Latent transition analysis of non-suicidal self-injury comorbidity patterns among adolescents: predictive roles of impulsivity, emotional dysregulation, and core self-evaluations

Abstract

Despite evidence that non-suicidal self-injury (NSSI) co-occurs with both emotional and behavioral problems in adolescence, little research has addressed their tripartite comorbidity. This study employed a person-centered approach to identify heterogeneous comorbidity patterns of NSSI with multiple emotional and behavioral problems, assess their stability and transitions, and explore associated risk and protective factors. A total of 3704 adolescents (57.1% males; Mage = 14.0, SD = 0.83) participated in a two-wave, nine months study. Three distinct patterns were identified, including one Low-symptoms and two comorbidity patterns, namely the Emotion-dominant, Low-NSSI Comorbidity and NSSI-dominant, High-symptoms Comorbidity. Across time, adolescents showed three directions of change as stability, symptoms improvement, and symptoms worsening. Moreover, adolescents with higher impulsivity and emotional dysregulation were more likely to be classified into the comorbidity patterns, whereas those with higher core self-evaluation were less likely to belong to the comorbidity patterns. Longitudinally, higher core self-evaluation reduced the likelihood of symptoms worsening. These findings reveal the heterogeneous and fluid nature of the tripartite comorbidity among NSSI, emotional problems, and behavioral problems, and point to the value of integrative interventions that target emotion regulation, impulse control, and core self-evaluation.

Introduction

Non-suicidal self-injury (NSSI) refers to the deliberate destruction of one’s own body tissue without suicidal intent (Nock, 2010). Adolescence is a high-risk period for the onset of NSSI, with an average prevalence of 23.2% (Xiao et al., 2022). Notably, NSSI often co-occurs with other emotional and behavioral problems during this period (Koposov et al., 2021; Nitkowski & Petermann, 2011; Sadath et al., 2023), and such comorbidities are associated with elevated risk for adverse developmental outcomes (Rong et al., 2025; Tanner et al., 2015). Despite its research and clinical significance, most existing studies have focused on the co-occurrence of NSSI with either emotional (e.g., anxiety, depression) or behavioral (e.g., aggression, delinquency) problems separately (Jiao et al., 2022; O’Donnell et al., 2015; Tanner et al., 2015; Zhao et al., 2024), with limited attention to more severe and complex patterns involving the comorbidity of NSSI, emotional problems and behavioral problems. In addition, prior research has predominantly adopted variable-centered approaches and cross-sectional designs, overlooking possible heterogeneous comorbidity patterns and their developmental changes over time. To address these gaps, this study adopts a person-centered approach and longitudinal design to identify heterogeneous patterns of NSSI comorbidity with multiple emotional and behavioral problems, as well as their transitions. Furthermore, guided by existing theories, this study also examines emotion dysregulation and impulsivity as potential risk factors, and core self-evaluation as a protective factor, for NSSI comorbidity patterns and their transitions.

Comorbidity of non-suicidal self-injury, emotional problems, and behavioral problems

Despite NSSI has become an individual diagnosis as “condition for further study” in DSM-5 (American Psychiatric Association, 2013), and it frequently operationalized as a standalone construct in empirical research, accumulating evidence suggests that NSSI seldom manifests independently. Instead, it commonly co-occurs with multiple emotional and behavioral problems, especially in adolescents (Nitkowski & Petermann, 2011). Currently, some empirical studies have examined the co-occurrence of NSSI with either emotional or behavioral problems separately. Regarding the emotional problems, most research has focused on the co-occurrence of NSSI and depression (He et al., 2023; Liang et al., 2023; Niu et al., 2024; Tilton-Weaver et al., 2019), while a few studies have identified a triadic co-occurrence of NSSI, depression and anxiety (Jiao et al., 2022; Zhao et al., 2024). For instance, a recent study (Rong et al., 2025) applied person-centered analyses and identified four heterogeneous subgroups among Chinese adolescents based on depression, anxiety, NSSI, suicide ideation and suicide attempt: a low-symptom class (70.8% of adolescents), a self-harm class (9.1%), an emotional symptom class (13.4%), and a high-symptom class (6.7%). Regarding behavioral problems, several studies have documented the co-occurrence of NSSI with aggression and certain form of delinquency. For example, a systematic review by O’Donnell et al. (2015) showed that self-injury, including NSSI, frequently co-occurs with aggression, with prevalence rates ranging from 5% to 74%. Tanner et al. (2015) reported a co-occurrence rate of 51.1% between NSSI and delinquency (i.e., fire setting) among Australia adolescents. However, more severe and complex comorbidity patterns, in which NSSI co-occurs simultaneously with both emotional and behavioral problems, have largely been overlooked in previous research.

From a theoretical perspective, the co-occurrence of NSSI with both emotional and behavioral problems can be understood from two complementary angles. One is the shared etiological perspective. For example, the Hierarchical Taxonomy of Psychopathology (HiTOP) system proposed by Kotov et al. (2017) posits that emotional and behavioral problems are part of a broad psychopathological spectrum network, sharing common genetic, environmental, and neurobiological risk factors. Within this framework, NSSI has also been shown to possess strong transdiagnostic features that link it to broader emotional and behavioral problems (Wang & Eaton, 2023). Another perspective is the emotion-behavior interplay perspective. According to Bresin’s (2020) unifying theory of dysregulated behaviors, the “dysregulated behaviors” such as NSSI, aggression and some delinquent behaviors (e.g., tobacco and alcohol use) are often used as a subset of behavioral avoidance strategies to cope with distressing thoughts and experiences arising from emotional problems. These theoretical frameworks jointly indicate that NSSI, emotional problems, and behavioral problems can co-occur within the same individual, and such co-occurrence may further reflect a higher level of coping vulnerability. In terms of empirical evidence, emerging studies have demonstrated that some adolescents simultaneously exhibit emotional and behavioral problems alongside NSSI (Koposov et al., 2021; Sadath et al., 2023). Additionally, recent neurophysiological research has identified shared neuro-affective mechanisms underlying NSSI and emotional and behavioral problems, such as dampened striatal responding while anticipating incentives (Schettini et al., 2021). Nevertheless, empirical research on the comorbidity of NSSI with both emotional and behavioral problems remain limited, highlighting the need for further investigation into its characteristics and potential mechanisms. Moreover, most existing studies have primarily employed variable-centered approaches, which are based on assumption of homogeneity among sample individuals (Koposov et al., 2021; Laursen & Hoff, 2006). To obtain a more ecologically valid and nuanced understanding of the co-occurrence of NSSI with emotional and behavioral problems, it is also important to consider the heterogeneity within NSSI comorbid groups. Person-centered approaches like latent profiles analysis (LPA) allow for the identification of underlying subgroups based on individuals’ presentation of NSSI, emotional problems and behavioral problems (Laursen & Hoff, 2006) and are considered more informative and statistically robust in capturing complex health comorbidity patterns (Sinha et al., 2021). Therefore, the current study would examine the heterogeneous comorbidity patterns of NSSI with both emotional and behavioral problems, utilizing the LPA approach.

In addition, established research has demonstrated that NSSI, along with emotional and behavioral problems, fluctuates across adolescence due to rapid biological, psychological, cognitive, and social development (Hastings et al., 2011; Sourander & Helstelä, 2005; Zahn-Waxler et al., 2000). Thus, adopting a developmental perspective is essential for understanding the stability and change (or transitions) among different NSSI comorbidity subgroups during this period, which can inform early interventions for at-risk adolescent groups. Latent transition analysis (LTA) provides an effective statistical technique for capturing developmental changes in subgroup membership (Abarda et al., 2020). Therefore, based on the identification of NSSI comorbidity patterns, the current study would further investigate the stability and change between and among these patterns.

Influencing factors associated with the comorbidity of non-suicidal self-injury, emotional problems, and behavioral problems

Given the heighted risk for adverse developmental outcomes among adolescents exhibiting comorbidity of NSSI with emotional and behavioral problems (Rong et al., 2025; Tanner et al., 2015), it is essential to explore key influencing factors underlying this issue to inform targeted prevention and intervention strategies. In the present study, we adopted a framework of risk-protective factors (Crews et al., 2007; Rutter, 1987) rooted in developmental psychopathology to examine the key influences on NSSI comorbidity patterns and their longitudinal transitions.

Regarding risk factors, emotion dysregulation and high impulsivity may represent important transdiagnostic processes associated with NSSI and its co-occurrence with emotional and behavioral problems. According to the cognitive-neural mechanism model of NSSI (Deng et al., 2022), emotion dysregulation and high impulsivity represent central risk factors underlying the occurrence of NSSI with emotional and behavioral problems, rooting in the overlapping brain region abnormalities. Established empirical research has also demonstrated that emotion dysregulation and impulsivity are prominent transdiagnostic factors involved in the onset and maintenance of NSSI, emotional problems and behavioral problems (Hasking & Claes, 2020; Jiang et al., 2024; Thompson et al., 2025). However, a limited number of studies directly investigated how emotion dysregulation and impulsivity relate to the comorbidity of NSSI with emotional problems or behavioral problems. One study by Chen et al. (2023) showed a significant association between emotion dysregulation and NSSI in adolescents with depression. Another research by Peters et al. (2019) found that mood instability may contribute to the emergence of NSSI among individuals with anxiety disorders. In terms of impulsivity, one study provided indirectly evidence about its association with the occurrence of NSSI and behavioral problems, showing that low childhood self-control differentiated adolescents who engaged in both self-harm and violent crime from those who engaged in self-harm alone (Richmond-Rakerd et al., 2019). However, no study has systematically examined how emotion dysregulation and impulsivity relate to the comorbidity of NSSI with both emotional and behavioral problems, nor clarified their distinct roles in different comorbidity patterns and transitions over time.

Concerning protective factors, core self-evaluation may serve as a crucial protective resource. Core self-evaluation refers to the fundamental evaluation individuals make about their own worth and competence (Judge et al., 1998). Li and Nie (2010) suggest that core self-evaluation influences mental health through capitalizing and coping mechanisms. Specifically, in positive situations, individuals with high core self-evaluations tend to respond positively to external support (Li & Nie, 2010), which helps them better recover from emotional and behavioral problems. In negative situations, they have a stronger sense of control about outcomes and are less likely to use avoidant coping strategies (Li & Nie, 2010), which reduces the risk of reciprocal transformation between emotional and behavioral problems. Therefore, core self-evaluation potentially serves as a significant protective factor for NSSI and its comorbidities. Empirically, only a limited number of studies have evidenced the associations between core self-evaluation and NSSI, emotional problems (e.g., depression & anxiety), and behavioral problems (e.g., aggression) separately (Cross et al., 2023; Huang et al., 2025; Wang & Zhang, 2020; Wang et al., 2024). Its protective role in the context of comorbidity has not been directly explored. Therefore, building upon the identification of NSSI comorbidity patterns and their stability and transitions over time, the present study further adopted a developmental psychopathology-informed risk–protective framework, in which emotion dysregulation and high impulsivity are examined as particularly salient risk factors, and core self-evaluation as a potential important protective factor, to identify the key factors influencing these comorbidity patterns and their longitudinal transitions.

In addition, sex differences may also play an important role in adolescents’ NSSI comorbidity with emotional and behavioral problems. Previous studies have documented sex differences in the prevalence and severity of NSSI, emotional problems, and behavioral problems among adolescents (Brown & Plener, 2017; Rodríguez-Mora et al., 2025). More directly, person-centered studies on the co-occurrence of NSSI with either emotional problems (Liang et al., 2023) or behavioral problems (Xiong et al., 2022) have also suggested that the severity or developmental characteristics of comorbid symptoms may differ by sex. However, whether sex differences exist in the comorbidity patterns and transitions involving NSSI, emotional problems, and behavioral problems simultaneously remains unclear. Therefore, the present study further examined sex differences in adolescents’ profile membership and transitions. Beyond sex, other demographic variables such as age and socioeconomic status may also be related to this issue. Although direct evidence regarding their roles in NSSI comorbidity remains limited, previous studies have indicated that age and socioeconomic status are associated with a broad range of psychopathological symptoms among adolescents (Kieling et al., 2024; Peverill et al., 2021; Polanczyk et al., 2015). Therefore, while examining the roles of emotion dysregulation, impulsivity, core self-evaluation, and sex, the present study also included age and socioeconomic status as covariates to control for their potential confounding effects.

The current study

This study addressed three primary objectives. First, given that existing research has examined the co-occurrence of NSSI with either emotional or behavioral problems separately, ignoring their tripartite comorbidity, as well as their heterogeneous comorbidity patterns, this study aimed to examine the heterogeneous comorbidity patterns of NSSI with both emotional and behavioral problems. Based on existing literature, it is hypothesized that several patterns would emerge, including a subgroup of adolescents with low levels of all the symptoms, a subgroup exhibiting the comorbidity between NSSI and emotional symptoms, a subgroup exhibiting the comorbidity between NSSI and behavioral symptoms, and a subgroup exhibiting the comorbidity of NSSI with both emotional and behavioral symptoms. Second, given the rapid development during adolescence, the stability and transition of each pattern identified in the first aim over nine months were analyzed. It is hypothesized that the subgroup with low levels of all the symptoms would exhibit the greatest stability across time, while other subgroups would display a minor to moderate degree of transition. Third, it also aimed to investigate the association between emotion dysregulation, impulsivity and core self-evaluation and the comorbidity patterns of NSSI identified in the first aim, as well as to assess how these factors influence the transition of patterns identified in the second aim. It is hypothesized that adolescents with higher levels of emotion dysregulation and impulsivity would be more likely to be classified and translated to the at-risk subgroups (i.e., subgroups with patterns of NSSI comorbidity). While adolescents with higher levels of core self-evaluation would be more likely to be classified into the subgroup with low levels of all the symptoms, and they are also more likely to translated from the at-risk subgroups to the subgroup with low symptoms. In addition, given the potential role of sex differences in NSSI and its comorbidity with emotional and behavioral problems, the current study further explored sex differences in membership in NSSI comorbidity patterns and in transitions between patterns over time.

Methods

Participants and procedures

Data were collected from three schools located in Northern China, including one junior high school, one nine-year school integrating primary and junior high education, and one twelve-year school integrating primary, junior high and senior high education. All participants were recruited from the junior high school grades. Two schools were located in urban areas, and one was located in a county-level town. Two surveys were administered at an interval of 9 months, starting in summer 2022 and continuing until spring 2023. At the first time point, a total of 3881 students in seventh or eighth grade were recruited for the study. Due to attrition at the second time point, 2999 students from the original sample participated in the study, with a retention rate of 77.3%. The known primary reason for dropout reported by participants was preparation for the senior high school entrance examination in ninth grade. Invalid questionnaires at T1 and T2 were then excluded based on instructed response items (Ward & Meade, 2023) and response times (Pérez-Rojas et al., 2021), resulting in a valid baseline sample of 3704 adolescents (57.1% males, Mage = 14.00, SD = 0.83, range = 11 to 17 years old). The educational levels of the parents of these adolescents were as follows, with 60.0% of fathers and 64.4% of mothers had a junior high school education or below, 25.9% of fathers and 22.7% of mothers had completed high school, and 14.2% of fathers and 12.9% of mothers had attained a college degree or above.

Missing data analyses were subsequently conducted based on this valid sample. Little’s Missing Completely at Random (MCAR) test (Little, 1988) indicated that the data were not missing completely at random (χ²/_df=9 = 372.521, p < 0.001), although the possibility of missing at random (MAR) could not be ruled out. Further group difference analyses showed no significant difference in subjective socioeconomic status between continuers and dropouts (t/_df=1394.417 = 0.410, p = 0.682), whereas significant differences were found for sex (χ²/_df=1 = 4.226, p < 0.05) and age (t/_df=1424.578 = 15.973, p < 0.001). The age difference was consistent with the fact that most dropouts were ninth-grade students. Therefore, all participants in the valid baseline sample were retained for subsequent analyses, and multiple imputations were used to handle missing data (Mustillo & Kwon, 2015).

This study received approval from the Institutional Review Board and Ethics Committee of Human Participant Protection, Faculty of Psychology at Beijing Normal University. Prior to the study’s commencement, participants and their parents or legal caregivers provided written informed consent. All students were clearly informed that their participation was voluntary. Their responses would be kept confidential, and they retained the right to withdraw from the study at any time during data collection. Then, students completed questionnaires under the standardized guidance of research assistants, with the completion of these questionnaires taking approximately 35 min. After the study, each participant was rewarded with stationery as an appreciation for their contribution.

Measures

Non-suicidal self-injury (T1 & T2)

Adolescents’ NSSI was measured using the Deliberate Self-harm Inventory: Nine-item Version (DSHI-9; Bjärehed et al., 2012; Gratz, 2001). This instrument comprises 9 items, each representing a distinct non-suicidal self-injury behavior, such as “burning with cigarette, lighter or match”. Adolescents reported the frequency of engaging in these behaviors over the past six months using a 7-point scale (0 = never, 6 = five or more times). The total scores were calculated, with higher scores indicating greater NSSI severity. The DSHI-9 has established good reliability and validity in Chinese adolescents (Xiong et al., 2022). In this study, Cronbach’s α coefficients were 0.90 at T1 and 0.88 at T2.

Depression (T1 & T2)

Adolescents’ depression was measured using the 10-item Centre for Epidemiological Studies Depression Scale (CES-D-10; Andresen et al., 1994; Radloff, 1977). Each item representing a core depressive symptom, such as “my sleep was restless”. Adolescents reported how often they had experienced these symptoms in the past week using a 4-point scale (1 = less than one day, 4 = most of the time [5 to 7 days]). Mean scores were computed, with higher scores indicating more severe depressive symptoms. The CES-D-10 has established good reliability and validity in Chinese adolescents (Guo et al., 2020). In this study, Cronbach’s α coefficients were 0.89 at T1 and 0.90 at T2.

Anxiety (T1 & T2)

Adolescents’ anxiety was measured using the “Level 2-Anxiety-Child Age 11–17″ scale (PROMIS Health Organization & PROMIS Cooperative Group, 2012), which was developed from the DSM-5 (American Psychiatric Association, 2013), and the Chinese version was adapted by Zhang et al. (2018). Adolescents reported their anxiety symptoms (e.g., “I felt like something awful might happen”) in the past week using a 5-point Likert-type scale (1 = never, 5 = almost always). Mean scores were computed, with higher scores indicating greater anxiety symptoms. The Chinese version of this instrument consists of 10 items and demonstrated good reliability and validity in Chinese adolescents (Cai et al., 2024; Zhang et al., 2018). In this study, Cronbach’s α coefficients were 0.92 at T1 and 0.93 at T2.

Aggression (T1 & T2)

Adolescents’ aggression was measured using the Chinese version of Aggression Questionnaire (AQ-CV; Buss & Perry, 1992; Li et al., 2011). We adopted physical aggression (e.g., “give enough provocation, I may hit another person”) and verbal aggression (e.g., “I often find myself disagreeing with people”) subscales to address aggression, with a total of 12 items. Each item was answered using a 5-point Likert-type scale (1 = extremely uncharacteristic, 5 = extremely characteristic). Mean scores were computed, with higher scores indicating higher levels of aggression. In this study, Cronbach’s α coefficients were 0.91 at both T1 and T2.

Delinquency (T1 & T2)

Adolescent’s delinquency was measured using the Problem Behaviors Checklist, which was developed by Fang et al. (2004). This instrument comprises 12-items and has been demonstrated to exhibit adequate reliability and validity among Chinese adolescents (Gao et al., 2025; Su et al., 2015). Adolescents reported the frequency of their delinquent behaviors (e.g., “stealing things,” “damaging the property of the public or others”) over the past semester using a 4-point Likert scale (1 = never, 4 = often). Mean scores were computed, with higher scores indicating higher levels of delinquency. In this study, Cronbach’s α coefficients were 0.84 at T1 and 0.85 at T2.

Impulsivity (T1 & T2)

Adolescents’ impulsivity was measured using the Chinese version of Barratt Impulsivity Scale (BIS-11, Li et al., 2011; Patton et al., 1995). We adopted motor impulsiveness (e.g., “I act ‘on impulse’”) and attentional impulsiveness (e.g., “When I am thinking about something, I can focus” – reverse scored) subscales to address impulsivity, with a total of 20 items. Each item was answered using a 5-point Likert scale (1 = never, 5 = always). Mean scores were computed, with higher scores indicating higher levels of impulsivity. In this study, Cronbach’s α coefficients were 0.90 at T1 and 0.91 at T2.

Emotion dysregulation (T1 & T2)

Adolescents’ emotion dysregulation was measured using the short form of the Difficulties in Emotion Regulation Scale (DERS-SF; Kaufman et al., 2016). This instrument comprises six subscales representing facets of emotion dysregulation: nonacceptance of emotional responses, difficulty engaging in goal-directed behavior, impulse control difficulties, lack of emotional awareness, limited access to emotion regulation strategies and lack of emotional clarity, with a total of 18 items. Each item was answered using a 5-point Likert scale (1 = almost never, 5 = almost always). Mean scores were computed, with higher scores indicating higher levels of emotion dysregulation. The DERS-SF has established good reliability and validity in Chinese adolescents (Jiang et al., 2022). In this study, Cronbach’s α coefficients were 0.89 at T1 and 0.90 at T2.

Core self-evaluation (T1 & T2)

Adolescents’ core self-evaluation was measured using the Chinese version of Core Self-Evaluation Scale (Judge et al., 2003; Ren & Ye, 2009). This instrument comprises 8 items (e.g., “Overall, I am satisfied with myself”), and each item was answered using a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). Mean scores were computed, with higher scores indicating higher levels of core self-evaluation. The Chinese version of the scale has been successfully used in previous studies involving Chinese adolescents and has demonstrated acceptable construct validity and internal consistency (Liu et al., 2024; Ren & Ye, 2009; Tian et al., 2021). In this study, Cronbach’s α coefficients were 0.88 at T1 and 0.90 at T2.

Statistical analysis strategy

Firstly, descriptive statistics were conducted for main study variables, including means, standard deviations, and correlation analysis. Secondly, the latent profiles analysis (LPA) was applied at each time point to identify subgroups of adolescents based on similar profile of NSSI comorbidity patterns, which use NSSI, emotional symptoms (depression & anxiety), and behavioral symptoms (aggression & delinquency) as observed indicators. All study variables were converted into Z score (M = 0, SD = 1) before conducting LPA. Thirdly, latent transition analysis (LTA), which is a longitudinal extension of LPA, was conducted to explore the stability and change across identified profiles from T1 to T2. Finally, multinomial logistic regression analysis was utilized to examine the effect of impulsivity, emotional dysregulation and core self-evaluation on the profile membership at each time point and the profile transition from T1 to T2. Sex was also included as a predictor to examine potential sex differences in profile membership and transitions, whereas age and subjective socioeconomic status (SES) were included as covariates to control for their potential confounding effects. To control for Type I error due to multiple comparisons, the Benjamini–Hochberg false discovery rate (FDR) correction was applied to the multinomial logistic regression analyses (Benjamini & Hochberg, 1995), with statistical significance determined at an FDR-corrected p < 0.05. Model parameters were estimated using the maximum likelihood estimator with robust standard errors (MLR). Data analyses were conducted by SPSS 23.0 and Mplus 8.3.

The Harman single-factor test was conducted to examine potential common method bias using unrotated principal component factor analyses at the two time points. At T1, fifteen common factors with eigenvalues greater than 1 were extracted, and the first factor accounted for 28.16% of the total variance. Similarly, at T2, fifteen factors were extracted, with the first factor accounting for 29.49% of the variance. As the variance explained by the first factor at both time points was below the recommended threshold of 40% (Tang & Wen, 2020), common method bias was unlikely to be a serious concern in this study.

Results

Descriptive statistics

Statistical summaries, including means, standard deviations, and bivariate correlations of the study variables at two time points, were provided in Table 1.

Table 1.

Correlations, means, and standard deviations of study variables.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19
1 T1 NSSI	1
2 T1 DEP	0.52⁎⁎⁎	1
3 T1 ANX	0.43⁎⁎⁎	0.74⁎⁎⁎	1
4 T1 AGG	0.39⁎⁎⁎	0.45⁎⁎⁎	0.5⁎⁎⁎	1
5 T1 DEL	0.50⁎⁎⁎	0.48⁎⁎⁎	0.49⁎⁎⁎	0.58⁎⁎⁎	1
6 T2 NSSI	0.68⁎⁎⁎	0.42⁎⁎⁎	0.36⁎⁎⁎	0.33⁎⁎⁎	0.43⁎⁎⁎	1
7 T2 DEP	0.45⁎⁎⁎	0.70⁎⁎⁎	0.62⁎⁎⁎	0.39⁎⁎⁎	0.43⁎⁎⁎	0.50⁎⁎⁎	1
8 T2 ANX	0.34⁎⁎⁎	0.58⁎⁎⁎	0.67⁎⁎⁎	0.39⁎⁎⁎	0.39⁎⁎⁎	0.36⁎⁎⁎	0.75⁎⁎⁎	1
9 T2 AGG	0.32⁎⁎⁎	0.37⁎⁎⁎	0.38⁎⁎⁎	0.66⁎⁎⁎	0.48⁎⁎⁎	0.33⁎⁎⁎	0.43⁎⁎⁎	0.47⁎⁎⁎	1
10 T2 DEL	0.41⁎⁎⁎	0.40⁎⁎⁎	0.39⁎⁎⁎	0.44⁎⁎⁎	0.56⁎⁎⁎	0.44⁎⁎⁎	0.45⁎⁎⁎	0.45⁎⁎⁎	0.6⁎⁎⁎	1
11 T1 IMP	0.35⁎⁎⁎	0.61⁎⁎⁎	0.56⁎⁎⁎	0.49⁎⁎⁎	0.47⁎⁎⁎	0.29⁎⁎⁎	0.50⁎⁎⁎	0.48⁎⁎⁎	0.41⁎⁎⁎	0.41⁎⁎⁎	1
12 T1 ED	0.41⁎⁎⁎	0.70⁎⁎⁎	0.67⁎⁎⁎	0.48⁎⁎⁎	0.43⁎⁎⁎	0.32⁎⁎⁎	0.58⁎⁎⁎	0.53⁎⁎⁎	0.37⁎⁎⁎	0.37⁎⁎⁎	0.69⁎⁎⁎	1
13 T1 CSE	−0.42⁎⁎⁎	−0.75⁎⁎⁎	−0.70⁎⁎⁎	−0.39⁎⁎⁎	−0.43⁎⁎⁎	−0.35⁎⁎⁎	−0.61⁎⁎⁎	−0.56⁎⁎⁎	−0.33⁎⁎⁎	−0.37⁎⁎⁎	−0.67⁎⁎⁎	−0.72⁎⁎⁎	1
14 T2 IMP	0.33⁎⁎⁎	0.52⁎⁎⁎	0.51⁎⁎⁎	0.43⁎⁎⁎	0.42⁎⁎⁎	0.32⁎⁎⁎	0.63⁎⁎⁎	0.62⁎⁎⁎	0.48⁎⁎⁎	0.47⁎⁎⁎	0.73⁎⁎⁎	0.58⁎⁎⁎	−0.57⁎⁎⁎	1
15 T2 ED	0.35⁎⁎⁎	0.57⁎⁎⁎	0.56⁎⁎⁎	0.44⁎⁎⁎	0.37⁎⁎⁎	0.35⁎⁎⁎	0.70⁎⁎⁎	0.68⁎⁎⁎	0.46⁎⁎⁎	0.44⁎⁎⁎	0.54⁎⁎⁎	0.67⁎⁎⁎	−0.57⁎⁎⁎	0.72⁎⁎⁎	1
16 T2 CSE	−0.36⁎⁎⁎	−0.61⁎⁎⁎	−0.60⁎⁎⁎	−0.37⁎⁎⁎	−0.38⁎⁎⁎	−0.36⁎⁎⁎	−0.73⁎⁎⁎	−0.69⁎⁎⁎	−0.40⁎⁎⁎	−0.43⁎⁎⁎	−0.55⁎⁎⁎	−0.59⁎⁎⁎	0.70⁎⁎⁎	−0.68⁎⁎⁎	−0.72⁎⁎⁎	1
17 Age	0.04*	0.08⁎⁎⁎	0.06⁎⁎⁎	0.08⁎⁎⁎	0.06⁎⁎	0.06⁎⁎⁎	0.08⁎⁎⁎	0.09⁎⁎⁎	0.10⁎⁎⁎	0.07⁎⁎⁎	0.07⁎⁎⁎	0.06⁎⁎⁎	−0.06⁎⁎⁎	0.07⁎⁎⁎	0.09⁎⁎⁎	−0.06⁎⁎	1
18 SES	−0.10⁎⁎⁎	−0.18⁎⁎⁎	−0.16⁎⁎⁎	−0.05⁎⁎	−0.09⁎⁎⁎	−0.07⁎⁎⁎	−0.13⁎⁎⁎	−0.11⁎⁎⁎	−0.05⁎⁎	−0.08⁎⁎⁎	−0.18⁎⁎⁎	−0.18⁎⁎⁎	0.21⁎⁎⁎	−0.11⁎⁎⁎	−0.12⁎⁎⁎	0.16⁎⁎⁎	−0.03	1
19 Sex	−0.09⁎⁎⁎	−0.09⁎⁎⁎	−0.11⁎⁎⁎	0.11⁎⁎⁎	0.06⁎⁎⁎	−0.07⁎⁎⁎	−0.07⁎⁎⁎	−0.09⁎⁎⁎	0.16⁎⁎⁎	0.11⁎⁎⁎	−0.08⁎⁎⁎	−0.04*	0.16⁎⁎⁎	−0.07⁎⁎⁎	0.02	0.16⁎⁎⁎	0.08⁎⁎⁎	−0.03	1
M	2.27	17.06	21.61	20.62	14.84	1.94	17.48	21.99	19.97	15.01	47.21	15.14	27.16	48.15	15.14	27.41	14	5.35	—
SD	6.43	5.92	8.58	8.7	3.87	5.87	6.03	8.36	8.76	4.17	12.37	4.15	6.71	12.34	4.16	6.93	0.83	1.39	—

^⁎p<0.05.

^⁎⁎p<0.01.

^⁎⁎⁎p<0.001.

NSSI non-suicidal self-injury, DEP depression, ANX anxiety, AGG aggression, DEL delinquency, IMP impulsivity, ED emotional dysregulation, CSE core self-evaluation, SES subjective socioeconomic status, code for sex: boys = 1, girls = 0.

Latent profile analysis

Separate cross-sectional LPAs were conducted for T1 and T2. The analysis began with a two-profile solution, followed by evaluation of models with additional latent profiles. Fit indices for the two- to five-profile solutions at two time points are presented in Table 2. At T1, the AIC, BIC, and A-BIC values decreased as additional solutions were added, and the BLRT results for all four models remained statistically significant. The LMR-LRT was not significant for the three-profile solution, but the three-profile solution added meaningful profile (NSSI-dominant, High-symptoms Comorbidity) than two-profile solution. For the four-profile and five-profile solutions, the smallest latent profile included <3% of the total sample. The entropy value (0.897>0.80) for the three-profile solution also indicated that this solution offered a clearer and more exact separation of profiles. At T2, a similar result of AIC, BIC, A-BIC, BLRT, entropy and smallest profile size was obtained. Moreover, The LMR-LRT was statistically significant for the three-profile solution at T2. These suggested that the three-profile solution was the most plausible at each time point. Three profile groups were depicted for each time point in Fig. 1.

Table 2.

Model fit information for LPA solutions ranging from 2 to 5 profiles at T1 and T2.

model		AIC	BIC	A-BIC	Entropy	Adj-LMR-LRT	BLRT	SPS (%)
T1	2-profile	46,906.18	47,005.66	46,954.82	0.986	0.0030	<0.001	6.0%
	3-profile	43,542.35	43,679.13	43,609.22	0.897	0.0790	<0.001	3.6%
	4-profile	41,440.28	41,614.36	41,525.39	0.902	0.0040	<0.001	1.7%
	5-profile	40,108.05	40,319.44	40,211.40	0.906	0.3180	<0.001	1.2%
T2	2-profile	47,172.65	47,272.12	47,221.28	0.987	0.0118	<0.001	6.2%
	3-profile	43,788.64	43,925.42	43,855.51	0.879	0.0000	<0.001	4.0%
	4-profile	41,765.19	41,939.27	41,850.30	0.888	0.1465	<0.001	2.3%
	5-profile	40,076.10	40,287.48	40,179.45	0.904	0.2508	<0.001	0.9%

Bold rows indicate preferred models with the best fitness.

AIC Akaike information criterion, BIC Bayesian information criterion, A-BIC Sample-size adjusted BIC, Adj-LMR-LRT Adjusted Lo-Mendell-Rubin LRT, BLRT Bootstrap likelihood ratio test, SPS smallest profile size.

Fig. 1.

Latent profiles of NSSI, emotional and behavioral symptoms at two assessment times.

As shown in Fig. 1., the three-profile models at T1 and T2 followed similar patterns. Specifically, the first profile (T1: 71.9%, n = 2664; T2: 65.9%, n = 2440) reflected adolescents who exhibited lower levels of NSSI, depression, anxiety, aggression and delinquency than the overall sample, this profile thus was labeled as “Low-symptoms (LS)”. The second profile (T1: 24.4%, n = 905; T2: 30.1%, n = 1116) reflected adolescents who exhibited slightly above average levels of NSSI, along with depression, anxiety, aggression and delinquency that exceeded the mean by >0.5 standard deviations, with depression and anxiety being particularly prominent. Thus, this profile was labeled as “Emotion-dominant, Low-NSSI Comorbidity (ED-LNC)”. The third profile (T1: 3.6%, n = 135; T2: 4.0%, n = 148) reflected adolescents who exhibited highest levels of NSSI, along with depression, anxiety, aggression and delinquency that exceeded the mean by >1 standard deviation, with NSSI being particularly prominent. Accordingly, this profile was labeled as “NSSI-dominant, High-symptoms Comorbidity (ND-HSC)”. Sex distributions across the three profiles at T1 and T2 are presented in Table 3.

Table 3.

Sex distributions across the latent profiles at two time points.

Time points	profile	boys	girls
Time 1	Low-symptoms	58.5%	41.5%
	Emotion-dominant, Low-NSSI Comorbidity	55.7%	44.3%
	NSSI-dominant, High-symptoms Comorbidity	44.7%	55.3%
Time 2	Low-symptoms	57.9%	42.1%
	Emotion-dominant, Low-NSSI Comorbidity	56.9%	43.1%
	NSSI-dominant, High-symptoms Comorbidity	46.7%	53.3%

Latent transition analysis

Prior to latent transition analysis, measurement invariance was explored. The Satorra–Bentler scaled chi–square test (SB χ² (15) = 50.11, p < 0.001; Satorra & Bentler, 2001) indicated significant differences between the constrained model and unconstrained model. Despite this statistical significance, the constrained model was selected for the following reasons. First, the SB χ² test is sensitive to large sample sizes (n = 3704), where even trivial differences in loglikelihood can yield significant results (Cheung & Rensvold, 2002). Second, the constrained model had a lower BIC (85706.568) than the unconstrained model (85775.275), reflecting a better balance of fit and parsimony (Schwarz, 1978). Additionally, this aligned with measurement invariance requirements for longitudinal comparisons (Yau et al., 2018), enhancing the interpretability of results. Thus, the constrained model was adopted to examine longitudinal profile transitions.

Table 4 reports the transition probabilities derived from the LTA model, which quantifies the probability of transitions among the three groups from T1 to T2. Adolescents classified into the LS group demonstrated the highest rate of stability, with 92.1% maintaining their classification at T2, 7.3% transitioned to the ED-LNC group, and 0.6% transitioned to the ND-HSC group. The ED-LNC group showed an 84.6% stability rate, with 10.5% of transitions directed to the LS group, and 4.9% to the ND-HSC group. The group with the lowest stability was ND-HSC group, where 54.8% of adolescents remained in this group, 42.3% transitioned to the ED-LNC group, and 2.9% transitioned to the LS group.

Table 4.

Latent transition probabilities from T1 to T2.

T1	T2
	LS	ED-LNC	ND-HSC
LS	0.921	0.073	0.006
ED-LNC	0.105	0.846	0.049
ND-HSC	0.029	0.423	0.548

LS Low-symptoms group, ED-LNC Emotion-dominant, Low-NSSI Comorbidity group, ND-HSC NSSI-dominant, High-symptoms Comorbidity group.

In addition, sensitivity analyses by grade level were conducted to examine the robustness of the latent profiles and profile transitions. The results indicated that the three-profile solution remained optimal for both the Grade 7 and Grade 8 subsamples, with profile characteristics corresponding to the Low-symptoms, Emotion-dominant, Low-NSSI Comorbidity, and NSSI-dominant, High-symptoms Comorbidity groups. Moreover, the transition trends and transition probabilities from T1 to T2 were generally consistent across the Grade 7 and Grade 8 subsamples and the total sample. These findings suggest that the present findings regarding latent profiles and profile transitions were relatively robust across grade levels. Detailed results are presented in Appendix A.

Influencing factors of profiles and transitions

Multinomial logistic regression analyses were conducted to examine how impulsivity, emotional dysregulation and core self-evaluation at both T1 and T2 contributed to the profile membership at each respective time point while also examining the predictive effects of T1 levels of these constructs on longitudinal profile transitions from T1 to T2. Table 5 presents the cross-sectional associations between impulsivity, emotional dysregulation, core self-evaluation, and profile membership at both T1 and T2. Specifically, for impulsivity, adolescents with higher levels of impulsivity were more likely to be classified into the ED-LNC group (T1: OR = 1.06; T2: OR = 1.07) and ND-HSC group (T1: OR = 1.06; T2: OR = 1.09) rather than the LS group at both time points. As for emotional dysregulation, adolescents with higher levels of emotional dysregulation were more likely to be classified into the ND-HSC group than all the other groups (T1: OR = = 1.16 ∼ 1.54; T2: OR = 1.14 ∼ 1.54), and these adolescents were more likely to be classified into the ED-LNC group than the LS group (T1: OR = 1.33; T2: OR = 1.36). Furthermore, as for core self-evaluation, adolescents with higher levels of core self-evaluation were less likely to be classified into the ED-LNC group (T1: OR = 0.79; T2: OR = 0.81) and ND-HSC group (T1: OR = 0.71; T2: OR = 0.73) than the LS group at both time points, and these adolescents were less likely to be classified into the ND-HSC group than the ED-LNC group (T1: OR = 0.90; T2: OR = 0.90). Regarding sex, boys were more likely than girls to be classified into the ED-LNC group rather than the LS group, at both T1 (OR = 1.46) and T2 (OR = 1.61).

Table 5.

Multinomial logistic regression of predictors on T1 and T2 profiles.

predictor	ED-LNC vs LS				ND-HSC vs LS				ND-HSC vs ED-LNC
	OR	SE	p-value	FDR Corrected p-value	OR	SE	p-value	FDR Corrected p-value	OR	SE	p-value	FDR Corrected p-value
Time 1
T1 IMP	1.06***	0.01	<0.001	<0.001	1.06***	0.01	<0.001	<0.001	1.01	0.01	0.4156	0.4156
T1 ED	1.33***	0.04	<0.001	<0.001	1.54***	0.07	<0.001	<0.001	1.16***	0.05	0.0002	0.0002
T1 CSE	0.79***	0.01	<0.001	<0.001	0.71***	0.02	<0.001	<0.001	0.90***	0.03	0.0001	0.0002
Time 2
T2 IMP	1.07***	0.01	<0.001	<0.001	1.09***	0.02	<0.001	<0.001	1.02	0.01	0.0514	0.0514
T2 ED	1.36***	0.04	<0.001	<0.001	1.54***	0.07	<0.001	<0.001	1.14***	0.04	0.0002	0.0003
T2 CSE	0.81***	0.01	<0.001	<0.001	0.73***	0.03	<0.001	<0.001	0.90***	0.03	0.0006	0.0007

LS Low-symptoms group, ED-LNC Emotion-dominant, Low-NSSI Comorbidity group, ND-HSC NSSI-dominant, High-symptoms Comorbidity group, IMP impulsivity, ED emotional dysregulation, CSE core self-evaluation.

Raw and FDR-corrected p-values were based on Wald tests of the corresponding multinomial logistic regression coefficients. Bold OR values indicate statistically significant effects after FDR correction. *p<0.05, ^⁎⁎p<0.01, ^⁎⁎⁎p<0.001.

Table 6 summarizes the longitudinal predictive effects of impulsivity, emotional dysregulation, and core self-evaluation at T1 on profile transitions from T1 to T2. Among these predictors, core self-evaluation at T1 uniquely predicted the transitions. Specifically, among adolescents in LS group at T1, higher levels of core self-evaluation were associated with lower odds of moving to ED-LNC group (OR = 0.94) than staying in the same group at T2. And among adolescents in ED-LNC group at T1, higher levels of core self-evaluation were associated with lower odds of moving to ND-HSC group (OR = 0.91) than staying in the same group at T2. In addition, sex also significantly predicted profile transitions. Among adolescents in the ED-LNC group at T1, boys were more likely than girls to transition to LS group at T2 rather than remain in the ED-LNC group (OR = 1.67).

Table 6.

Multinomial logistic regression of predictors on profile transitions.

Predictor		LS					ED-LNC					ND-HSC
		Logit	SE	OR	p-value	FDR Corrected p-value	Logit	SE	OR	p-value	FDR Corrected p-value	Logit	SE	OR	p-value	FDR Corrected p-value
T1 IMP	LS	REF					0.01	0.01	1.01	0.2238	0.3357	0.01	0.03	1.01	0.8025	0.8025
	ED-LNC	−0.01	0.01	0.993	0.5591	0.8387	REF					−0.01	0.02	0.995	0.7273	0.7273
	ND-HSC	−0.03	0.04	0.98	0.5711	0.6854	−0.03	0.02	0.97	0.0484	0.2904	REF
T1 ED	LS	REF					0.05	0.03	1.05	0.0940	0.2820	−0.11	0.11	0.90	0.3198	0.3838
	ED-LNC	−0.06	0.04	0.95	0.1578	0.4734	REF					0.02	0.05	1.02	0.6444	0.7733
	ND-HSC	−0.12	0.16	0.89	0.4486	0.6729	−0.02	0.06	0.98	0.7149	0.7149	REF
T1CSE	LS	REF					−0.06**	0.02	0.94	0.0012	0.0074	−0.10	0.07	0.91	0.1421	0.2843
	ED-LNC	0.02	0.02	1.02	0.4943	0.9885	REF					−0.10*	0.03	0.91	0.0036	0.0216
	ND-HSC	0.09	0.11	1.10	0.3933	0.7865	0.07	0.04	1.07	0.0818	0.2453	REF

REF represents the reference group. LS Low-symptoms group, ED-LNC Emotion-dominant, Low-NSSI Comorbidity group, ND-HSC NSSI-dominant, High-symptoms Comorbidity group, IMP impulsivity, ED emotional dysregulation, CSE core self-evaluation.

Bold OR values indicate statistically significant effects after FDR correction. *p<0.05, ^⁎⁎p<0.01, ^⁎⁎⁎p<0.001.

Discussion

NSSI might co-occur with both emotional and behavioral problems (Browning & Muehlenkamp, 2024). However, this issue has received limited attention in existing literature. The few available studies on NSSI comorbidity mainly relied on cross-sectional data and variable-centered approaches, which are insufficient to capture the heterogeneous and dynamic nature of NSSI comorbidity patterns. Moreover, the key risk and protective factors underlying both NSSI comorbidity and its changes over time remain largely unclear. Therefore, this study employed a person-centered approach to identify distinct patterns of NSSI comorbidity with both emotional and behavioral problems, examine their stability and transitions over time, and investigate how emotion dysregulation, impulsivity and core self-evaluation contribute to the NSSI comorbidity patterns in a longitudinal sample of adolescents.

Comorbidity patterns of non-suicidal self-injury, emotional problems and behavioral problems

This study identified three distinct profiles based on NSSI, emotional problems, and behavioral problems at both time points, providing new evidence for the heterogeneous nature of NSSI comorbidity. Specifically, most adolescents were classified into a low-symptoms group, which showed lowest risk, with no NSSI and no emotional or behavioral problems. This aligns with previous findings on adolescent NSSI (Liang et al., 2023; Xiao et al., 2022). This suggests that although the profound biological, emotional, cognitive, and social changes of adolescence pose challenges to adaptation (Mastorci et al., 2024), low-risk adolescents still account for the largest proportion.

The remaining adolescents were classified into two comorbidity subgroups. The first was the emotion-dominant, low-NSSI comorbidity group, which comprised a larger share of the comorbidity subgroups (T1: 24.4%, T2: 30.1%). For these adolescents, NSSI and behavioral problems may reflect “spillover” responses to emotional problems. According to the neurobiological model of adolescent development (Casey et al., 2008), emotion-related limbic regions mature earlier than prefrontal regions involved in self-regulation, creating a developmental imbalance that heightens emotional vulnerability and limited regulatory capacity during adolescence. Consequently, when emotional problems intensify, some adolescents may engage in NSSI and other behavioral problems as maladaptive regulation strategies to alleviate emotional distress (Bresin, 2020; Selby & Joiner, 2013), leading to the comorbidity pattern of prominent emotional problems accompanied by behavioral problems and NSSI. The second comorbidity group was the NSSI-dominant, high-symptoms comorbidity group. Although this group was the smallest in size (T1: 3.6%, T2: 4.0%), these adolescents showed the highest risk. Notably, they reported six or more NSSI incidents in the past nine months at T1, and eleven or more incidents at T2, reaching the threshold of repetitive NSSI (Brunner et al., 2007; Lüdtke et al., 2018). Previous studies have shown that repetitive NSSI may share characteristics of behavioral addiction (Brunner et al., 2007; Worley, 2020) and linked to greater psychological burden (Brausch & Boone, 2015; Brunner et al., 2014; Manca et al., 2014). In the present study, adolescents in this subgroup also exhibited the highest levels of emotional and behavioral problems. Therefore, adolescents with this comorbidity pattern may be at heightened risk for symptom escalation and more adverse developmental outcomes, underscoring the need for greater clinical attention.

In addition, the present study did not identify the hypothesized profiles characterized by the comorbidity of NSSI with only emotional problems, or only behavioral problems. Instead, two distinct profiles characterized by the tripartite comorbidity of NSSI, emotional problems and behavioral problems were identified. This finding suggests that, among adolescents engaging in NSSI, emotional and behavioral problems may not constitute mutually exclusive comorbidity patterns but rather tend to co-occur to varying degrees across different maladjustment profiles. The specific mechanism underlying the onset of NSSI may help explain this finding. Previous theoretical and empirical studies have suggested that self-injury typically arises from the combined influences of emotional distress and behavioral disinhibition (O’Connor & Kirtley, 2018; Zhou et al., 2026), with emotional distress contributing to the formation of self-injurious thoughts, whereas deficits in behavioral inhibition facilitate the translation of these thoughts into actual self-injurious behavior (O’Connor & Kirtley, 2018). Consequently, individuals engaging in NSSI may simultaneously exhibit vulnerabilities related to emotional distress and behavioral disinhibition. Notably, emotional distress and behavioral disinhibition also represent core mechanisms underlying emotional and behavioral problems, respectively (Sheppes et al., 2015; Young et al., 2009). Therefore, NSSI may be more likely to co-occur with both emotional and behavioral problems. This finding highlights the importance of jointly examining emotional and behavioral problems in adolescent NSSI comorbidity.

Transition of comorbidity patterns of non-suicidal self-injury, emotional problems and behavioral problems

Adopting a developmental perspective, our findings offer important insight into the stability and transitions of subgroup membership overtime. Regarding stability, most adolescents demonstrated moderate to high stability across time. Specifically, >90% of those classified in the low-symptoms group at T1 remained in this group at T2, suggesting that most adolescents with initially low symptoms levels tended to maintain a healthy status over time, in line with previous longitudinal research (Liang et al., 2023; Xiong et al., 2022). However, it is concerning that 84.6% of adolescents in the emotion-dominant, low-NSSI comorbidity group and 54.8% of adolescents in the NSSI-dominant, high-symptoms comorbidity group remained in the same subgroup nine months later. This suggests a tendency toward consolidation of at-risk emotional and behavioral patterns, potentially placing them at considerable risk for severe psychopathological outcomes.

As for the transitions, adolescents exhibited two distinct directions of change. One involved symptoms improvement or even remit over time. For example, 10.5% of adolescents in the emotion-dominant, low-NSSI comorbidity group at T1 transitioned to the low-symptom group at T2. Similar direction of change has also been observed in previous longitudinal studies on NSSI comorbidity (Liang et al., 2023; Steinhoff et al., 2023; Tilton-Weaver et al., 2019). Such improvements may reflect developmental maturation in self-regulation and social-emotional functioning (Compas et al., 2017; Crone & Dahl, 2012; Kwong et al., 2019), together with possible increases in external support, which enable adolescents to better manage emotion distress and resist maladaptive behaviors. The other transition direction, which deserves particular attention, involved symptoms worsening over time. Specifically, 7.3% of adolescents in the low-symptoms group at T1 transitioned to the emotion-dominant, low-NSSI comorbidity group at T2, and 4.9% of those initially in the emotion-dominant, low-NSSI comorbidity group moved to the NSSI-dominant, high-symptoms Comorbidity group, which is characterized by more severe NSSI and comorbid symptoms. Adolescents exhibiting such symptoms worsening may possess insufficiencies in internal self-regulatory resources, such as lower impulse control, difficulties in emotion regulation, and negative self-related beliefs. These insufficiencies may increase their susceptibility to the onset or escalation of NSSI and comorbid emotional and behavioral problems (Compas et al., 2017; Donnellan et al., 2005; Sowislo & Orth, 2013; You et al., 2015), particularly under stressful conditions. From a contextual perspective, these worsening transitions may also be related to the increasing environmental stress faced by adolescents during this developmental period. Participants in the present study were largely in early to middle adolescence, a period during which academic, family, and interpersonal stressors tend to increase (Branje, 2018; Giota & Gustafsson, 2021). These contextual stressors may activate or amplify pre-existing vulnerabilities, thereby increasing the likelihood that susceptible adolescents would transition toward more severe symptom patterns.

Role of emotion dysregulation, impulsivity and core self-evaluation

According to the developmental psychopathology of self-injury (Yates, 2004), deficits in emotional competence, including emotion regulation and impulse control, and in attitudinal competence, such as self-representations, are key contributors to NSSI. As for the comorbid problems of NSSI, the cognitive-neural mechanism model further identifies emotion dysregulation and heightened impulsivity as key risk factors (Deng et al., 2022), while core self-evaluation, as a positive self-representation, may serve as a protective factor. This study thus focuses on emotion dysregulation, impulsivity, and core self-evaluation to clarify the risk and protective factors underlying NSSI comorbidity patterns and their transitions, providing empirical insight into etiology and informing intervention strategies.

The findings discovered that emotion dysregulation and impulsivity are important risk factors for the comorbidity of NSSI with emotional and behavioral problems. As hypothesized, adolescents with higher levels of emotion dysregulation or impulsivity were more likely to be classified in the at-risk groups (i.e., emotion-dominant, low-NSSI comorbidity group & NSSI-dominant, high-symptoms comorbidity group) than in the low-symptoms group. This result provides empirical support for the cognitive-neural mechanism model of NSSI (Deng et al., 2022). According to this model (Deng et al., 2022), NSSI and its comorbid symptoms share common neurophysiological abnormalities in emotion- and control-related brain regions (e.g., the amygdala, prefrontal cortex and cingulate gyrus), with emotion dysregulation and impulsivity acting as key cognitive-affective processes that link these neurophysiological mechanisms to NSSI and its comorbidities. Furthermore, within the two at-risk groups, adolescents with higher levels of emotion dysregulation were more likely to fall into the NSSI-dominant, high-symptoms comorbidity group than the emotion-dominant, low-NSSI comorbidity group. This suggests that higher “loading” of emotion dysregulation serve as specific risk factor associated with the NSSI-dominant, high-symptoms comorbidity. Indeed, such heightened emotion dysregulation may reflect a general vulnerability in coping capacity (Wolff et al., 2019), predisposing adolescents to rely more frequently on multiple maladaptive behaviors (e.g., NSSI, aggression, delinquency) as means of temporarily alleviating stress and emotional distress (Hamza et al., 2015; Laporte et al., 2021; Wolff et al., 2019). Such maladaptive coping, in turn, intensifies emotional distress (Fang et al., 2025; Selby et al., 2013) and reinforces the cycle of NSSI and broader emotional-behavioral problems, eventually giving rise to the comorbidity pattern characterized by repetitive NSSI and high symptom severity.

More importantly, the present study further revealed the protective role of core self-evaluation in both NSSI comorbidity patterns and their transitions. For comorbidity patterns, adolescents with higher levels of core self-evaluation were more likely to belong to the low-symptoms group than in the at-risk groups. Several theoretical perspectives have emphasized the critical role of a healthy self-system in resisting psychopathological issues (Cicchetti & Rogosch, 2002; Harter, 2012; Kotov et al., 2017). As a higher-order personality trait, core self-evaluation represents an essential component of self-system, improving individuals’ internal functioning by reducing cognitive vulnerability, enhancing self-regulation, and reinforcing adaptive coping strategies (Judge et al., 2003; Kernis, 2003). It has been linked to a wide range of psychopathological outcomes (Li & Nie, 2010) and may also serve a cross-diagnostic protective role in mitigating the effects of NSSI and its comorbid symptoms. For transitions of comorbidity patterns, adolescents with higher levels of core self-evaluation at T1 were less likely to transition into the patterns characterized by more severe symptoms at T2, suggesting that core self-evaluation may act as a buffer that prevents progression toward more severe NSSI comorbidity patterns over time. From the standpoint of symptom onset and escalation, negative self-related constructs, such as low self-esteem, diminished self-worth, and low self-efficacy, are important driving forces of NSSI (Witcher et al., 2025), emotional problems (Metalsky et al., 1993; Paersch et al., 2025), and behavioral problems (Donnellan et al., 2005). However, heightened NSSI, emotional and behavioral problems would further reinforce these negative self-cognitions by increasing feeling of guilt and shame (Witcher et al., 2025), reducing self-worth and confidence (Li et al., 2023; Orth & Robins, 2013), and through negative feedback from the social environment (Murray et al., 2021), thereby leading to the escalation of comorbid symptoms. Within this process, positive core self-evaluations could function as psychological buffers that interrupt the transition from negative self-cognitive motives to the onset of NSSI and its comorbid emotional and behavioral problems, as well as block the escalation of symptoms over time. From the standpoint of adolescent’s developmental and educational contexts, during the study interval, participants progressed to higher grades, which likely exposed them to increased academic demands and examination-related pressure (Chao et al., 2024; Li et al., 2023). In the Chinese educational context, academic achievement is strongly emphasized through family, school, and broader societal expectations (Jiang et al., 2021; Shi et al., 2023), closely linking adolescents’ self-evaluation and sense of personal worth to their academic performance. Previous research on Chinese adolescents has further suggested that academic stress can undermine self-related processes, thereby increasing the risk of mental health problems (Chen et al., 2024; Jiang et al., 2021). Accordingly, in the context of increasing academic pressure, a positive core self-evaluation may serve as a protective factor, buffering adolescents against the onset or escalation of NSSI and comorbid emotional and behavioral problems.

Sex differences

An additional contribution of the present study is the identification of sex differences in NSSI comorbidity profiles and their longitudinal transitions. Specifically, first, the results showed that boys were more likely than girls to be classified into the emotional-dominant, low-NSSI comorbidity group rather than the low-symptom group. This finding may reflect a heightened susceptibility among adolescent boys to a comorbidity pattern in which emotional distress predominates, accompanied by lower level of NSSI and behavioral problems. As noted above, the pattern of emotional-dominant, low-NSSI comorbidity may reflect a “spillover” response to emotional problems, whereby lower levels of NSSI and behavioral problems serve as maladaptive strategies for regulating intense emotional distress (Bresin, 2020; Selby & Joiner, 2013). Compared with girls, boys may encounter greater barriers in expressing negative emotions (Chaplin, 2015; Wang, 2025) and in seeking help for emotional stress (Sheikh et al., 2025), particularly in sociocultural contexts where masculine norms emphasize emotional restraint, independence, and toughness (Vogel et al., 2011). Consequently, when confronted with the complex developmental pressures of early adolescence (Branje, 2018; Giota & Gustafsson, 2021), boys may be more likely to exhibit the emotional-dominant, low-NSSI comorbidity pattern due to insufficient skills to cope with emotional stress. Second, regarding profile transitions, boys in the emotional-dominant, low-NSSI comorbidity group at the first time point were more likely to transition to the low-symptom group nine months later, whereas girls tended to remain in the original group. This finding indicates that, although boys were more likely to exhibit the emotional-dominant, low-NSSI comorbidity pattern, it was less likely to persist among boys than among girls, suggesting that girls with this comorbidity pattern may be at greater risk for symptom persistence and chronicity. Notably, previous research has also reported that, compared with boys, adolescent girls’ NSSI and emotional problems are more likely to persist or recur over time (Gutman & Codiroli McMaster, 2020; Keyes & Platt, 2024; Moloney et al., 2025). These findings imply that interventions for girls at risk of NSSI-related comorbidity, especially those with prominent emotional distress, may need to incorporate sustained intervention components, such as skills reinforcement and relapse-prevention strategies (Cox et al., 2012; Mehlum et al., 2014) to address the potential persistence and chronicity of comorbid symptoms.

Implications

The present findings have both theoretical and practical implications. At the theoretical level, this study provides an empirical context for integrating several perspectives, including the HiTOP system, Bresin’s unifying theory of dysregulated behaviors, and the cognitive-neural mechanism model of NSSI, by revealing the associations between risk factors (emotion dysregulation and impulsivity) and protective factor (core self-evaluation) and the NSSI comorbidity. These findings help to advance current understanding of the co-occurrence of NSSI with multiple emotional and behavioral problems among adolescents. Beyond this, the present study further reveals the heterogeneous patterns and dynamic nature of NSSI comorbidity from a person-centered perspective, while identifying specific influencing factors that effectively distinguish different comorbidity patterns, providing a more nuanced characterization of adolescent NSSI comorbidity. At the practical level, the present findings also offer important implications for targeted prevention and intervention efforts. First, two distinct comorbidity patterns of NSSI with both emotional and behavioral problems were identified. This finding underscores the need to move beyond a “one-size-fits-all” approach. Instead, tailored intervention strategies should be developed to meet the specific needs of adolescents with different comorbidity patterns. For example, for adolescents exhibiting emotion-dominant, low-NSSI comorbidity pattern, emotional problems take a central position and may act as a driving force behind the escalation of both NSSI and behavioral problems (Bresin, 2020; Selby & Joiner, 2013). Thus, interventions simultaneously addressing NSSI and co-occurring symptoms, with particular attention to emotional problems, may yield greater benefits. Second, the longitudinal findings revealed that although a substantial proportion of adolescents remained in the low-symptoms pattern nine months later, some still transitioned to higher-risk patterns or remained in at-risk comorbidity patterns, highlighting the need for closer attention to these adolescents. This also suggests that school administrators should conduct periodic mental health evaluations and provide timely, targeted support for those adolescents whose NSSI comorbid symptoms persist or worsen. Third, this study revealed that emotion dysregulation and impulsivity emerged as major risk factors, whereas core self-evaluation functioned as a significant protective factor associated with adolescents’ NSSI comorbidity patterns. These findings suggest that intervention development should avoid focusing exclusively on either risk reduction or the enhancement of protective resources. Instead, they point to the value of integrative approaches that simultaneously target emotion-regulation difficulties and impulsive tendencies, while strengthening core self-evaluation–related protective factors such as sense of self-worth, self-efficacy, and perceived control (Elliott et al., 2013; Judge et al., 2003). Such a dual-pathway perspective provides a more comprehensive framework for informing the prevention and intervention of NSSI-related comorbid problems among adolescents.

Limitations and future directions

The following limitations should be considered when interpreting the findings of this study. First, the three schools included in the present study were all located in Northern China, which may limit the representativeness of the sample. Therefore, caution should be exercised when generalizing the current findings to adolescents from other regions. Future studies are needed to replicate and extend the current findings using more diverse and representative samples. Second, although Harman’s single-factor test suggested that common method bias was not a serious concern, all variables were assessed using adolescents’ self-report measures, which may still introduce shared method variance and potentially inflate the observed associations. Future studies could incorporate multiple assessment sources (e.g., parent or teacher reports, clinical assessments) to further reduce potential common method bias. Third, this study adopted a two-wave longitudinal design focusing on early adolescence, which may constrain the generalizability of some findings. It would be better for future studies to investigate the stability and transition of NSSI comorbidity patterns across a longer time frame. Fourth, the current study did not assess potential external influences during the study period, such as psychological interventions or mental health support received by participants, which may have influenced profile transitions. Future studies should further incorporate relevant external factors to improve the comprehensiveness and rigor of longitudinal investigations. Fifth, although this study examined the developmental transitions of distinct NSSI comorbidity patterns, it was unable to further capture the more fine-grained dynamic processes underlying these transitions, including symptom-level associations among NSSI and co-occurring emotional and behavioral problems, as well as the specific mediating mechanisms linking core self-evaluation to profile transitions. Future studies may consider employing longitudinal network analytic approaches (McNally, 2021; Miers et al., 2020) or longitudinal structural equation modeling approaches to provide a more nuanced understanding of symptom-level dynamics and developmental mechanisms underlying transitions in NSSI comorbidity patterns. Sixth, although the current study identified personal-level risk and protective factors of adolescents’ NSSI comorbidity, the role of environmental factors also warrants further attention. Prior research suggests that some distal and proximal environmental factors (e.g., childhood abuse; bully victimization) may also be associated with NSSI comorbidity (Liang et al., 2023; Ye et al., 2012). Future research should consider personal and environmental factors simultaneously to provide more comprehensive evidence for identifying specific risk factors associated with different comorbidity patterns, thereby deepening our understanding of this issue.

Conclusion

The coexistence of NSSI with emotional and behavioral problems warrants greater scientific attention. However, empirical knowledge about the heterogeneous patterns and developmental process of NSSI comorbidity among adolescents remains limited. Furthermore, understanding the underlying risk and protective factors represents a crucial step toward more effective prevention and intervention efforts. Accordingly, the current study conducted LPA and LTA on adolescents’ NSSI, emotional problems, and behavioral problems over a 9-month interval. Three subgroups were identified as low-symptoms, emotion-dominant, low-NSSI comorbidity, and NSSI-dominant, high-symptoms comorbidity. Latent transition analysis revealed three directions of change as stability, symptoms improvement, and symptoms worsening. Emotion dysregulation and impulsivity were identified as key risk factors for comorbidity patterns, with a high loading of emotion dysregulation specifically associated with the membership in NSSI-dominant, high-symptoms comorbidity. Moreover, core self-evaluation was identified as a significant protective factor, not only being cross-sectionally associated with membership in lower-risk patterns, but also exerting a longitudinal buffering effect that prevented progression toward more severe NSSI comorbidity over time. These findings provide important insights for the prevention and intervention of NSSI comorbid problems among adolescents.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Ethics approval

This study was approved by the ethics committee of Institutional Review Board of the Faculty of Psychology, Beijing Normal University (IRB Number: 202112100080). The procedures used in this study adhered to the tenets of the Declaration of Helsinki. Participants provided active informed assent, and their guardians gave written informed consent for the assessment.

Funding

This work was supported by the National Natural Science Foundation of China [Grant Number: 32471115].

Glossary

NSSI: non-suicidal self-injury

HiTOP: the hierarchical taxonomy of psychopathology system

LPA: latent profiles analysis

LTA: latent transition analysis

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.