that could potentially offer support may also serve as the source of stress. Bullying victimization may involve interpersonal rejection, shame, fear of negative evaluation, and distrust toward peers, limiting adolescents’ willingness or ability to seek support [25, 55]. Even when support is theoretically available, it may not be mobilized effectively to counteract the psychological burden of peer victimization. Furthermore, measurement considerations are important when interpreting these findings. Conceptually, the composite social support score captures adolescents’ overall perception of support availability across multiple interpersonal contexts; however, peer, family, and teacher support may represent distinct protective resources and may not be fully interchangeable [16, 26]. As such, a general support score may be less sensitive to whether support is contextually matched to the source of stress. For example, family or teacher support may not directly repair damaged peer status or reduce ongoing peer exclusion, which may partly explain why perceived social support appeared more relevant in the context of family adversity than in the context of bullying victimization. More broadly, these findings may suggest a potential severity gradient in the buffering relevance of social support. Although this interpretation should be considered cautiously due to the cross-sectional design, one plausible explanation is that the protective association of general social support may have limited buffering capacity. Social support may be more relevant when adversity is concentrated in a single domain but less sufficient when adversity is chronic, cumulative, or spans both family and peer contexts. This interpretation aligns with Hobfoll’s Conservation of Resources theory, which posits that the buffering function of resources weakens when stress exposure is severe and resource demands exceed available coping capacity [56, 57]. Accordingly, social support should not be seen as a universal protective factor; rather, its relevance may depend on the type, source, and clustering of adversity exposure. The findings of this study have important implications for identifying and supporting adolescents with elevated depressive symptoms. Given the association of both ACEs and bullying victimization with increased depressive symptoms, screening for these adversities in school and community settings may help identify adolescents in need of further mental health evaluation and intervention. Furthermore, the observed interaction between perceived social support and ACE exposure suggests that social support may play a critical role in mitigating depressive symptoms related to family adversity. In contrast, the lack of a similar interaction for bullying victimization indicates that addressing bullying-related depressive symptoms may require more targeted, school-based interventions, including anti-bullying policies, teacher training, and trauma-informed support strategies. Programs such as the KiVa and Olweus Bullying Prevention Program have proven effective in reducing peer victimization and improving school climate in previous research [4, 58, 59]. Future prevention efforts in China may benefit from exploring the cultural adaptation of these programs to local school environments, taking into account class-based management structures, teacher-led pastoral systems, and variations in peer interaction norms. In conclusion, these findings highlight the importance of multi-level prevention and intervention strategies that differentiate between family adversity, peer victimization, and their co-occurrence. Despite the strengths of a large sample and rigorous analyses, several limitations must be acknowledged. First, the cross-sectional design precludes causal inference and does not establish temporal relationships between ACEs, bullying victimization, perceived social support, and depressive symptoms. Second, all variables were assessed through self-report, which may introduce recall bias and social desirability bias. Although the participants were early adolescents, certain subjective and family-related adversities may have been under-reported due to limited self-awareness or reluctance to disclose sensitive experiences. Third, the entropy of 0.761 of LCA model indicates moderate classification uncertainty, which could cause attenuation bias and underestimate standard errors in regression. Additionally, the sample was drawn from junior high school students in a single province in China, which may limit the generalizability of the findings. Moreover, residual confounding cannot be excluded. Perceived social support was measured using three single-item indicators across peer, family, and teacher domains; while these were combined into an overall support score, this approach may obscure source-specific effects. Future research should utilize longitudinal designs, multidimensional support measures, and broader assessments of adversity. Conclusions This study demonstrated that both ACEs and peer bullying victimization were common among Chinese adolescents and independently associated with elevated depressive symptoms. Adolescents exposed to both ACEs and bullying victimization exhibited the highest depressive symptom levels, supporting a cumulative association between family and peer adversity. Furthermore, perceived significantly moderated the relationship between ACEs and depressive symptoms, while no significant buffering effect was observed for bullying victimization. These findings suggest that the impact of perceived social support on depressive symptoms may differ depending on the context of adversity. Given the cross-sectional design, these results should be interpreted as associative rather than causal. Nevertheless, they highlight the need for differentiated prevention and intervention strategies, including the strengthening of supportive family and school relationships and the implementation of targeted, school-based responses to bullying victimization. Supplementary Information Supplementary Material 1. Acknowledgements We extend our sincere gratitude to all the researchers and participants for their invaluable support in data collection for this study. Abbreviations Adverse childhood experiences Emotional abuse Physical abuse Sexual abuse Emotional neglect Physical neglect Latent class analysis Hypothalamic—Pituitary—Adrenal Axis Post—Traumatic stress disorder Odds ratio Confidence interval Regression coefficient Standard deviation Akaike information criterion Bayesian information criterion Adjusted Bayesian information criterion Lo—Mendell—Rubin Likelihood Ratio Test Bootstrapped likelihood ratio test Authors’ contributions Qingyuan Liu, Yuan Li conceived the study, analyzed the data, and drafted the paper. Cong Wang, Yunyun Zhao, and Wanqi Ren: Collected the data and completed the initial analyses and interpretation. Puyu Su, Gengfu Wang and Qunshan shen: Conceived the study, completed the initial analyses and interpretation and revised and reviewed the paper. All authors read and approved the final manuscript. Funding This study was funded by the grants from the National Natural Science Foundation of China (Grant No. 82204071, 82473655, 81874268 and 82173539), Natural Scientific Research Priority Project of Anhui Higher Education Institution (2025AHGXZK30203), Project of MOE Key Laboratory of Population Health Across Life Cycle (JK20245). Data availability The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. Declarations Ethics approval and consent to participate The study procedures received approval from the Biomedicine Ethics Committee of Anhui Medical University (NO. 20180083). The participants were fully informed of the purpose of the study and were invited to participate voluntarily. We obtained written informed consent from the participants, their parents and teachers. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Qingyuan Liu and Yuan Li contributed equally to this work. Contributor Information Gengfu Wang, Email: wanggenfu@ahmu.edu.cn. Qunshan Shen, Email: drshen@ahmu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.