Introduction

Suicide is one of the leading causes of death, accounting for nearly 800,000 deaths each year worldwide (World Health Organization, 2018). Rates of suicide within the United States continue to increase, particularly among college-aged individuals (Centers for Disease Control and Prevention, 2018). Suicidal thoughts and behaviors are even more prevalent than deaths by suicide (Nock et al., 2008), and lead to substantial societal and economic costs (Shepard et al., 2016). In 1995, nearly 10% of college students reported seriously considering suicide, and 2% reported a suicide attempt within the past year (Kisch et al., 2005). More recent estimates suggest that over half of college students endorse suicidal ideation (Drum et al., 2009), and in 2019, the suicide rate among youth aged 15–24 years was 13.95 per 100,000 individuals (American Foundation for Suicide Prevention, 2019). In a recent study sampling undergraduate and graduate students across 79 universities in the U.S., about 13% reported past-year suicidal ideation, 6% reported making a suicide plan in the past year, and just over 1% reported a past-year suicide attempt (Oh et al., 2021). Given the high risk of suicidal thoughts and behaviors in college samples and the seriousness of this problem, research that allows us to better understand the factors that contribute to heightened suicide risk is urgently needed. Two such factors that may contribute to elevated risk in this population are emotion dysregulation (Law et al., 2015) and engagement in self-damaging behaviors (Barrios et al., 2000).

Broadly, emotion regulation refers to the process by which individuals modulate their emotions in response to environmental demands and personal goals (Gross, 1998). As such, emotion dysregulation encompasses nonacceptance of emotional responses, difficulties engaging in goal-directed behavior, impulse control difficulties, lack of emotional awareness, limited access to emotion regulation strategies, and lack of emotional clarity (Gratz & Roemer, 2004). Although some research points to emotion dysregulation as a risk factor for suicidal thoughts and behaviors (Arria et al., 2009; Brezo et al., 2008), other studies offer only partial or mixed support (Anestis et al., 2011; Law et al., 2015). Recent findings indicate that this relationship is complex and likely influenced by other variables, such as engagement in behaviors that serve to regulate emotions and that increase suicidal capacity by reducing fear of death and increasing tolerance of physical pain (e.g., Law et al., 2015). One well-established theory of suicide, the interpersonal-psychological theory of suicidal behavior (Joiner, 2005; Van Orden et al., 2010), proposes that people engage in suicidal behaviors (i.e., attempts) because (a) they want to (what is termed suicidal desire, often due to perceived burdensomeness, disconnection, or other sources of distress); and (b) they can (what is termed capability). This model rests on the assumption that many people have innate barriers to directly inflicting damage to their bodies. Thus, those individuals who engage in suicidal behaviors must have the capability to overcome these barriers. This model also proposes that repeated experiences with dangerous and painful events may, over time, increase the capability for suicide. An implication of this model is that psychological pain alone may lead to suicidal desire, but not behaviors, in the absence of some capability for self-inflicted injury. From this standpoint, emotion dysregulation may reflect elevated distress and the inability to regulate this distress, thereby contributing to elevated desire for suicide. Indeed, low distress tolerance, one component of emotion dysregulation (Gratz & Roemer, 2004), is associated with views of oneself as a burden to others and perceived lack of meaningful interpersonal relationships (Anestis et al., 2011), which are both theorized to generate suicidal desire (Joiner, 2005). However, not all who have the desire for suicide are capable of acting on these thoughts/urges (i.e., it is relatively difficult and painful to die by suicide). This capability for suicide may be in part accomplished through engaging in self-damaging behaviors (via reduced fear of death and increased pain tolerance; Joiner, 2005; Law et al., 2015). Therefore, among those who have not developed this reduced fear of death and increased pain tolerance (and consequently the enhanced capability for suicide), the link between emotion dysregulation and suicide risk may be relatively weak.

One factor that may influence the strength of the association between emotion dysregulation and suicide risk is engagement in self-damaging behaviors. These behaviors encompass any deliberate behavior with a high potential for causing physical harm to oneself (Turner et al., 2015), including non-suicidal self-injury (NSSI), substance misuse, and disordered eating (St. Germain & Hooley, 2012). While the former behavior (i.e., NSSI) involves direct damage to bodily tissue, the latter two behaviors (i.e., substance misuse, disordered eating) may be conceptualized as “indirect self-injury,” involving the potential for physical damage over time and socioemotional difficulties (St. Germain & Hooley, 2012), albeit in the absence of any intent to cause self-injury. Such behaviors are prevalent among college students: up to 38% of college student samples have endorsed some history of NSSI (Gratz et al., 2002), 40% have reported misusing substances (O’Malley & Johnston, 2002), and approximately 14% have reported clinically significant eating disorder symptoms (Eisenberg et al., 2011). Although these behaviors can frequently co-occur (e.g., Serras et al., 2010), past research has tended to examine specific self-damaging behaviors on their own, complicating our understanding of the links between multiple self-damaging behaviors and maladaptive outcomes. Several studies have documented the associations between self-damaging behaviors and suicidal behaviors (e.g., Turner et al., 2013). Meta-analyses have demonstrated links between both substance misuse and NSSI and increased suicide risk (Franklin et al., 2017; Ribeiro et al., 2016). Furthermore, studies have shown that NSSI (Franklin et al., 2011), substance misuse (Bohnert et al., 2017), and disordered eating (Franko & Keel, 2006) are each associated with increased suicide capability/decreased pain perception, increased risk for death by suicide, and increased risk of suicide attempts and death by suicide, respectively. Specifically, among college students, a range of potentially self-damaging behaviors, including substance misuse, physical fights, and lack of safety precautions (i.e., rarely or never wearing a seat belt in the car; carrying a weapon; riding with a driver who had been drinking alcohol; driving after drinking alcohol), have been associated with increased suicidal ideation (Barrios et al., 2000). Suicide capability is theorized to stem from repeated exposure and eventual habituation to the fear and pain involved in hurting oneself (Van Orden et al., 2008). Thus, engagement in self-damaging behaviors may serve as “practice” for the very act of harming oneself, thereby increasing suicide capability (e.g., Anestis et al., 2015; Van Orden et al., 2008).

It is important to note that these self-damaging behaviors are typically not engaged in for the purpose of suicide. In fact, the very definition of NSSI entails a lack of suicidal intent (ISSS, 2018), and the most common reason people report for engaging in NSSI is to reduce negative emotions (Nock & Prinstein, 2004). Although not everyone who engages in NSSI also attempts suicide (Hamza & Willoughby, 2013), self-injury remains one of the most robust predictors of suicide (Joiner et al., 2005). The interpersonal-psychological theory (Joiner, 2005) helps us to understand this, suggesting that, regardless of intent, repetitive engagement in these behaviors may erode fears of death and simultaneously increase tolerance of painful stimuli. In this manner, NSSI is somewhat of a double-threat–it may be both a marker of distress, which could lead to suicidal desire, and also increase, over time, the capability of acting on that desire.

In line with this theory, the number of both past suicide attempts, and painful and provocative events in an individual’s life (e.g., NSSI, purging), are associated with elevated suicide capability (Van Orden et al., 2008), and elevated pain tolerance has been found to account for this association (Franklin et al., 2011). Despite these established associations, it remains to be understood how engagement in multiple forms of self-damaging behaviors may create an additive effect with respect to suicide risk. This is an important empirical question given that risky behaviors tend to co-occur (Serras et al., 2010).

Engagement in numerous self-damaging behaviors may magnify the relation between emotion dysregulation and suicide risk. That is, it is possible that individuals with multiple co-occurring self-damaging behaviors may be at particularly heightened risk for suicide (Turner et al., 2013). Such a relationship is critical to understand among college students, in light of evidence suggesting that this population often engages in multiple forms of self-damaging behaviors (e.g., Serras et al., 2010). College students with emotion dysregulation who also engage in multiple forms of self-damaging behaviors may be at particularly high risk for suicide. It is also likely that emotion dysregulation per se may increase suicide risk in college students (Kisch et al., 2005). Furthermore, recent research has proposed that emotion dysregulation may be associated with elevated suicide risk, particularly among those who utilize painful and/or provocative coping methods, such as NSSI, to regulate intense and distressing emotions (Law et al., 2015). In keeping with the interpersonal-psychological theory (Joiner, 2005), if emotion dysregulation contributes to suicidal desire via increased perceptions of oneself as a burden and as lacking in meaningful relationships (Anestis et al., 2011), risk for suicidal behavior is likely to increase in the presence of suicide capability (Van Orden et al., 2008). In other words, the presence of an elevated desire for suicide does not necessarily increase one’s risk for engaging in lethal self-directed behavior; one must also be able to act on such desires or have the capability to do so. Therefore, suicide risk may be particularly high in the presence of both emotion dysregulation (high suicidal desire) and multiple self-damaging behaviors (high suicide capability).

Another plausible way of understanding the link between these constructs is to conceptualize self-damaging behaviors as a possible mediator of the relationship between emotion dysregulation and suicide risk. In other words, emotion dysregulation may be indirectly associated with increased risk for suicide by increasing engagement in self-damaging behaviors. Engagement in self-damaging behaviors may reflect one’s propensity to act out on the body as a way to cope with internal distress (Brausch et al., 2011). Given that emotion dysregulation encompasses difficulties controlling impulses and having less effective ways of regulating intense emotions (Gratz & Roemer, 2004), these deficits may lead to increased engagement in risky and impulsive behaviors (e.g., Van der Kolk et al., 1996) as maladaptive ways of coping with this distress and attempting to reduce negative emotion (e.g., Bonn-Miller et al., 2008; Nock & Prinstein, 2004; Whiteside et al., 2007). Research supports the association between emotion dysregulation and engagement in a range of self-destructive behaviors (Weiss et al., 2015), including disordered eating and substance misuse (Aldao et al., 2010). Theories of NSSI also suggest that this behavior helps individuals escape from unwanted emotional experiences (Chapman et al., 2006). Taken together, research underscores the link between emotion dysregulation and a range of self-damaging behaviors, as these behaviors may serve to help cope with or escape from distressing negative emotions, at least in the short-term. In turn, engagement in multiple forms of these behaviors could then, via habituation to the pain and fear inherent in repeatedly hurting oneself, lead to increased risk for suicide attempts or death by suicide (Van Orden et al., 2008). NSSI has indeed been found to mediate the link between emotion dysregulation and suicide attempts (Anestis et al., 2014). However, no studies to our knowledge have examined engagement in multiple forms of self-damaging behaviors as a potential mechanism linking emotion dysregulation and suicide risk. Backed by theoretical and some empirical support, both moderation and mediation models are logically feasible, although no studies have simultaneously tested both within the same sample.

Our study’s aim was therefore to examine and clarify associations between emotion dysregulation, self-damaging behaviors, and suicide risk within a sample of college students. Specifically, we first examined the association between the interaction of emotion dysregulation and multiple self-damaging behaviors (i.e., NSSI, alcohol misuse, substance misuse, and disordered eating), and suicide risk. We hypothesized that engaging in more forms of self-damaging behaviors would confer higher risk for suicide, and that this association would be particularly strong in the context of greater emotion dysregulation. Second, we examined whether engagement in multiple forms of self-damaging behaviors might also serve as a mediator of the link between emotion dysregulation and suicide risk. We hypothesized that emotion dysregulation would be indirectly associated with elevated suicide risk through engagement in multiple forms of self-damaging behaviors.

Method

Participants

Participants (N = 181; Mage = 20.0 years, SD = 2.2; 82.9% female) were recruited through the psychology subject pool at a large northeastern university for a parent study (n = 183, with 2 excluded due to missing data on the dependent variable for the current analyses) on emotion regulation, borderline personality disorder, and risky behaviors in daily life. We included individuals with a range of borderline personality disorder (BPD) features based on the Personality Assessment Inventory–Borderline Features Scale (PAI-BOR; Morey, 1991), and oversampled individuals with elevated BPD features (PAI-BOR score > 38 based on a departmental mass prescreening survey) by sending personalized emails to these individuals, although no specific score was required to enter the parent study. Inclusion criteria for the parent study specified that participants be at least 18 years old, be able to read and complete online questionnaires in English, and be able to speak fluently in English. Racial/ethnic breakdown based on self-report was as follows: 66.9% White, 15.5% Asian/Southeast Asian, 6.6% Black/African American, 5.0% multiracial, 3.9% Hispanic/Latinx, and 2.2% another race/ethnicity. Among participating students in the current analyses (n = 181), 14.4% reported currently taking medications for a psychiatric disorder, and 19.3% reported a history of psychological treatment (i.e., medication or therapy/counseling) for a psychiatric disorder. Mean yearly family/household income was $53,092 (SD = $35,236).

Measures

Demographics

Participants completed a questionnaire assessing various aspects of their social identities and other demographics (i.e., race/ethnicity, age, sex, sexual orientation, relationship status, education status, yearly household income, employment status, current psychiatric or other medication use, lifetime psychiatric treatment, weight, height, use of hormonal contraceptives, menstruation status).

Emotion Regulation Difficulties

The Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) is a 36-item measure that assesses habitual difficulties with regulating emotions (i.e., nonacceptance of emotional responses, difficulties engaging in goal-directed behavior, impulse control difficulties, lack of emotional awareness, limited access to emotion regulation strategies, and lack of emotional clarity). Participants rate items on a 5-point Likert scale in which higher scores indicate greater emotion dysregulation and certain items are reverse-scored. The DERS total score ranges from 36 to 180 and has demonstrated excellent internal consistency among a sample of undergraduates (Cronbach’s α = .93), good test-retest reliability over four to eight weeks (ICC = .88), and good construct and predictive validity (Gratz & Roemer, 2004; Gratz & Tull, 2010). In our study, the internal consistency of the total scale was α = .95.

Self-Damaging Behaviors

We operationalized self-damaging behaviors as the total number (out of a maximum of 4) of distinct self-damaging behaviors reported (i.e., NSSI, alcohol misuse, drug misuse, disordered eating behavior) that were above established clinical thresholds (each behavioral measure listed next). Specifically, we first determined the presence of each type of self-damaging behavior, and then summed the total number of self-damaging behaviors in excess of their respective thresholds.

Non-Suicidal Self-Injury (NSSI)

We used the Deliberate Self-Harm Inventory (DSHI) to assess lifetime engagement in NSSI across 17 behaviors that included cutting, burning, and severe scratching (Gratz, 2001). For our study, we calculated total frequency across behaviors and the total number of methods. Those who endorsed having any history of NSSI (i.e., frequency > 0) were considered to exhibit this specific self-damaging behavior (i.e., given a score of “1” vs. “0”). The total number of methods ranged from 0 to 9. The DSHI has demonstrated good internal consistency (Cronbach’s α = .82), adequate test-retest reliability, and adequate construct, discriminant, and convergent validity in undergraduate samples (Gratz, 2001).

Alcohol Use

The Alcohol Use Disorders Identification Test (AUDIT) is a 10-item measure that assesses patterns of general alcohol consumption, drinking behaviors, and past-year alcohol-related problems (Babor & Grant, 1989). The total AUDIT score could range from 0 to 40. Consistent with established guidelines, participants who received a total score of 8 or higher on the AUDIT out of the possible 40 were considered to engage in harmful alcohol use (Bohn et al., 1995), and in our study were given a rescore of “1” (vs. “0”). The AUDIT has demonstrated good reliability and validity (Saunders et al., 1993). In our sample, the internal consistency of the AUDIT was α = .86.

Drug Use

The Drug Use Disorders Identification Test (DUDIT) is an 11-item measure that assesses general drug use and past-year drug-related problems (Berman et al., 2005). The first 9 items are scored on a 5-point Likert scale (ranging from 0-4), and the last 2 items are scored on a 3-point Likert scale (0, 2, or 4). The total DUDIT score could range from 0 to 44. Consistent with past research, we used a cut-off score of 8 out of the possible 44 to identify which participants qualified as having drug use problems (Voluse et al., 2012), and those with 8 or higher were rescored to a “1” (vs. “0”). The DUDIT has demonstrated good reliability and was significantly associated with DSM-IV substance use disorders in both clinical and non-clinical samples (Berman et al., 2005). In our sample, the internal consistency of the DUDIT was α = .89.

Disordered Eating Behaviors

The Eating Disorder Diagnostic Scale (EDDS) is a 22-item measure that assesses current (i.e., past 3 months) body image concerns and engagement in Eating Disorder (ED) behaviors, including food restriction, binge eating, and other compensatory behaviors (Stice et al., 2000). Scoring the EDDS involves examining whether a participant met DSM-IV criteria for any of the diagnosable EDs assessed by the measure (i.e., anorexia nervosa, bulimia nervosa, binge eating disorder; Stice et al., 2000). For our study, any participant who met diagnostic criteria for any of the three probable ED diagnoses was given a score of “1” (vs. “0”). The EDDS has demonstrated good criterion validity with interview-based diagnoses, convergent validity with relevant risk factors, and internal consistency (Stice et al., 2004). Internal consistency of the EDDS in our sample was α = .79.

Total Self-Damaging Behaviors

We computed a total self-damaging behaviors score for each participant. Each self-damaging behavior (i.e., NSSI, alcohol misuse, drug misuse, and disordered eating behavior) for which a participant met criteria earned that participant a score of 1, with total scores ranging from 0 to 4.

Suicide Risk

The Suicidal Behaviors Questionnaire-Revised (SBQ-R) is a 4-item measure that assesses the following variables related to suicidality: (1) lifetime suicidal ideation and/or suicide attempts (scored on a Likert scale ranging from 1 [never] to 4 [I have attempted to kill myself, and really hoped to die]); (2) past-year frequency of suicidal ideation (scored on a Likert scale ranging from 1 [never] to 5 [very often – 5 or more times]); (3) lifetime disclosure of suicidal desire or intent to others (scored on a Likert scale ranging from 1 [no] to 3 [yes, more than once, and really wanted to do it]; and (4) likelihood of a future suicide attempt (scored on a Likert scale ranging from 0 [never] to 6 [very likely]; Osman et al., 2001). Based on previous research with undergraduate samples, a total score of 7 or above (out of a possible continuous range of 3-18) indicates risk for suicide, with higher scores indicating higher suicide risk (Osman et al., 2001). Internal consistency of the SBQ-R in our sample was α = .83.

Procedure

We obtained written (virtually) informed consent from all participants included in the study. Participants completed an online survey using Qualtrics survey software, which consisted of various self-reported measures of psychological and behavioral functioning. Participants were compensated with course credit for completing the study, and those who completed the study were virtually displayed a debriefing form that contained mood improvement activities (e.g., funny video clip links), as well as local and national mental health resources (including the National Suicide Prevention Lifeline). All procedures were approved by the University of Massachusetts Amherst Institutional Review Board.

Data Analytic Plan

Preliminary Analyses

We first examined descriptive statistics related to the study’s primary variables. In order to assess the associations between overall emotion dysregulation, self-damaging behaviors, and suicide risk, we then computed Pearson and point-biserial (for binary variables) correlations.

Primary Analyses

For our test of moderation, we first examined the main effects of overall emotion dysregulation and number of self-damaging behaviors on suicide risk in Step 1 of a multiple regression model. We included the interaction effect in Step 2 using the SPSS Process macro. We probed significant interactions with simple slopes (M ± 1SD) and Johnson-Neyman regions of significance. We centered independent variables at their respective sample means prior to entry in the models and the calculation of the interaction term. For the mediation model, we used SPSS Process with overall emotion dysregulation as the independent variable, number of self-damaging behaviors as the mediator, and suicide risk as the dependent variable. We report the effects of the independent variable on the dependent variable (path c), the independent variable on the mediator (path a), and the mediator on the dependent variable (path b; controlling for the independent variable). We examined the bootstrapped (5000 bootstraps) confidence intervals to determine significance of the indirect effect of the independent variable on the dependent variable via the mediator (ab; Preacher & Hayes, 2004). We also examined the direct effect of the independent variable on the dependent variable with the mediator in the model (path c’).

Results

Preliminary Analyses

Descriptive statistics of the primary study variables are shown in Table 1. Table 2 displays the intercorrelations among primary study variables.

Table 1 Descriptive statistics of main study variables
Full size table
Table 2 Intercorrelations among study variables
Full size table

Primary Analyses

Self-Damaging Behaviors as a Moderator

Table 3 displays the results of the regression analyses. A multiple regression analysis demonstrated significant main effects of both self-damaging behaviors and emotion dysregulation on suicide risk. Specifically, when holding emotion dysregulation constant at the sample mean, a 1-unit increase in self-damaging behaviors was associated with a .94-unit increase in suicide risk. Holding self-damaging behaviors constant at the sample mean, a 1-unit increase in emotion dysregulation was associated with a .05-unit increase in suicide risk.

Table 3 Results of primary and exploratory analyses
Full size table

There was a significant interactive effect of overall emotion dysregulation and self-damaging behaviors on suicide risk (see Figure 1). The effect of emotion dysregulation on suicide risk was significant, and larger at higher (M+1SD: β = 0.07, SE = 0.01, p < .001) vs. lower (minimum observed value [M-.99SD]: β = 0.03, SE = 0.01, p = .009) levels of self-damaging behaviors. The Johnson-Neyman test indicated that the transition point was outside of the range of data on the self-damaging behaviors variable. Therefore, we exchanged the independent and moderator variables. The Johnson-Neyman test indicated that the association between self-damaging behaviors and suicide risk was only significant (and positive) for individuals with scores on the DERS above 74. The model with the interaction term included explained 36% of the variance in suicide risk.

Fig. 1
figure 1

Interaction between emotion dysregulation and self-damaging behaviors in predicting suicide risk. The effect of emotion dysregulation on suicide risk was significant and larger at higher vs. lower levels of self-damaging behaviors. Emotion dysregulation is plotted at 1 SD above and below the mean, and self-damaging behaviors is plotted at 1 SD above the mean, and 0.99 SD below the mean.

Full size image

Self-Damaging Behaviors as a Mediator

Table 3 also displays the results of the mediation analyses. Results revealed that all paths a, b, and c were significant. Further, there was a significant indirect effect of overall emotion dysregulation on suicide risk through number of self-damaging behaviors. The direct effect of emotion dysregulation on suicide risk with self-damaging behaviors in the model (path c’) was also significant. This model explained 34% of the variance in suicide risk, which is relatively comparable to that of the moderation model.

Exploratory Analyses

We also explored whether the severity of self-damaging behaviors would both interact with and mediate the effect of emotion dysregulation on suicide risk. To answer these questions, we calculated z scores for self-damaging behaviors (i.e., NSSI frequency; number of NSSI methods; alcohol misuse; drug misuse; and presence of probable anorexia nervosa, bulimia nervosa, and binge eating disorder). We calculated means for NSSI (across frequency and number of methods) and probable eating disorders (across the three disorders). We computed severity of self-damaging behaviors scores by summing the z scores for NSSI, alcohol misuse, drug misuse, and disordered eating. Given non-normal distributions, we log-transformed (base 10; after adding 3.31 units) scores, and then centered at the sample mean for the moderation analysis.

Analyses revealed effects of both self-damaging behavior severity (β = 2.72, SE = 0.79, p = .001) and emotion dysregulation (β = 0.05, SE = 0.008, p <.001) on suicide risk, in addition to a significant interaction between severity of self-damaging behaviors and emotion dysregulation (β = 0.06, SE = 0.03, p = 0.046). The association of emotion dysregulation on suicide risk was stronger at high (M+1SD: β = 0.07, SE = 0.01, p < 0.001) vs. low (M-1SD: β = 0.04, SE = 0.01, p = 0.005) levels of self-damaging behavior severity. Likewise, mediation analyses revealed a significant indirect effect of emotion dysregulation on suicide risk through severity of self-damaging behaviors (see Table 3).

Discussion

Our study’s aim was to clarify, within a sample of college students, the role of multiple self-damaging behaviors (i.e., NSSI, alcohol and substance misuse, and probable eating disorders) in the relationship between emotion dysregulation and suicide risk by examining both moderation and mediation models. Consistent with past work, both emotion dysregulation and number of self-damaging methods were associated with increased suicide risk. These predictors interacted to incrementally confer greater suicide risk, supporting a moderation model. Further, self-damaging behaviors also helped explain the link between emotion dysregulation and elevated suicide risk, supporting a mediation model.

Preliminary correlational analyses indicated that emotion dysregulation demonstrated significant zero-order associations with multiple forms of self-damaging behaviors, consistent with past work (Law et al., 2015). Emotion dysregulation was positively associated with drug and alcohol misuse (Dvorak et al., 2014), as well as NSSI frequency (Gratz, 2003) and methods, findings that are consistent with past research in college samples. Of the probable eating disorders assessed, only bulimia nervosa was significantly associated with emotion dysregulation at the zero-order level. Not surprisingly, students’ self-damaging behaviors were positively inter-correlated, aligning with research suggesting that risky behaviors tend to co-occur (Serras et al., 2010). In addition, NSSI frequency and methods, alcohol misuse, and probable bulimia nervosa were all positively correlated with suicide risk (e.g., Franko & Keel, 2006; Victor & Klonsky, 2014).

Importantly, our findings offer clarification of the mixed support for the relationship between emotion dysregulation and suicide risk found in prior studies (Anestis et al., 2015). Correlational analyses suggested a positive link between emotion dysregulation and suicide risk. Furthermore, results highlight the moderating role of self-damaging behaviors in this association. Our findings suggest that if the average individual endorses both emotion dysregulation and multiple self-damaging behaviors, they are at particularly high risk for suicidal thoughts and behaviors. Of note, our results revealed that the relationship between number of self-damaging behaviors and suicide risk was not present among those with low emotion dysregulation (DERS<74), just below average scores in university student samples (Gratz & Roemer, 2004). Therefore, for those who do not struggle with regulating their emotions, the presence of multiple risky behaviors may not increase their risk for suicide, a finding that is somewhat reassuring given the prevalence of such behaviors among college students (e.g., Eisenberg et al., 2011; Gratz et al., 2002). Whereas emotion dysregulation may confer risk for suicidal desire, frequent engagement in a range of distinct self-damaging behaviors may provide a pathway to acquire the capability to overcome innate barriers to harming oneself, thereby increasing one’s capacity for suicide. These two factors may be a particularly potent combination, interacting to contribute to heightened suicide risk, consistent with theoretical models of suicide (Joiner, 2005). Of note, we replicated this pattern of findings when considering self-damaging behavior severity as an indicator, consistent with previous research examining individual risky behaviors (e.g., Paul et al., 2015; Stewart et al., 2017). This suggests some correspondence between multiple forms of self-damaging behaviors and greater severity.

An alternate mediation model supported the role of number of self-damaging behaviors in explaining the link between overall emotion dysregulation and elevated suicide risk. This model explained slightly less variance in suicide risk than the moderation model, although the effects were relatively comparable. This is consistent with past research pointing to engagement in NSSI as one mechanism explaining the relationship between emotion dysregulation and suicide attempts in particular (Anestis et al., 2014). However, to our knowledge, our study was the first to examine the role of multiple self-damaging behaviors in explaining this link. Individuals who struggle with regulating intense emotions may be more likely to turn to maladaptive ways of coping through these risky, self-destructive behaviors in order to reduce this distress (e.g., Bonn-Miller et al., 2008; Whiteside et al., 2007). However, over time, these self-destructive behaviors may erode barriers to physically harming oneself, consistent with the interpersonal-psychological theory (Joiner, 2005), via habituation to pain and/or reduction of fear (Van Orden et al., 2008), thereby increasing suicide risk.

There are several potential reasons that the number of distinct self-damaging behaviors employed by an individual is particularly risky in terms of suicidal outcomes. First, an individual who resorts to multiple different forms of self-damaging behaviors may more quickly overcome various hurdles for engagement in their destruction of their bodily tissue, thereby increasing their suicide capability. Second, an individual who indiscriminately engages in multiple modes of self-damaging behaviors may be particularly desperate for relief from distressing negative emotions. The specific mechanism by which multiple self-damaging behaviors leads to heightened suicide risk warrants further examination.

This study had several limitations. First, because we utilized cross-sectional data, we were unable to disentangle the temporal sequence of these distinct factors. Future research utilizing a longitudinal study design would allow us to examine how the associations between emotion dysregulation, self-damaging behaviors, and suicide risk evolve over time. This type of study design would also allow us to explore whether the presence of emotion dysregulation leads to engagement in self-damaging behaviors, or whether self-damaging behaviors reinforce use of maladaptive emotion regulation strategies in the short-term, and lead to increased difficulties in regulating distressing emotions over time. Second, this study relied exclusively on self-report measures, which are subject to many problems including recall bias. Future research with student samples would benefit from use of behavioral measures of emotion dysregulation or implicit measures of suicide to more objectively measure the constructs of interest and reduce such bias. Furthermore, we used a count variable for number of self-damaging behaviors, restricting the range of this variable and thus limiting variability and destabilizing regression coefficients. As such, the use of a continuous measure in future studies will be important. This issue, which is not unique to count variables, applies to all predictors with small variance (Coxe et al., 2009). Third, we had low rates of clinically significant levels of anorexia nervosa, which may reflect the relatively lower prevalence rates of anorexia than bulimia nervosa in the general population (Hudson, Hiripi, Pope Jr., & Kessler, 2007). Fourth, the model including the interaction effect only resulted in a modest (2%) improvement in the prediction of variance in suicide risk. Thus, results should be interpreted with caution, and further work is needed to better understand variables that substantially contribute to heightened risk for suicide. Finally, this study examined a nonclinical college sample (with participants reporting a range of BPD features); the sample was relatively homogeneous in terms of its racial make-up, and it is important to note that participants recruited through the psychology subject pool may not represent all college students (e.g., they may be more psychologically-minded or curious; they may be more or less likely to have dealt with their own mental health concerns). Although rates of suicidal thoughts and behaviors are reported at concerningly high rates among this population (Nock et al., 2008; Wilcox et al., 2010), findings may not be generalizable to treatment-seeking samples, other age groups, or more diverse populations.

This line of research has important clinical implications for treatment providers who work with college-aged individuals and college administrations broadly. While we focused on a narrow set of self-damaging behaviors, the accumulation of any set of risky behaviors or potent stressors in a student’s profile may additively affect risk. Given the high prevalence of suicide risk and self-damaging behaviors in college samples, it is vital that school administrators proactively seek to prevent suicidal thoughts and behaviors by allocating resources to provide support to students at a population-wide level. Administrators should consider reducing barriers to counseling access (e.g., Downs & Eisenberg, 2012), incorporating mental health screening measures into routine health center visits, training resident advisors in listening skills and referring for help when needed (e.g., Tompkins & Witt, 2009), and having instructors flag students whose grades appear to be suffering or whose behavior seems erratic. Taking a preventive approach may offset the cost of suicidal thoughts and behaviors on young adults’ health and their associated societal costs (Goldman-Mellor et al., 2014).

Among those who are already at elevated risk for suicide outcomes, findings highlight the dire need for access to quality mental health care on college campuses. Given the high prevalence of self-damaging behaviors among college students, it is crucial that providers understand that an individual who struggles with intense emotions and who may turn to self-damaging behaviors to regulate these emotions may be at heightened risk of suicide; risk may be considered to be even greater for individuals who endorse multiple forms of self-damaging behaviors. Therefore, it is important for clinicians to conduct a thorough assessment of such behaviors, as well as risk for suicide, including suicidal thoughts, urges, reasons for living, access to means, and other well-known risk factors for suicide. Our findings also highlight the need for understanding an individual’s repertoire of emotion regulation skills when assessing suicide risk, particularly in the context of multiple self-damaging behaviors. Given that these behaviors can serve as maladaptive emotion regulation strategies for some (Nock & Prinstein, 2004), teaching more adaptive emotion regulation skills to at-risk college students might be an important step in reducing their suicide risk.