Introduction

Non-suicidal self-injury (NSSI) is broadly defined as the self-inflicted destruction of body tissue in the absence of suicidal intent (^{Nock, 2010}). It affects between 13% and 32% of adolescents and young adults in community samples (^{Muehlenkamp et al., 2012}; ^{Swannell et al., 2014}), becoming a growing public health concern. Alarmingly, NSSI has been associated with interpersonal difficulties, high levels of psychological suffering, and high risk for suicide (^{Ribeiro et al., 2016}; ^{Vega et al., 2018}). While growing research has significantly increased knowledge about NSSI, the factors involved in the motivations of this high-risk behavior (i.e., functions of NSSI) have yet to be fully elucidated.

Previous studies have supported a two-factor higher-order model to explain functions of NSSI. On the one hand, intrapersonal NSSI-functions involve motivations associated with changes in one's internal state, such as an emotional state or thoughts (e.g., affect regulation) and, on the other hand, interpersonal functions involve motivations associated with changes in the external environment (e.g., interpersonal influence: ^{Klonsky et al., 2015}; ^{Turner et al., 2012}). Among the wide range of NSSI functions commonly reported by people who engage in self-injury, intrapersonal functions are the most prevalent, indicating that individuals typically engage in self-injury to cope with inner tension (^{Taylor et al., 2018}). Consistently, previous results show that intrapersonal functions are related to high emotional reactivity as well as poor emotion regulation skills, while interpersonal functions are mainly associated with social difficulties (^{Andover, 2014}; Turner et al., 2012). For example, in a sample of college students, individuals with high levels of emotion dysregulation reported that they engaged in NSSI mainly with an intrapersonal motivation, whilst those reporting greater difficulties in social relationships incurred in NSSI more commonly with an interpersonal motivation (^{Vega et al., 2017}).

Although emotion dysregulation is considered an important transdiagnostic risk factor for a wide range of psychopathology in adolescents and adults (^{Aldao et al., 2010}; ^{Sheppes et al., 2015}), previous work has reported a specific robust association between emotion (dys)regulation and NSSI (^{Fox et al., 2015}; ^{Wolff et al., 2019}). In this regard, individuals who incur in NSSI exhibit poor emotion regulation compared to non-NSSI individuals (^{Taylor et al., 2018}), and the degree of emotion dysregulation is correlated with the frequency of NSSI (^{Andover, 2014}).

It is therefore not surprising that NSSI is strongly related to emotional disorders such as Borderline Personality Disorder (BPD) (^{Glenn & Klonsky, 2013}). Affective dysregulation and poor emotion regulation skills are principal characteristics of BPD (^{Linehan, 1993}). The emotional cascade model suggests that individuals with BPD experience intense negative affect and excessive rumination that may motivate engaging in dysregulated behaviors (e.g., self-injury) to reduce emotional arousal (^{Selby & Joiner, 2009}).

Closely related to emotion dysregulation, negative urgency is an impulsivity trait that is conceptualized as a disposition or a proclivity to act rashly when experiencing strong negative emotions (^{Cyders & Smith, 2008}). Most importantly, negative urgency is implicated in NSSI (^{Peckham et al., 2020}). Engaging in NSSI has been associated with higher scores in negative urgency (^{Bresin et al., 2013}), indicating a difficulty to control the urge to hurt oneself when experiencing negative emotions. In addition, a prospective study revealed that adolescents with baseline negative urgency are more vulnerable to later engagement in NSSI (^{You et al., 2016}).

While negative urgency has been identified as the impulsivity measure that is most robustly associated with NSSI (^{Hamza et al., 2015}), some dimensions of emotion dysregulation have also been established as more relevant for NSSI than other dimensions. For example, a meta-analysis concluded that limited access to emotion regulation strategies, followed by nonacceptance of emotions, were most robustly associated with NSSI engagement (^{Wolff et al., 2019}). Additionally, a recent study has reported that adolescents with severe NSSI show higher negative affect, less emotion regulation strategies, and more difficulties with impulse control, relative to their less severely affected counterparts (^{Chen & Chun, 2019}). Collectively, these findings highlight the importance of considering measures of both emotion dysregulation and impulsivity to better characterize NSSI behavior.

Notwithstanding the above, there is a dearth of literature exploring the role of specific dimensions of emotion regulation in the engagement of NSSI and with regards to the motivation (intrapersonal vs. interpersonal function) of NSSI. A better understanding of this relationship, both in clinical and nonclinical samples, may help refine interventions for individuals who engage in self-injury. To bridge the gap in extant literature, the current study aimed to investigate the association of specific dimensions of emotion dysregulation with the functions and frequency of NSSI, in a sample of young adults. Innovatively, we included both a clinical group of patients with BPD and NSSI and a subclinical group of college students with NSSI.

Based on prior research suggesting that NSSI is associated with poor emotion regulation skills (^{Taylor et al., 2018}; ^{Wolff et al., 2019}; Fox et al., 2016), it was hypothesized that participants with NSSI (both the clinical and subclinical groups) would present more clinical symptoms and more difficulties with emotion regulation relative to the healthy control group. In turn, this would indicate that NSSI is associated with a worse clinical status and higher emotion dysregulation (hypothesis 1). Second, it was predicted that emotion (dys)regulation (assessed by DERS subscales) would show a strong positive association with intrapersonal functions of NSSI and a weaker or no association with interpersonal functions of NSSI. Specifically, given recent meta-analytic findings (Wolff et al., 2019), it was postulated that limited access to emotion regulation strategies and non-acceptance of emotions would be more strongly associated with intrapersonal (i.e., affect-regulating) functions of NSSI and frequency of NSSI (hypothesis 2). Lastly, given prior findings portraying negative urgency as a vulnerability factor for self-harm (^{You et al., 2016}), it was anticipated that negative urgency would predict NSSI frequency (hypothesis 3). Given no prior research regarding the role of negative urgency in motivations for self-harm, exploratory analyses were conducted to explore whether this impulsivity construct would predict intrapersonal or interpersonal NSSI-functions.

Materials and Methods

Results

Demographics and NSSI-characteristics

Demographical information and NSSI characteristics are shown in Table 1 and psychometric properties of scales in Table S1. The three groups were matched by age (χ² = 2.83, p = .24) and sex (χ² = .60, p = .74). The HC group had more years of formal education than the BPD-NSSI and ST-NSSI groups. Regarding NSSI-functions within self-harm groups, intrapersonal functions were significantly more prevalent than interpersonal functions both in BPD-NSSI (Z = -4.45, p < .001) and ST-NSSI group (Z = -4.37, p < .001). Between-group analysis can be found in Table S2. Notably, intrapersonal NSSI-functions were significantly more prevalent in the BPD-NSSI group than in the ST-NSSI group and, specifically, anti-suicide NSSI-function was higher in the BPD-NSSI group. There were no significant differences in interpersonal NSSI-functions between groups.

Table 1. Sociodemographic characteristics. 

^Note.BPD-NSSI = Border Personality Disorder with Non-suicidal self-injury; ST-NSSI = Student with Non-suicidal self-injury; HC = healthy control group; Others = unemployment or sickness disability

χ²†= Kruskall-Wallis test;

Z = U Mann-Whitney test.

NSSI, emotion dysregulation, and clinical symptoms in 3 groups (hypothesis 1)

Table 2 illustrates between-group differences in clinical measures. As shown, BPD patients with NSSI (BPD-NSSI group) and college students with NSSI (ST-NSSI group) scored higher than healthy controls on all scales, suggesting a higher clinical severity associated with the presence of NSSI. Additionally, the BPD-NSSI group scored higher than the ST-NSSI group in Negative Urgency (UPPS-P), symptoms of Depression (DASS-21), and inability to access emotion regulation Strategies (DERS). No differences (BPD-NSSI vs. ST-NSSI) were found in levels of emotion dysregulation (except on lack of emotion regulation strategies), suggesting that the presence of NSSI is associated with poor emotion regulation. Importantly, BPQ discriminated between the three groups, and specifically, the ST-NSSI group differed significantly from the BPD-NSSI group in borderline pathology.

Table 2. Multiple comparisons for the three groups in self-report measures. 

^Note.BPD-NSSI = Borderline Personality Disorder with Non-Suicidal Self-Injury; ST-NSSI = Student with Non-suicidal self-injury; HC = healthy control group; BPQ = Borderline Personality Questionnaire; UPPS = The Impulsivity Scale; DASS-21 = Depression, Anxiety and Stress Scale; DERS-18 = Difficulties in Emotion Regulation Scale.

^*p < .05,

^**p < .01,

^***p < .001.

χ²† = Kruskall-Wallis test.

Emotion dysregulation, Negative Urgency and NSSI-functions (hypothesis 2)

We assessed whether difficulties in emotion regulation are differentially associated with intrapersonal and interpersonal NSSI functions (Table 3). In these analyses, we controlled for Group (BPD-NSSI vs. ST-NSSI) to account for potential differences associated with the presence of a BPD diagnosis. In addition, since significant correlations were observed between predictors, we assessed multicollinearity by calculating the variance inflation factor (VIF). All VIF values were under 3, indicating that there were no multicollinearity problems (^{Hair, 2010}).

Table 3. Multiple linear regression analysis. 

^Note.ISAS = Inventory of Statements about Self-Injury; BPQ = The Borderline Personality Questionnaire.

First, we used NSSI-intrapersonal functions as the outcome variable. We entered Group and DERS-scales (Awareness, Clarity, Goals, Impulse, Nonacceptance, Strategies) as predictors. Results showed that only the lack of acceptance of one's emotions (i.e., Nonacceptance DERS-scale) significantly predicted NSSI-intrapersonal functions (R² = .47, F(7, 43) = 5.64, p < .001).

We repeated the previous analysis with NSSI-interpersonal functions as the outcome. When Group (BPD-NSSI vs. ST-NSSI) and DERS-scales were entered simultaneously as predictors, none of these variables significantly predicted NSSI-interpersonal functions. However, there was a moderate trend for clarity about the nature of one's emotions (i.e., Clarity DERS-scale) (Table 3).

We also evaluated whether Negative urgency, together with Group (BPD-NSSI vs. ST-NSSI), significantly predicted intrapersonal and interpersonal NSSI-functions. Surprisingly, negative urgency did not predict any NSSI-functions (all p >.17).

All analyses were repeated using BPQ (i.e., severity of borderline symptoms) as an additional covariate, and results remained the same (supplementary material). Collectively, these results suggest that individuals with high negative urgency in the current sample are not more likely to incur to NSSI with an intrapersonal or an interpersonal motivation. However, results also suggest that out of all emotion regulation difficulties (DERS-scales), participants with specific difficulties in acceptance of emotions were more likely to engage in NSSI with an intrapersonal motivation, even after controlling for BPD severity.

Frequency of NSSI: Emotion dysregulation and Negative urgency (hypothesis 3)

According to hypothesis 3, we explored whether DERS-scales and Negative Urgency could potentially predict NSSI-lifetime frequency. In a first model, Group (BPD-NSSI vs. ST-NSSI) and DERS-scales were entered simultaneously as predictors, yet none of these scales predicted NSSI-lifetime frequency (all p > .13). Lastly, when negative urgency was included in a second model with Group as a control variable, it showed no predictive effect on NSSI-lifetime frequency (all p < .62).

Additionally, we examined possible interactions between DERS-scales and Negative urgency. Results revealed a significant interaction between Strategies DERS-scale and Negative urgency in predicting NSSI-lifetime frequency (Strategies by Urgency = .32, p < .05; R² = .20, F(3, 40) = 3.46, p <. 05), even after controlling for BPQ (R² =.21, F(4, 39) = 2.71, p < .05). To decompose the interaction, we conducted simple slopes analysis to assess whether Strategies DERS-scale predicts NSSI-lifetime frequency for individuals with high and low Negative urgency (^{Preacher et al., 2006}). We draw on previous studies to consider Negative Urgency as a moderator (^{Bresin et al., 2013}). First, we mean-centered the two predictors of the interaction and then estimated individuals with high and low (±1 S.D) negative urgency. We found that DERS-strategies was significantly associated with NSSI-lifetime frequency (B = .24, p < .05) in individuals with high negative urgency, but not in individuals with low negative urgency (B = -.04, p = .58) (Figure 1).

Figure 1. Association between Lack of Emotion Regulation Strategies and NSSI-frequency moderated by Negative Urgency. 

Discussion

Innovatively, the present study investigated whether specific dimensions of emotion dysregulation and negative urgency are differentially associated with NSSI functions and NSSI frequency. Innovatively, we recruited both a clinical and a subclinical group of young adults characterized by frequent NSSI engagement (and a group of healthy controls). Our findings indicate that a specific dimension of emotion dysregulation, namely nonacceptance of emotions, uniquely predicted intrapersonal motivations for engaging in NSSI, even after controlling for BPD symptom severity. Importantly, results also revealed an interaction between lack of emotion regulation strategies and negative urgency in predicting NSSI-frequency.

While not previously documented, the unique association of nonacceptance of emotions (DERS-scale) with intrapersonal NSSI-functions may cohere with prior work in the field. For instance, a recent meta-analysis (^{Taylor et al., 2018}) found that individuals most commonly engage in self-injury with an intrapersonal motivation (between 66 to 81%), most commonly to avoid or escape distressing emotions. A lack of acceptance of emotions implies rejecting states of distress by attempting to extinguish the emotion altogether or responding to the emotion with secondary negative emotions such as shame or guilt (^{Gratz & Roemer, 2004}). Accordingly, non-acceptance of emotions could be perceived as an experiential avoidance strategy that may promote engaging in NSSI when individuals are faced with intense negative emotions (^{Chapman et al., 2006}; ^{Selby & Joiner, 2009}). Indeed, previous studies have found that experiential avoidance and thought suppression strategies -constructs related to nonacceptance of emotions (Gratz & Roemer, 2004)- have been associated with an increased risk of NSSI (^{Chapman et al., 2005}, ²⁰⁰⁶). Importantly, nonacceptance of emotions has also been associated with an increased risk of maintenance of NSSI. Compared to college students who stopped engaging in self-injury one year ago, students who continued engaging in NSSI behaviors showed significantly less acceptance of emotional responses, regardless of gender (^{Anderson & Crowther, 2012}). Interestingly, efforts to suppress or avoid a negative emotion may have a counter-effect, increasing rather than decreasing psychological distress (^{Kashdan et al., 2006}). Therefore, recurrent use of these strategies in the presence of high-intensity emotions may serve as a stepping-stone towards other maladaptive strategies, such as NSSI, when emotional avoidance alone is no longer effective.

Consistent with our hypothesis, we did not find a significant association between any emotion dysregulation dimensions and interpersonal NSSI-functions. As reported in previous literature, interpersonal motivations for NSSI are less common (^{Taylor et al., 2018}), especially among individuals with high levels of emotion dysregulation (^{Vega et al., 2017}), as is the case in the current sample. Interpersonal NSSI functions such as communicating distress or seeking support (Taylor et al., 2018) reflect deficits in interpersonal emotion regulation skills that are not necessarily captured by the DERS scale. Only lack of emotional clarity DERS-scale approached a significant association with interpersonal NSSI-functions. This trend may reflect deficits in identification and communication of emotions (^{Gratz & Roemer, 2004}; ^{Gross, 2015}), which may stem from invalidating environments and dysfunctional parenting that detriment the emotion regulation capacity (^{Crowell et al., 2009}; ^{Linehan, 1993}). Indeed, a recent study found that an invalidating environment characterized by childhood maltreatment was specifically related to deficits in emotional clarity, which were in turn associated with increased probability of NSSI in young adults (^{Guérin-Marion et al., 2020}). These deficits, added to a context of interpersonal conflicts (^{Adrian et al., 2011}), may motivate engagement in NSSI. However, these relationships should be explored in larger samples.

Additionally, we found that the association between lack of emotion regulation strategies (DERS-dimension) and NSSI-frequency was moderated by negative urgency. Specifically, among individuals who reported a lack of emotion regulation strategies, only those with higher negative urgency, that is, a proclivity to act rashly when experiencing intense negative emotions, were likely to engage in more frequent NSSI. These findings are consistent with meta-analyses that identify limited access to emotion regulation strategies and negative urgency as the most robust predictors of NSSI (^{Sheppes et al., 2015}; ^{You et al., 2016}) and with biosocial models that characterize NSSI by emotion dysregulation and impulsivity (^{Kashdan et al., 2006}; ^{Lieb et al., 2004}). Effectively, the observed interaction highlights negative urgency as a critical risk factor for NSSI and corroborates prior research reporting that negative affect, specifically daily sadness, only predicted NSSI engagement in college students with high negative urgency but not with low urgency (^{Peckham et al., 2020}).

Notably, while negative urgency predicted NSSI-frequency (in interaction with lack of emotion regulation strategies), it did not predict motivations for self-harm. This null finding is not surprising, given that negative urgency interferes with NSSI only after NSSI urges surpass a temporal or intensity threshold. As recently study demonstrated, negative urgency in NSSI is not associated with deficits in restraining an emotionally impulsive act before it is initiated, but it is rather associated with deficits in terminating an ongoing negative emotional response after it is initiated (^{Allen & Hooley, 2019}). Such findings suggest that negative urgency may not be involved in the motivations for engaging in NSSI, but it may be involved in readily transforming the NSSI urge into an NSSI act.

Additionally, when comparing the NSSI groups with healthy controls, individuals with NSSI scored higher on all clinical measures. As predicted, the BPD-NSSI group presented higher severity (e.g., psychopathology) than the student group, which is also reflected in the higher prevalence of intrapersonal NSSI-functions. Importantly, both the BPD-NSSI and ST-NSSI groups showed similar levels of emotion dysregulation and did not differ in age onset, frequency, and number of methods used in NSSI. Complementarily, both BPD and student groups engaged in NSSI with intrapersonal functions more frequently than with interpersonal functions, which reflects that the main motivation for NSSI in both groups was to regulate an internal state of emotional distress. These findings are consistent with a study reporting that intrapersonal NSSI-functions are associated with affective dysregulation in college students (^{Vega et al., 2017}). Given the growing prevalence of NSSI among young adults in community populations (^{Muehlenkamp et al., 2012}; ^{Swannell et al., 2014}), exploring the characteristics of NSSI in subclinical samples may be particularly pertinent. Findings indicate that the phenotypic characteristics of NSSI may not differ for clinical and subclinical groups with frequent NSSI.

Our results suggest that specific dimensions of emotion dysregulation may provide unique information about different NSSI features. Precisely, nonacceptance of emotions may predict the motivation for engaging in NSSI (i.e., NSSI used as an experiential avoidance strategy), while lack of emotion regulation strategies (in the presence of negative urgency) may predict increased frequency and maintenance of NSSI. Addressing these emotion regulation deficits in therapy could have a direct impact on NSSI. Interestingly, Dialectical Behavioral Therapy (DBT), specifically DBT skills training (^{Linehan, 1993}; 2014), has been shown to effectively reduce NSSI acts (^{Linehan et al., 2015}) by improving acceptance skills (^{Krantz et al., 2018}). Therefore, therapeutic interventions which target emotion acceptance and emotion regulation (e.g., DBT) may be a necessary goal for patients with NSSI. Correspondingly, by learning acceptance and mindfulness skills, patients can learn to tolerate disruptive emotions that can trigger urgency of NSSI, and by learning emotion regulation skills they can learn to identify and describe emotions that precede the NSSI act and obtain greater cognitive and behavioral control (^{Linehan et al., 2014}).

Despite promising findings, this study has some important limitations that need to be addressed to inform cautious interpretation of the data. Firstly, the low sample size may have led to less robust estimates as well as some null findings, therefore the results presented should be confirmed by future research with larger sample sizes. Secondly, the BPD-NSSI and ST-NSSI groups demonstrated considerable homogeneity in emotion dysregulation, which may have attenuated associations and reduced power to detect subtle differences. Also, our data is based on self-reports, therefore, information about retrospective NSSI may be under or overestimated (or may entail memory biases). In addition, the clinical measures only assessed the past few weeks, while NSSI-frequency was measured over a longer timeframe; therefore, comparison of these measures could be biased. Ecological momentary assessment may offer a better alternative to capture clinical information related to NSSI acts in real time. Lastly, we used a cross-sectional design, whereas longitudinal studies could be more informative of the NSSI trajectory.

Conclusion

Results reveal facets of emotion dysregulation that are uniquely associated with the functions and frequency of NSSI and they highlight the need for emotion regulation skills and distress tolerance (to prevent negative urgency) as a common treatment target in psychotherapy. While preliminary, our findings are expected to have clinical implications and may serve as a guide for more proactive interventions, especially in subclinical groups of people with NSSI.