Introduction

Non-suicidal self-injury (NSSI) is defined as the socially unacceptable behavior of an individual causing deliberate, repetitive harm to their own body without the purpose of suicide1. Cutting, biting, hitting, and scratching the skin are common NSSI behaviors2. NSSI usually begins in early adolescence, increases throughout adolescence, and gradually decreases in adulthood3. A meta-analysis showed that 22.1% of adolescents worldwide engage in NSSI behavior4. In adolescents with psychiatric disorders such as major depressive disorder (MDD), prevalence ranges from 30% to 80%5,6. A more recent study discovered that over 53.2% of individuals with first-episode MDD engaged in NSSI7. In addition to causing physical harm, NSSI is associated with a variety of adverse outcomes, including cognitive impairment, poor interpersonal relationships, and potential involvement in violent offenses8,9. Additionally, NSSI significantly predicts suicidal behavior10,11, making it a significant public mental health issue. Thus, it is urgent to further explore the modifiable risk factors for NSSI for the early identification of at-risk populations and the implementation of timely preventive measures.

The reasons for adolescent NSSI are intricate and not entirely understood, involving a multitude of factors, including various sociodemographic aspects (e.g., gender), family adversity12, emotion regulation13, and psychiatric and psychological factors14. A growing body of research has linked environmental risk factors, particularly adverse childhood experiences (ACEs), with the occurrence, development, and prognosis of MDD15 and NSSI16. ACEs are potentially traumatic events that occur in childhood17, ranging from emotional, physical, or sexual abuse and neglect to parental separation or divorce and family or community violence18,19,20. ACEs, serving as distant risk factors, may precipitate NSSI behaviors among adolescents with MDD21,22,23. A multicenter study of 833 adolescent patients with depression reported that those who had experienced abuse and neglect were more than twice as likely to exhibit NSSI behavior24. In addition, ACEs exhibit a cumulative effect; individuals with 1–2 ACEs are 2.2 times more likely to engage in NSSI than those with no ACEs, and there is a 6-fold greater likelihood of engaging in NSSI with exposure to more than three ACEs25,26. ACEs have been confirmed to not only directly affect NSSI but also indirectly through certain mediating factors (e.g., emotional dysregulation and psychological resilience)27,28. However, the physiological regulatory pathways through which ACEs influence NSSI in adolescent patients with MDD are poorly understood.

One possibility is that ACEs shape mental health risks through biological rhythm changes. Biological rhythms pertain to cyclical variations in physiological and behavioral functions, including the sleep-wake cycle, eating patterns, social activity, hormone secretion, and other important bodily functions modulated by the circadian clock29,30. Individuals with ACEs have an increased risk of circadian rhythm anomalies31,32. Perturbation of biological rhythms by external environmental factors or internal regulatory imbalances influences brain development33, resulting in an increased incidence of affective disorders and risk-taking behaviors34. NSSI is a common risk-taking behavior linked to biological rhythm disorder (BRD). Researchers have proposed that individuals with BRD, such as those with irregularity in the sleep-wake cycle and high adherence to unhealthy eating pattern rhythms, are at greater risk of intense NSSI urges and behaviors35,36. In addition, cumulative clinical and biological studies have shown that persons with BRD are 1.2 to 1.8 times more likely to develop MDD, and disturbance of biological rhythms is associated with more severe depressive symptoms37,38. Furthermore, the severity of depressive symptoms significantly increases the level of self-harm in patients with MDD38.

Based on these results, there may be a relationship among ACEs, BRD, depression severity, and NSSI. However, the relationships among these factors have been under-studied, especially in adolescents with MDD, and it is unclear whether specific risk factors indirectly affect NSSI behaviors. The purpose of the present study was to explore the relationships among ACEs, BRD, depression severity, and NSSI. We hypothesized that ACEs affect NSSI by mediating the effects of BRD and depression severity.

Methods

Participants

A total of 270 adolescents with first-episode depression were recruited from the psychiatry outpatient clinic of the General Hospital of Ningxia Medical University, China. All patients were diagnosed by two psychiatrists using the Chinese version of the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V).

The inclusion criteria were (1) age between 10 and 19 years, (2) meeting the diagnostic criteria for MDD according to the DSM-V, (3) first episode with no prior antidepressant or antipsychotic treatment, and (4) possessing the ability and intelligence to comprehend and complete questionnaires. Exclusion criteria were as follows: (1) current or previous history of other psychiatric disorders, such as bipolar disorder; (2) severe physical illness or brain injury, such as epilepsy, liver or kidney disease, or heart disease; and (3) suspected or confirmed history of alcohol or drug abuse. After screening out incomplete questionnaires, 257 valid responses were obtained, yielding a retention rate of 95.2%.

Ethics approval and consent to participate

The study protocol was approved by the Ethics Committee of the General Hospital of Ningxia Medical University (ethics approval number: KYLL-2024-0051) and performed in compliance with the Declaration of Helsinki. Written informed consent was obtained from all participants and/or their legal guardians before the study commenced.

Measures

Adverse childhood experiences

ACEs were measured using the Chinese version of the Adverse Childhood Experiences Questionnaire (ACEQ)39,40. It has undergone cross-cultural validation in samples of adolescents from mainland China and has demonstrated acceptable validity and reliability, with Kuder-Richardson 20 and Guttman split-half reliability being 0.830 and 0.750, respectively40. The questionnaire consists of 14 categories covering emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, interparental violence, household substance abuse, household mental illness, criminal behavior, parental divorce, low socioeconomic status (SES), peer victimization, peer isolation/rejection, and exposure to community violence. Each ACE was coded dichotomously based on the presence or absence of each experience in the participant’s childhood. A positive response to a specific category indicated the occurrence of the corresponding ACE. An overall ACE score ranging from 0 to 14 was created by summing the number of ACE assessment categories that received a positive response. The Cronbach’s alpha coefficient of the ACEQ in this study was 0.731.

Biological rhythm disorder

We used the Self-Rating of Biological Rhythm Disorder for Adolescents (SBRDA) to assess BRD in the participants41. The SBRDA is an original Chinese-language scale specifically designed for Chinese adolescents, which demonstrates good reliability and validity, with a Cronbach’s α coefficient of 0.95. It comprises 29 items that include four dimensions: sleep rhythm, eating habit rhythm, activity rhythm, and digital media use. Participants respond using a 5-point Likert scale (1 = completely inconsistent, 5 = completely consistent), and the total rhythm disorder scores range from 29 to 145. The higher the questionnaire score, the more severe the self-rating of biological rhythm disorders. We used the 90th percentile of the total scores as the cutoff point to indicate the presence of BRD among adolescents according to the Chinese norm. Cronbach’s alpha coefficient of the SBRDA in this study was 0.840.

Severity of depression

The severity of depression was assessed using the 24-item Hamilton Depression Rating Scale (HAMD-24)42. This scale evaluates depression across seven domains: anxiety/somatization, weight, cognitive impairment, diurnal variation, retardation, sleep disturbances, and hopelessness. Each of the 24 items is rated on a five-point Likert scale, with higher scores indicating more severe depression. If the total score is greater than 20, the respondent is considered to be at risk of depression. The examiners were experienced psychiatrists who received consistent training before the study began.

Non-suicidal self-injury

NSSI was assessed using the Adolescent Non-suicidal Self-Injury Assessment Questionnaire (ANSAQ)43. The ANSAQ comprises behavioral and functional sections. This study employed the behavioral segment, which contains 12 questions that assess two dimensions: self-injurious behavior with significant tissue damage and self-injurious behavior without significant tissue damage. The questionnaire uses a five-point Likert scale, whereby scores of 0 − 4 correspond to frequencies of “none,” “occasionally,” “sometimes,” “often,” and “always,” respectively. Elevated scores suggest increased severity of NSSI behaviors. The Cronbach’s alpha value of the ANSAQ in this study was 0.953.

Emotional regulation

The Chinese version of the Regulatory Emotional Self-Efficacy Scale (RESE) was used to assess self-efficacy in emotion regulation among adolescents with first-episode depression44. This scale includes 12 items that assess three dimensions: perceived self-efficacy in expressing positive affect (POS), managing despondency/distress (DES), and managing anger/irritation (ANG). It utilizes a five-point Likert scale (1 = very uncharacteristic, 5 = very characteristic). Higher scores indicate greater self-efficacy in emotion regulation. The Cronbach’s alpha coefficient of the RESE in this study was 0.826.

Statistical analysis

Participants’ characteristics were described by means, standard deviations (SD), and percentages. Pearson’s correlations were used to investigate the relationships between ACEQ, SBRDA, HAMD, and ANSAQ scores. Statistical significance was set at P < 0.05.

When examining the mediating effects of BRD and depression severity on the relationship between ACEs and NSSI, gender and age were included as basic demographic covariates in sequential mediation models. RESE was also included, given its protective role against NSSI and association with internalizing problems (e.g., depression)45. Since POS reflects a unique domain of emotional efficacy—one that differs from the other two dimensions of RESE that focus on negative emotion management—its specific influence on the research variables cannot be fully captured or substituted by the total RESE score. Thus, POS was also considered as a covariate. Furthermore, to avoid the interference of multicollinearity, two separate models were designed: Model 1 adjusted for gender, age, and RESE; Model 2 adjusted for gender, age, and POS.

Regression coefficients were calculated for each mediation model. Bootstrapping was performed using 5000 resamples with 95% confidence intervals to analyze the significance of the mediation models. Effects with a confidence interval (CI) excluding zero were considered statistically significant. All statistical analyses were performed using SPSS version 27.0 and PROCESS Macro 4.1 (IBM, Chicago, IL, USA).

Results

Participant characteristics

Of the 257 adolescents with first-episode depression, the mean age and standard deviation were 15.66 ± 2.035 years. 186 (72.4%) were female, and the mean nightly sleep time and standard deviation were 5.61 ± 1.575 h. Due to the high proportion of NSSI behaviors among adolescents with first-episode depression, the analyses were performed as continuous variables. Moreover, continuous scores enable the detection of subtle variability, leading to greater precision and improved statistical power. Table 1 presents the characteristics of the variables.

Table 1 Descriptive statistics of the variables.
Full size table

Common method bias test

To further improve the scientific rigor of this study, a common method bias test utilizing Harman’s one-factor approach without rotation revealed the presence of 23 factors with eigenvalues greater than 1. Notably, the variance explained by the first factor amounted to 22.46% (< 40%)46. These findings indicate the absence of significant common method bias in the present study.

Associations between ACEs, BRD, severity of depression, and NSSI

Figure 1 presents the results of correlation analyses. ACEs, BRD, severity of depression, and NSSI were positively and significantly correlated with each other. Specifically, ACEs were positively associated with BRD (r = 0.368, P < 0.001), severity of depression (r = 0.376, P < 0.001), and NSSI (r = 0.404, P < 0.001). BRD was positively associated with severity of depression (r = 0.365, P < 0.001) and NSSI (r = 0.304, P < 0.001). Severity of depression was also positively associated with NSSI (r = 0.546, P < 0.001). Additionally, RESE was negatively associated with ACEs (r = −0.220, P < 0.001), BRD (r = −0.208, P < 0.001), severity of depression (r = −0.357, P < 0.001), and NSSI (r = −0.356, P < 0.001).

Fig. 1
figure 1

Pearson correlation analyses between the variables. RESE: The Regulatory Emotional Self-Efficacy Scale; ACEQ: The Adverse Childhood Experiences Questionnaire; SBRDA: The Self-Rating of Biological Rhythm Disorder for Adolescents; HAMD: Hamilton Depression Rating Scale; ANSAQ: The Adolescent Non-suicidal Self-Injury Assessment Questionnaire.

Full size image

Sequential mediating effects between ACEs, BRD, severity of depression, and NSSI

The results of regression analysis of the mediation model after adjusting for the effects of gender, age, and RESE (i.e., Model 1) are shown in Fig. 2A. Regarding ACEs, the paths from ACEs to BRD (β = 0.322, P < 0.001) and HAMD (β = 0.200, P < 0.001) were positive and significant. However, there was no significant correlation between BRD and NSSI (β = 0.112, P = 0.061) after adjusting for HAMD as a mediator. Table 2 presents the bootstrap analysis of the sequential mediation effects. There was a significant mediating effect of HAMD on the relationship between ACEs and NSSI (95% CI: 0.074–0.389), but the effect of BRD on the relationship between ACEs and NSSI was not significant (95% CI: −0.023–0.413). The total effect of ACEs on NSSI was significant (β = 0.323, 95% CI: 1.071–2.193). ACEs indirectly affected the NSSI, with a sequential mediating effect of BRD and HAMD (95% CI: 0.025–0.136).

Fig. 2
figure 2

Sequential mediation model of the effect of BRD and depression severity on ACEs and NSSI. A: after adjusting for the effects of gender, age, and regulatory emotional self-efficacy. B: after adjusting for gender, age, and perceived self-efficacy in expressing positive affect. *** P < 0.001, * P < 0.05. ACE: Adverse Childhood Experiences; BRD: Biological Rhythm Disorder; HAMD: Hamilton Depression Rating Scale (depression severity); NSSI: Non-suicidal Self-Injury. The pathway denoted by a dashed line is of no significance.

Full size image
Table 2 The sequential mediating effects among variables.
Full size table

Interestingly, compared with Model 1, adjusting for POS in Model 2 enhanced the mediating effect of BRD, which became statistically significant (95% CI: 0.043–0.537). A detailed description of the mediation model is presented in Fig. 2B, and a comparison of the differences in mediating effects between the two models is provided in Table 2.

Discussion

In this clinical sample of 257 adolescents with first-episode depression, ACEs, BRD, and depression severity were significantly and positively associated with NSSI. Furthermore, the effect of ACEs on NSSI was mediated by BRD and depression severity after adjusting for gender, age, and perceived self-efficacy in expressing positive affect.

The current study confirms that ACEs are strongly associated with a higher incidence of NSSI behaviors among adolescents with MDD, which is consistent with previous studies27,47. One potential mechanism is that ACEs contribute to the development of negative coping strategies in adolescents such as avoidance, denial, and suppression48. The experience avoidance model proposes that self-injurious behaviors primarily facilitate the avoidance of undesired or aversive internal states encompassing bodily sensations and emotional experiences in adolescents49. In this context, NSSI may be understood as a modality through which adolescents articulate their underlying psychological distress, solicit attention, or escape from the realities they confront. Furthermore, this study aligns with prior research that has established a positive correlation between ACEs and both the severity of depression50 and NSSI behavior51. Additionally, our study revealed significant positive correlations between BRD and ACEs, depression severity, and NSSI. This interplay could increase the complexity and difficulty of addressing NSSI behaviors among adolescents with depressive disorders. Therefore, it is necessary to comprehensively consider ACEs, BRD, and the degree of depression to develop personalized treatment plans.

The present study is the first to explore whether BRD mediates the relationship between ACEs and NSSI, offering deeper insights into the mechanisms by which ACEs increase the risk of NSSI. Furthermore, the results of the sequential mediation model showed that the direct effect of ACEs on NSSI was significant, and the mediation effect (ACE → BRD → severity of depression → NSSI) was also significant, suggesting that for adolescents with first-episode depressive disorders, ACEs not only directly cause NSSI, but also mediate this risk through BRD and depression severity. These findings significantly contribute to our understanding of the pathophysiology of NSSI. Adolescents experience critical physiological and psychological development52, and instability in biological rhythms may exacerbate difficulties in emotion regulation. Logan & McClung53 revealed a close link between BRD and depressive symptoms, potentially through effects on neurotransmitter function and endocrine system balance54. Consistent with previous studies, our study indicates that adolescents who experience ACEs are more prone to BRD than are those without ACEs, which subsequently increases the risk of depressive symptoms. The findings indicate that integrating BRD assessment into MDD and NSSI screening for adolescents enables the targeted identification of high-risk individuals with positive BRD status. Such identification underpins the delivery of timely interventions, which ultimately prevents the initiation or progression of MDD and NSSI. Depression severity is another mediator that further contributes to the relationship between ACEs and NSSI. Depressive symptoms may lead individuals to engage in self-negating thoughts and behaviors while also reducing their ability to cope with difficulties and challenges55. We found that adolescents with higher levels of depression were more likely to exhibit NSSI behaviors potentially related to self-identity crises, social isolation, and disruptions in coping mechanisms induced by depressive symptoms56. By identifying the two intermediary factors between ACEs and NSSI (i.e., BRD and depression severity), this study offers novel perspectives on the prevention and treatment of depressive disorders and NSSI among adolescents. For instance, BRD can be rectified through improvements in sleep quality and adjustments to daily schedules, whereas the severity of depression can be mitigated by combining antidepressants with psychological therapies, thereby reducing the risk of NSSI.

One interesting finding of the present study is that after adjusting for gender, age, and RESE, the effect of BRD on the relationship between ACEs and NSSI was not significant (95% CI: −0.023–0.413). However, when further adjusting for POS, the aforementioned chained mediation pathway effects were all significant. The core reason for this shift may lie in that POS, as a confounding factor, masks the true mediating role of BRD. Individuals with high self-efficacy in regulating positive emotions are more likely to employ active coping mechanisms in response to stress and negative experiences57,58. This ability to actively regulate emotions may buffer the impact of ACEs on biological rhythms and depression59,60, thereby weakening the chain-mediating effect in the absence of positive emotion regulation self-efficacy. These findings underscore the complexity and importance of emotion regulation in maintaining mental health. From a clinical perspective, the model-dependent mediating effect of BRD implies that interventions for depressed adolescents with ACEs exposure should adopt a “dual-target” strategy: stabilize biological rhythms (i.e., target the BRD-mediated pathway) via approaches like sleep hygiene training and light therapy61,62, while enhancing POS through emotion regulation techniques such as positive emotion journaling63.

This study has some limitations. First, the sample size was relatively modest, and the findings should be interpreted and generalized to a larger population in the future. Second, given that all participants were recruited exclusively from a single tertiary hospital in Ningxia, the sample’s representativeness was insufficient, although some patients were from neighboring provinces and autonomous regions such as Inner Mongolia, Shaanxi, and Gansu. Future studies were suggested to conduct multicenter research or include cross-cultural samples, which would further expand the sample coverage and thereby improve the representativeness and external validity of the research findings. Third, this was a cross-sectional study, which precluded the establishment of causal relationships between variables, and it remains unclear how these factors (i.e., ACEs, BRD, severity of depression, and NSSI) would change over time in response to the intervention, and whether these factors are stable or state based. Future cohort studies are recommended to clarify both the causal associations and the patterns of change. Fourth, the severity of depression was analyzed using clinician rating scales, whereas ACEs, BRD, and NSSI were assessed using self-rating scales. Bias may arise from the nature of the self-rating scales: participants might underreport or misrecall experiences related to ACEs, BRD, or NSSI, which could slightly compromise the accuracy of these measured variables. Additionally, the Chinese version of the ACEQ may have potential biases in cultural adaptability. However, these scales typically show very good reliability and validity, and Harman’s one-factor approach showed no common method bias. Finally, this study only focuses on the overall mediating role of BRD. Subsequent research can further explore which specific type of BRD intervention (e.g., sleep rhythm, eating habit rhythm, activity rhythm) is most convenient and effective for depressed adolescents with high ACEs exposure.

In conclusion, our findings indicate that BRD and depression severity partially mediate the relationship between ACEs and NSSI in adolescents with first-episode depression. Improving biological rhythms in adolescents with MDD who experience ACEs could help decrease the severity of depression and prevent them from engaging in NSSI behaviors.