Introduction
Nonsuicidal self-injury (NSSI) refers to deliberately damaging one’s body tissue without lethal intent; NSSI typically involves cutting, scratching, burning, and banging [
1]. Epidemiological research consistently indicates higher prevalence rates of NSSI in teenagers than in adults [
2]. The overall lifetime prevalence of NSSI is 19.4–26.7% among adolescents, and it is more common in girls than in boys (risk ratio 1.72) [
3,
4]. NSSI is increasingly recognized as a significant public health concern because of its high prevalence and its association with several internalizing and externalizing disorders [
5] and NSSI is considered to be a strong predictor of suicidal behaviour [
6]. Identifying risk factors for adolescent NSSI is critical to understanding its mechanism and providing early prevention and treatments.
Growing research focuses on the potential mechanisms of child maltreatment (CM) experiences leading to NSSI. As distal risk factors for NSSI [
1], CM includes five types: emotional abuse (EA), physical abuse (PA), sexual abuse (SA), emotional neglect (EN), and physical neglect (PN). A recent meta-analysis showed that CM and its subtypes are associated with NSSI, with the exception of childhood emotional neglect [
7]. CM is generally considered a risk factor for many psychopathological and behavioural problems. A substantial amount of research has demonstrated that CM experiences are associated not only with higher rates of NSSI [
8‐
10] but also with depression [
11] and difficulty in emotion regulation (DER) [
12]. Moreover, recent data also suggest that these two variables have mediating properties in the relationship between CM and NSSI. Brown et al. [
9] found a partial mediating effect of EA and full mediating effects of SA and PN by depression on NSSI. Similarly, Titelius et al. [
13] found that DER mediated the relationship between CM and NSSI frequency in a small clinical sample. Finally, DER partially mediated the effects of an invalidating family environment (a latent factor consisting of CM) and NSSI [
14]. Thus, depressive symptoms and DER may indeed be important factors explaining the pathway from CM to self-injury. However, these proposed mediators often cooccur, and fewer studies have tested mediating effects when all variables are assessed simultaneously. Shenk et al. [
15] found that when all variables (posttraumatic stress symptoms, depressive symptoms, and psychological dysregulation) were included in the same model, only posttraumatic stress symptoms mediated the relationship between CM and self-injury. Moreover, moderating factors such as sex and how CM and NSSI is measured also play a role between CM and NSSI [
16]. Therefore, the multifactorial mediating mechanism between CM and self-injury needs further research to complement the current literature.
Multiple theoretical models of NSSI have been proposed to explain self-injury. However, numerous theories and related studies have converged upon defining NSSI as a problem of emotion dysregulation. Research focusing on the function of NSSI suggests that adolescents may engage in NSSI to reduce negative affect [
17] and avoid unwanted emotions [
18]. These findings are consistent with Linehan’s Biosocial Model of emotion dysregulation [
19] postulates that invalidating environments contribute to deficits in emotion-regulating capacities, which increase the likelihood of engaging in NSSI behaviour to cope with distress. DER arises from biological anomalies combined with exposure to dysfunctional environments throughout development [
20]. This theoretical model corresponds to a broader literature in which CM is frequently identified as an invalidating environment associated with various poor outcomes, including DER, depression, and NSSI [
8,
10,
11]. To extend Linehan’s Biosocial Model above, Gratz and Roemer propose the “difficulties in emotion regulation” (DER) model [
21], which addresses several deficiencies in emotion regulation (e.g., impulse control difficulties, self-perceived limited access to strategies, a lack of emotional clarity). Recent research has found that these aspects of DER are mechanisms of change in DER-model-based treatment for adolescents with NSSI disorder [
22]. Despite robust relations among CM, DER, and NSSI, additional research is warranted to delineate the specificity of how DER might mediate relations from CM to NSSI.
The current study examined relations between CM and NSSI within a diverse sample of boys and girls in a psychiatric hospital setting. Considering the diagnosis of patients was mainly depressive disorders in our sample, additional research is warranted to delineate the specificity of how depressive symptoms might mediate relations from CM to NSSI. First, we tested the mediating role of DER and depressive symptoms in these relations. Second, we further tested the mediating chain effect of DER and depression from CM to NSSI. The previous research emphasizes that DER precedes the onset of depressive symptoms and can predict depressive symptom trajectory from early to middle adolescence [
23]. Thus, the DER might mediate the relationship between childhood emotional abuse and current depression [
24]. See Additional file
1: Figure S1 for details of the theoretical model. Simultaneously, each direct or indirect path was estimated to determine its effect size in explaining the pathway from CM to NSSI.
The present study investigated adolescent NSSI within a significant clinical sample through robust structural equation modeling [
25]. In particular, to focus on NSSI with clinical significance and accurately describe its severity, the severity of NSSI behaviors by clinicians use the Clinician-Rated Severity of NonSuicidal Self-Injury (CRSNSSI-DSM-5) through structured interviews. To our knowledge, this is the first study identifying the mediating pathways of depressive symptoms and DER from childhood maltreatment to NSSI through a chain mediation model in adolescents, further elucidating the roles of DER for NSSI. According to the mentioned literature, DER is a mediator with a stronger effect. The current study represents an essential empirical contribution to the biosocial theory of NSSI.
Discussion
This study aimed to examine the mediation pathway from CM exposure to adolescent NSSI behaviours based on a sample of adolescent patients in a psychiatric setting. The findings supported our hypothesis: CM exposure has indirect effects on the severity of NSSI via the mediating effect of DER severity and depressive symptoms. To our knowledge, this is the first investigation of the distal effects in the pathway from CM to NSSI severity via the chain mediating effects of DER and depression.
In our study, the prevalence of NSSI was over forty percent; girls had higher rates of NSSI, and depressive disorder was the primary psychiatric diagnosis. This is consistent with characteristics in previous adolescent samples [
4,
35,
36]. Women were more likely to engage in NSSI than men in clinical samples [
4], especially adolescent girls (16–19 years old) [
37]. Further model comparison analysis found significant sex differences in the chain model. The chain mediation model explained girls well but not boys (Additional file
1: Table S2, Fig S2). Gender may play a role in NSSI mechanisms, and a recent review argues that CM seems less deleterious in males than females NSSI [
7]. This difference may be due to lower levels of EA and EN (Additional file
1: Table S1), lower prevalence of NSSI (27.1%) to girls (51.1%), or smaller sample size (n = 48) for boys, which is lower than the recommended minimum sample size of 99. However, the main reason may be that only EA, except other CM types, was related to NSSI in our sample.
From our results, there were some unique features in the prevalence of different types of CM.The prevalence of SA in our sample was 12.5%, which was consistent with the result of a Chinese clinical sample study (12.5%) [
38] and slightly higher than a meta-analysis for Chinese individuals (8.9% for women; 9.1% for men) [
39]. In a review of SA studies worldwide, the prevalence in Asian countries was significantly lower than in other regions [
40], and many reviewed studies were from China. Therefore, it might be evident that China did exhibit lower SA rates. It might explain why SA was not associated with NSSI severity in our study, unlike some studies in western countries [
16]. The research regarding the relationship between SA and NSSI is somewhat inconclusive. Some studies have reported that sexual abuse was not directly associated with NSSI [
9,
41]. Weierich and Nock, in contrast, found a relationship only between sexual abuse and NSSI and not between non-sexual abuse and NSSI [
42]. The EA prevalence for girls in our study was significantly higher than that in North America (50.2% vs. 28.4%) and was twice that in an Asian sample (26.9%) [
40]. However, the prevalence of EA for boys was comparable to the median prevalence rates in an Asian sample (29.2% vs. 33.2%) [
40]. The prevalence of PA in our study was similar to recent international prevalence rates but lower than that in China a few years ago [
43]. This is different from the prevalence in Asia, which is the same for boys and girls [
40]. In conclusion, the girls in our study had higher rates of EA, lower rates of SA, and a similar prevalence of PA than those in the international sample and higher rates of EA and EN than the boys in our sample. These results may be related to China’s “one-child policy” and traditional Chinese parenting methods [
44]. Chinese parents focus on their children's academic performance [
45], not their emotional needs [
46]. Compared with safety or sexual threats, children in Chinese families experience more emotional abuse and neglect, which may increase depression and emotional dysregulation [
47].
At the same time, our study is different from that of Shenk, which may be due to different sample characteristics. Shenk’s study sample, recruited from a Child Protective Service (CPS) agency, had high levels of SA (58%), PA (34%) and posttraumatic stress symptoms [
15], while our sample had lower rates of SA (12.5%), lower rates of Trauma and Stressor-Related Disorders (18.8%), and higher rates of EA (46.1%). Two recent studies from clinical and nonclinical samples were in good agreement with our results. A study in clinical samples found that DER mediated the relationship between EA and NSSI frequency and that EA was associated with NSSI but not SA. Another study in a nonclinical sample also found a partial mediating effect of EA, with a higher EA rate (72%) in participants with NSSI.
According to previous research results, CM is a distal associative factor of NSSI and a shared risk factor for depression and DER. For the properties of CM as a distal risk factor, the time bound for maltreatment was before the age of 12 years. Therefore, we speculate that CM mainly affects NSSI through these recent risk factors. The chain mediation model supports our hypothesis, and we found a mediation of the association between CM and NSSI with a significant indirect effect but no remaining direct effect. Whether CM is understood as a direct risk factor for NSSI is still debated [
9,
48]. However, CM as a distal risk factor creates more indirect vulnerability for mental disorders, which increases the likelihood of engaging in NSSI [
49].
DER was the strongest mediator, followed by depressive symptoms, in our model. Regardless, our findings are consistent with a previous study, which showed that the mediating effect of DER was higher than that of depression symptoms in the relationship between CM and NSSI [
15,
50]. Early exposure to abusive or neglectful environments may disrupt children’s development of healthy emotion regulation skills and socioemotional competencies [
51,
52]. Individuals with CM experience tend to use more maladaptive strategies, including inhibition and rumination [
53], and have an impaired ability to use adaptive emotion regulation strategies, such as acceptance, reassessment, and problem-solving, which are associated with enhanced positive emotions and better mental health outcomes [
53]. Without adaptive emotion regulation skills, self-harm behaviours may function as compensatory strategies to cope with overwhelming emotions. Specifically, self-harm may be used to distract oneself from distress and to regain a sense of control and self-efficacy [
54]. We also proved a potential chain-mediated pathway of DER and depression in the relationship between CM and NSSI, which may add to our understanding of the relationship between CM and NSSI. The chain mediating effects of DER and depression was in line with the role of CM, as it impaired the development of self-regulation on emotional cognitive levels, resulting in poor emotional regulation and a depressogenic attributional style. Significantly, these cognitive and emotional sequelae of CM then increase the risk for the later development of depression symptoms [
55,
56] and ultimately contribute to an increased possibility of NSSI [
57]. Our findings potentially align with the biological conceptual model of NSSI [
49]. According to this theory, early exposure to CM impairs the emotion regulation circuit of the brain through environmental-biological interactions in long-term adolescent development, causing DER and depression. Then, adolescents may adopt NSSI as a coping strategy for regulating aversive emotional experiences. Of course, this requires further imaging and biological research.
Our findings also have some clinical implications. First, our results contribute to the understanding that CM and NSSI are common among Chinese inpatient adolescents with psychiatric disorders, as shown in a recent study [
35]. Many adolescents feel reluctant to talk about their CM or NSSI experiences in face-to-face settings. Hence, a mixed battery of self-reports and assessments by physicians may be more suitable in Chinese culture. Second, distal risk factors (i.e., CM) and proximal risk factors (especially DER) might work together to induce the onset of NSSI. Therefore, assessing CM, DER and depressive symptoms in adolescents with NSSI behaviours is also of practical importance, and we should enhance healthy emotion regulation strategies. Various components of DER, especially limited access to strategies [
12], will be targeted for future NSSI interventions. Current interventions facilitating healthy emotion regulation might help modify maladaptive cognitions and behaviours [e.g., Dialectical Behaviour Therapy, DBT; Emotion Regulation Individual Therapy for Adolescents (ERITA)] [
22,
58,
59]. However, more study is needed, as research on NSSI prevention is preliminary.
Conclusions and limitations
Overall, this study explored the mechanism between the distal risk factors of CM and NSSI behaviour using a sample of inpatient adolescents in China, demonstrating a simple and chained mediating effect of DER and/or depressive symptoms. Importantly, our research shows that DER seems to be a mediator with a stronger indirect effect compared to depressive symptoms. Furthermore, we chose the CRSNSSI (DSM-5) as the NSSI evaluation index through structured interviews, which reflects the severity of NSSI better than self-rating NSSI scales. Despite such strengths, several limitations of the current study need to be acknowledged. First, the participants were patients in one hospital in Hangzhou, and male participants were underrepresented. Therefore, the sample source would limit the generalizability of these findings. Second, although we controlled the time boundaries of CM, we used cross-sectional data on DER and depression, and correlational data were used to test a causal model. Future research should utilize a longitudinal design while considering other possible mediators (e.g., life stressors). Third, we employed retrospective self-report questionnaires, and adolescents would unlikely be able to retrospectively report on maltreatment occurring in infancy, toddlerhood, or early childhood. Also, adolescents who are depressed, having difficulty regulating their emotions, and/or engaging in NSSI may be more likely to view their childhood experiences and family relationships more negatively. These might mean our data were not sufficiently objective. Therefore, these data may be affected by bias to some extent. Future studies should employ a large sample, multicentre, longitudinal design and adopt tools other than self-report questionnaires (e.g., expert opinions or other objective evidence). Future research should also examine subdomains of DER and their roles in NSSI behaviours. Because the recruitment period overlaps with the outbreak of COVID-19 in Hangzhou. Therefore, some impacts of COVID-19 need to be considered in our interpretation of results.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.