164
Suicide and Life-Threatening Behavior 45 (2) April 2015 © 2014 The American Association of Suicidology DOI: 10.1111/sltb.12120
How Many Times and How Many Ways: The Impact of Number of Nonsuicidal Self-Injury Methods on the Relationship Between Nonsuicidal Self-Injury Frequency and Suicidal Behavior MICHAEL D. ANESTIS, PHD, LAUREN R. KHAZEM, BA, AND KEYNE C. LAW, BA (HONS.)
Several variables have been proposed as heavily influencing or explaining the association between nonsuicidal self-injury (NSSI) and suicidal behavior. We propose that increased comfort with bodily harm may serve as an incrementally valuable variable to consider. We sought to indirectly test this possibility by examining the moderating role of number of NSSI methods utilized on the relationship between NSSI frequency and lifetime number of suicide attempts, positing that increased variability in methods would be indicative with a greater general comfort with inflicting harm upon one’s own body. In both a large sample of emerging adults (n = 1,317) and a subsample with at least one prior suicide attempt (n = 143), results were consistent with our hypothesis. In both samples, the interaction term was significant, with the relationship between NSSI frequency and suicidal behavior increasing in magnitude from low to mean to high levels of NSSI methods. Although frequency of NSSI is robustly associated with suicidal behavior, the magnitude of that relationship increases as an individual engages in a wider variety of NSSI methods. We propose that this may be due to an increased comfort with the general concept of damaging one’s own body resulting from a broader selection of methods for self-harm.
It has been repeatedly shown that nonsuicidal self-injury (NSSI) is associated with both suicidal ideation (e.g., Laye-Gindhu & Schonert-Reichl, 2005; Lloyd-Richardson, MICHAEL D. ANESTIS, LAUREN R. KHAZEM, and KEYNE C. LAW , Department of Psychology, University of Southern Mississippi, Hattiesburg, Mississippi, USA. Address correspondence to Michael D. Anestis, PhD, Nina Bell Suggs Professor of Psychology, Department of Psychology, University of Southern Mississippi, 118 College Drive, Box 5025, Hattiesburg, MS 39406; E-mail: michael. [email protected]
Perrine, Dierker, & Kelley, 2007) and future suicide attempts (e.g., Cooper et al., 2005; Nock, Joiner, Gordon, Lloyd-Richardson, & Prinstein, 2006; Whitlock & Knox, 2007). Indeed, NSSI has been identified as one of the most robust predictors of suicidal behavior (Arasnow et al., 2011; Wilkinson, Kelvin, Roberts, Dubicka, & Goodyear, 2011). In this sense, although NSSI and suicidal behavior are distinct in a number of ways, they nonetheless maintain a robust association with one another. Although no single comprehensive explanation has been put forth, past research has suggested several
ANESTIS
ET AL.
plausible mechanisms (mediators) and contextual factors (moderators) driving the relationship between these behaviors. One proposed mechanism is shared psychopathology. For instance, depressive symptoms have been consistently found to be an indicator of suicide risk (Davidson, Wingate, Grant, Judah, & Mills, 2011; Holma et al., 2010). In fact, individuals who have engaged in both NSSI and suicide attempts have reported elevated levels of depression compared with those who engage in NSSI only (Dougherty et al., 2009). At the same time, Klonsky, May, and Glenn (2013) found the relationship between NSSI and attempted suicide to be stronger than the relationship between depression and attempted suicide. In fact, in their sample, NSSI and not depression was significantly associated with attempted suicide. Emotion dysregulation has also been associated with both NSSI and suicidal behavior. Specifically, the ability to regulate emotions has been found to be negatively related to suicidal ideation (Ciarrochi, Dean, & Anderson, 2002; Rajappa, Gallagher, & Miranda, 2012), plans, and attempts (Tamas et al., 2007; Zlotnick, Donaldson, Spirito, & Pearlstein, 1997; Zlotnick, Wolfsdorf, Johnson, & Spirito, 2003). Furthermore, emotion dysregulation has been found to be greater in individuals with multiple attempts compared with those with single attempts (Esposito, Spirito, Boergers, & Donaldson, 2003). Moreover, it has been suggested that emotion dysregulation may increase levels of NSSI by lowering pain perception, and thus increasing the willingness to experience the pain associated with selfinjury (Franklin, Aaron, Arthur, Shorkey, & Printsein, 2012). Past research has also found that emotion dysregulation is associated with a diminished acquired capability for suicide, and that greater levels of painful and provocative experiences (NSSI) in some emotionally dysregulated individuals mediates the relationship between emotion dysregulation and suicidal behavior (Anestis, Bagge, Tull, & Joiner, 2011; Anestis, Coffey, Schumacher, & Tull, 2011; Anestis, Gratz, Bagge, & Tull, 2012; Anestis & Joiner, 2012; Pennings &
165 Anestis, 2013). In this sense, emotion dysregulation may facilitate suicidal desire while simultaneously serving as an obstacle to the capacity for suicidal behavior in emotionally dysregulated individuals who do not engage in highly elevated rates of painful and provocative behaviors. Another variable that may either explain or amplify the relationship between NSSI and suicidal behavior is the acquired capability for suicide. According to the interpersonal-psychological theory of suicide (Joiner, 2005; Van Orden et al., 2010), suicidal desire alone does not lead to death by suicide, as most individuals lack the ability to act on their suicidal desire. Only after repeated exposure to painful and provocative events, through which an individual can habituate to physiological pain and develop a fearlessness regarding death, is it possible for individuals to acquire the capability to die by suicide. It is important to note that NSSI has already been proposed as a means to acquire the capability for suicide (Joiner, 2005; Van Orden, Witte, Gordon, Bender, & Joiner, 2008). In this sense, the relationship between NSSI and suicidal behavior could theoretically be explained by the longterm impact of NSSI on an individual’s relationship with and response to pain and death, eventually facilitating suicidal behavior among those with suicidal desire. While the aforementioned theories regarding the relationship between NSSI and suicidal behavior are certainly plausible, another possibility lies within factors that have not been directly measured thus far. As discussed, individuals engaging in NSSI may have increased levels of acquired capability for suicide. However, it remains unclear, precisely how NSSI influences acquired capability. Repeated NSSI has been shown to be associated with increased pain tolerance (Bresin & Gordon, 2013; Hooley, Ho, Slater, & Lockshin, 2010; McCoy, Fremouw, & McNeil, 2010; Russ, Campbell, Kakuma, Harrison, & Zanine, 1999; Russ, Roth, Lerman, & Kakuma, 1992); however, it has also been proposed that the acquired capability involves a decreased fear of bodily
166 harm (Joiner, 2005), and this fear of bodily harm is not currently assessed in any measure of the acquired capability. We propose that NSSI—and particularly the number of NSSI methods used—may increase comfort with self-inflicted bodily harm. Previous research has supported this possibility, as the number of different NSSI methods used has been shown to be associated with the amount of lifetime suicide attempts (Nock et al., 2006), severity of suicidal behavior, frequency of suicidal ideation, and overall suicide risk (Turner, Layden, Butler, & Chapman, 2013). Additionally, compared with individuals who only reported NSSI behaviors involving cutting, individuals who reported NSSI behaviors involving cutting, burning, and bruising were found to be at a greater overall risk for suicide (Turner et al., 2013). Once the number of methods used for NSSI was accounted for, however, the frequency of NSSI was only associated with the severity of suicidal behavior, indicating that a willingness to utilize more than one method may be pivotal in enhancing risk. We propose that versatility in NSSI methods may foster comfort with bodily harm in general rather than simply reflecting comfort with one specific method of selfinjury. An individual who engages in a single NSSI method would theoretically be relatively comfortable engaging in that particular form of self-directed violence, given the evidence that NSSI is driven primarily by the drive to reduce the intensity of negative affect (e.g., Chapman & Dixon-Gordon, 2007; Nock & Prinstein, 2004). It is unclear, however, to what extent that comfort extends beyond that specific behavior. On the other hand, individuals who repeatedly engage in a variety of methods of self-inflicted bodily harm may find reinforcement less from the sensations associated with a specific familiar behavior but rather from the general infliction of bodily harm. If this were the case, frequently engaging in NSSI may directly impact the pain tolerance component of the acquired capability regardless of versatility in methods, but versatility may enhance the degree to which engagement in NSSI impacts
HOW MANY TIMES/WAYS an individual’s comfort with the possibility of inflicting harm upon his or her body. We do not posit that other variables proposed to impact the relationship between NSSI and suicidal behavior are unfounded; however, we believe that comfort with bodily harm may represent an additional factor capable of providing incrementally valuable information. To our knowledge, no direct measure of this construct exists (measures of acquired capability only measure fear of death), thereby requiring indirect tests of our model. If our proposed explanation of the relation of NSSI to suicidal behavior were true, we would expect the relationship between how often an individual engages in NSSI and lifetime number of suicide attempts to increase in strength as the number of NSSI methods increases. The frequency of the individual’s engagement in the behavior would not be seen as irrelevant to suicidal behavior—as it likely increases pain tolerance, facilitating only one component of the acquired capability for suicide—but its relevance would be more pronounced in individuals who engage in a greater variety of NSSI methods. Such individuals would likely exhibit not only a familiarity with harming themselves in a limited number of specific ways, but likely also a comfort with the notion of bodily harm in general. The primary aim of the current study was to examine how the number of methods used for NSSI impacts the relationship between NSSI frequency and suicidal behavior. Specifically, we wanted to determine whether or not the amount of NSSI methods an individual has utilized moderates the relationship between lifetime NSSI frequency and lifetime number of suicide attempts. In this sense, our study builds on extant research by looking specifically at suicidal behavior rather than composite suicide risk variables that include a range of disparate outcomes (e.g., suicidal ideation, talking about suicide; e.g., Turner et al., 2013). We first aimed to test our hypothesis in a large sample of emerging adults to determine the extent to which this model is supported in a sample with varying levels of experience with
ANESTIS
ET AL.
167
both NSSI and suicidal behavior. We then examined the same model only within individuals with a past history of suicidal behavior to clarify the extent to which NSSI methods would distinguish individuals who have experienced various levels of suicidal behaviors (as opposed to largely differentiating those with and without a past history of suicidal behavior). Considering the broad range of theoretically plausible mechanisms past research has suggested might influence suicidal behaviors (e.g., depression, emotion dysregulation, acquired capability, perceived burdensomeness, thwarted belongingness, and fear of death), we tested our hypothesized model with and without controlling for the aforementioned risk factors to demonstrate a clear, robust fit for our model.
METHOD
Participants Participants in the full sample were 1,317 adults (78.8% female) ranging in age from 18 to 69 (M = 21.11; SD = 5.46). The sample was drawn from undergraduates enrolled in an introductory psychology course who registered for and completed
the study online through a secure link. Procedures were approved by the institutional review board of the University of Southern Mississippi and all participants provided informed consent prior to participation. Participants in the subsample of attempters were 143 adults (79.0% female) ranging in age from 18 to 45 (M = 21.38; SD = 4.87). Participants in this subsample were selected from the larger sample based on their self-reported history of suicidal behavior. All participants received course credit for their participation (Table 1). Measures Predictor/Moderator. The Deliberate Self-Harm Inventory (DSHI; Gratz, 2001) is a 17-item self-report questionnaire assessing the lifetime frequency, duration, and severity of NSSI behaviors across a range of methods. A total score equivalent to the number of lifetime NSSI episodes across methods was computed for each participant (referred to as NSSI Frequency). Additionally, a total score equivalent to the number of methods each participant endorsed utilizing at least once was computed (referred to as NSSI Methods).
TABLE 1
Demographic Information for the Full Sample (n = 1317) and the Subsample of Attempters (n = 143) Full sample Sex Male Female Family Income $0–$10,000 $10,001–$25,000 $25,001–$50,000 $50,001–$75,000 $75,001–$100,00 $100,000+ Race/Ethnicity White African American Other
N (%) 274 (20.8%) 1038 (78.8%) 123 189 324 296 199 182
(9.3%) (14.4%) (24.6%) (22.5%) (15.1%) (13.8%)
727 (55.2%) 507 (38.5%) 75 (5.7%)
Attempters Sex Male Female Family Income $0–$10,000 $10,001–$25,000 $25,001–$50,000 $50,001–$75,000 $75,001–$100,00 $100,000+ Race/Ethnicity White African American Other
N (%) 28 (19.6%) 113 (79.0%) 20 26 37 29 14 16
(14.0%) (18.2%) (25.9%) (20.3%) (9.8%) (11.2%)
73 (60.4%) 42 (34.7%) 5 (4.1%)
168 Covariates. Covariates were chosen based on variables that have previously been proposed as suicide risk factors and which could represent alternative variables that might better account for suicidal behavior than would our proposed independent variable and moderator. The Depression Anxiety Stress Scale (DASS; Lovibond & Lovibond, 1995) is a 21-item self-report questionnaire assessing severity of depression, anxiety, and stress symptoms over the course of the previous 7 days. The Depression subscale was utilized as a covariate in the current study and Cronbach’s alpha for this subscale in the present sample was .88. The Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) is a 36-item self-report measure that assesses individuals’ general capacity for identifying, understanding, accepting, and adaptively responding to negative affective states. The scale yields six subscale scores as well as a total score indicative of overall emotion dysregulation. Items are scored on a Likert-type scale ranging from 1 (Almost never [0–10%]) to 5 (Almost always [91– 100%]), with higher scores indicating greater emotion dysregulation. Total score was utilized as a covariate, and the alpha coefficient in this sample was .94. The Interpersonal Needs Questionnaire-15 (INQ-15; Van Orden et al., 2008; Van Orden, Witte, Cukrowicz, & Joiner, 2012) is a questionnaire that was developed to measure thwarted belongingness (nine items) and perceived burdensomeness (six items). Items are rated on a scale from 1 (Not at all true for me) to 7 (Very true for me). The alpha coefficients for burdensomeness and belongingness in this sample were .94 and .91, respectively. The Acquired Capability for Suicide Scale–Fear of Death Scale (ACSS-FDS; Ribeiro et al., 2014) is a 7-item version of the original 20-item ACSS (Bender, Gordon, Bresin, & Joiner, 2011) used to measure an individual’s fearlessness about lethal selfinjury. The 7-item version was developed through factor analysis and has been utilized
HOW MANY TIMES/WAYS previously in undergraduate samples (Witte, Gordon, Smith, & Van Orden, 2012). Ratings for each item ranged from 1 (Not at all like me) to 5 (very much like me). The alpha coefficient in this sample was .76. Outcome. The Measure of Episodic Planning of Suicide (MEPOS; Anestis, Pennings, & Williams, 2014) is a brief selfreport questionnaire based closely on the Lifetime Suicide Attempt Self-Injury Interview (L-SASI; Linehan & Comtois, 1996). First, participants are asked, “Throughout the course of your life, have you ever intentionally harmed yourself with at least some intent to cause your own death?” Participants who indicate that they have done so at least once are then asked a series of follow-up questions. Items assess total number of past attempts (Over the course of your entire life, how many times have you intentionally harmed yourself with at least some intention of causing your own death?), intent to die for past attempts (Of those events, how many times were you certain you wanted to die? Of those events, how many times were you uncertain if you truly wanted to die?), level of medical attention received in response to past attempts (Of those events, how many times did your behavior require medical attention [e.g., visit to hospital or other medical facility]?), method used for the most recent attempt (What method did you use in your most recent suicide attempt [e.g., firearms, prescription medication]?), level of intent for the most recent attempt (During your most recent attempt, to what extent did you want to die?), and level of planning for most recent attempt (Prior to your most recent attempt, how much had you thought about trying to die by suicide using the method you ultimately chose?). For items asking the number of times an event had occurred, participants entered a number (software prevented participants from entering text). In this project, participants were included from the larger sample if they endorsed at least one prior incident of suicidal behavior with clear or ambivalent intent to die. The number of such instances (total attempts with clear intent plus total attempts with ambivalent intent) was then computed for each participant and served as the outcome variable.
ANESTIS
ET AL.
Data Analytic Procedure. Following transformations (described below), we conducted a hierarchical multiple regression analysis to examine the potential moderating effect of NSSI Methods on the relationship between NSSI Frequency and lifetime number of suicide attempts. Both the predictor (NSSI Frequency) and the moderator (NSSI Methods) were grand-mean centered to facilitate the interpretation of a significant two-way interaction. In step one of the equation, the covariates and the main effects of the predictor and moderator were entered. In step two, the interaction of the predictor and moderator were entered. Lifetime number of suicide attempts (with ambivalent or clear intent to die) served as the outcome variable. Simple slopes analyses were used to test the nature and direction of the two-way interaction.
RESULTS
Descriptive data and intercorrelations for the variables utilized in the primary analyses are provided in Table 2.
Variable Transformations NSSI Frequency (Full sample skew = 29.03; Full sample kurtosis = 933.95; Attempter skew = 7.12; Attempter kurtosis = 56.78), NSSI Methods (Full sample skew = 2.94; Full sample kurtosis = 11.00; Attempter skew = 1.04; Attempter kurtosis = 1.14), and suicide attempts (Full sample skew = 8.36; Full sample kurtosis = 102.87; Attempter skew = 3.89; Attempter kurtosis = 20.25) exhibited significant skew and/or kurtosis. In an effort to approximate normality within distributions, we performed rank transformations using Blom’s formula, which resulted in acceptable levels (Kline, 2005) of skewness (
Suicide and Life-Threatening Behavior 45 (2) April 2015 © 2014 The American Association of Suicidology DOI: 10.1111/sltb.12120
How Many Times and How Many Ways: The Impact of Number of Nonsuicidal Self-Injury Methods on the Relationship Between Nonsuicidal Self-Injury Frequency and Suicidal Behavior MICHAEL D. ANESTIS, PHD, LAUREN R. KHAZEM, BA, AND KEYNE C. LAW, BA (HONS.)
Several variables have been proposed as heavily influencing or explaining the association between nonsuicidal self-injury (NSSI) and suicidal behavior. We propose that increased comfort with bodily harm may serve as an incrementally valuable variable to consider. We sought to indirectly test this possibility by examining the moderating role of number of NSSI methods utilized on the relationship between NSSI frequency and lifetime number of suicide attempts, positing that increased variability in methods would be indicative with a greater general comfort with inflicting harm upon one’s own body. In both a large sample of emerging adults (n = 1,317) and a subsample with at least one prior suicide attempt (n = 143), results were consistent with our hypothesis. In both samples, the interaction term was significant, with the relationship between NSSI frequency and suicidal behavior increasing in magnitude from low to mean to high levels of NSSI methods. Although frequency of NSSI is robustly associated with suicidal behavior, the magnitude of that relationship increases as an individual engages in a wider variety of NSSI methods. We propose that this may be due to an increased comfort with the general concept of damaging one’s own body resulting from a broader selection of methods for self-harm.
It has been repeatedly shown that nonsuicidal self-injury (NSSI) is associated with both suicidal ideation (e.g., Laye-Gindhu & Schonert-Reichl, 2005; Lloyd-Richardson, MICHAEL D. ANESTIS, LAUREN R. KHAZEM, and KEYNE C. LAW , Department of Psychology, University of Southern Mississippi, Hattiesburg, Mississippi, USA. Address correspondence to Michael D. Anestis, PhD, Nina Bell Suggs Professor of Psychology, Department of Psychology, University of Southern Mississippi, 118 College Drive, Box 5025, Hattiesburg, MS 39406; E-mail: michael. [email protected]
Perrine, Dierker, & Kelley, 2007) and future suicide attempts (e.g., Cooper et al., 2005; Nock, Joiner, Gordon, Lloyd-Richardson, & Prinstein, 2006; Whitlock & Knox, 2007). Indeed, NSSI has been identified as one of the most robust predictors of suicidal behavior (Arasnow et al., 2011; Wilkinson, Kelvin, Roberts, Dubicka, & Goodyear, 2011). In this sense, although NSSI and suicidal behavior are distinct in a number of ways, they nonetheless maintain a robust association with one another. Although no single comprehensive explanation has been put forth, past research has suggested several
ANESTIS
ET AL.
plausible mechanisms (mediators) and contextual factors (moderators) driving the relationship between these behaviors. One proposed mechanism is shared psychopathology. For instance, depressive symptoms have been consistently found to be an indicator of suicide risk (Davidson, Wingate, Grant, Judah, & Mills, 2011; Holma et al., 2010). In fact, individuals who have engaged in both NSSI and suicide attempts have reported elevated levels of depression compared with those who engage in NSSI only (Dougherty et al., 2009). At the same time, Klonsky, May, and Glenn (2013) found the relationship between NSSI and attempted suicide to be stronger than the relationship between depression and attempted suicide. In fact, in their sample, NSSI and not depression was significantly associated with attempted suicide. Emotion dysregulation has also been associated with both NSSI and suicidal behavior. Specifically, the ability to regulate emotions has been found to be negatively related to suicidal ideation (Ciarrochi, Dean, & Anderson, 2002; Rajappa, Gallagher, & Miranda, 2012), plans, and attempts (Tamas et al., 2007; Zlotnick, Donaldson, Spirito, & Pearlstein, 1997; Zlotnick, Wolfsdorf, Johnson, & Spirito, 2003). Furthermore, emotion dysregulation has been found to be greater in individuals with multiple attempts compared with those with single attempts (Esposito, Spirito, Boergers, & Donaldson, 2003). Moreover, it has been suggested that emotion dysregulation may increase levels of NSSI by lowering pain perception, and thus increasing the willingness to experience the pain associated with selfinjury (Franklin, Aaron, Arthur, Shorkey, & Printsein, 2012). Past research has also found that emotion dysregulation is associated with a diminished acquired capability for suicide, and that greater levels of painful and provocative experiences (NSSI) in some emotionally dysregulated individuals mediates the relationship between emotion dysregulation and suicidal behavior (Anestis, Bagge, Tull, & Joiner, 2011; Anestis, Coffey, Schumacher, & Tull, 2011; Anestis, Gratz, Bagge, & Tull, 2012; Anestis & Joiner, 2012; Pennings &
165 Anestis, 2013). In this sense, emotion dysregulation may facilitate suicidal desire while simultaneously serving as an obstacle to the capacity for suicidal behavior in emotionally dysregulated individuals who do not engage in highly elevated rates of painful and provocative behaviors. Another variable that may either explain or amplify the relationship between NSSI and suicidal behavior is the acquired capability for suicide. According to the interpersonal-psychological theory of suicide (Joiner, 2005; Van Orden et al., 2010), suicidal desire alone does not lead to death by suicide, as most individuals lack the ability to act on their suicidal desire. Only after repeated exposure to painful and provocative events, through which an individual can habituate to physiological pain and develop a fearlessness regarding death, is it possible for individuals to acquire the capability to die by suicide. It is important to note that NSSI has already been proposed as a means to acquire the capability for suicide (Joiner, 2005; Van Orden, Witte, Gordon, Bender, & Joiner, 2008). In this sense, the relationship between NSSI and suicidal behavior could theoretically be explained by the longterm impact of NSSI on an individual’s relationship with and response to pain and death, eventually facilitating suicidal behavior among those with suicidal desire. While the aforementioned theories regarding the relationship between NSSI and suicidal behavior are certainly plausible, another possibility lies within factors that have not been directly measured thus far. As discussed, individuals engaging in NSSI may have increased levels of acquired capability for suicide. However, it remains unclear, precisely how NSSI influences acquired capability. Repeated NSSI has been shown to be associated with increased pain tolerance (Bresin & Gordon, 2013; Hooley, Ho, Slater, & Lockshin, 2010; McCoy, Fremouw, & McNeil, 2010; Russ, Campbell, Kakuma, Harrison, & Zanine, 1999; Russ, Roth, Lerman, & Kakuma, 1992); however, it has also been proposed that the acquired capability involves a decreased fear of bodily
166 harm (Joiner, 2005), and this fear of bodily harm is not currently assessed in any measure of the acquired capability. We propose that NSSI—and particularly the number of NSSI methods used—may increase comfort with self-inflicted bodily harm. Previous research has supported this possibility, as the number of different NSSI methods used has been shown to be associated with the amount of lifetime suicide attempts (Nock et al., 2006), severity of suicidal behavior, frequency of suicidal ideation, and overall suicide risk (Turner, Layden, Butler, & Chapman, 2013). Additionally, compared with individuals who only reported NSSI behaviors involving cutting, individuals who reported NSSI behaviors involving cutting, burning, and bruising were found to be at a greater overall risk for suicide (Turner et al., 2013). Once the number of methods used for NSSI was accounted for, however, the frequency of NSSI was only associated with the severity of suicidal behavior, indicating that a willingness to utilize more than one method may be pivotal in enhancing risk. We propose that versatility in NSSI methods may foster comfort with bodily harm in general rather than simply reflecting comfort with one specific method of selfinjury. An individual who engages in a single NSSI method would theoretically be relatively comfortable engaging in that particular form of self-directed violence, given the evidence that NSSI is driven primarily by the drive to reduce the intensity of negative affect (e.g., Chapman & Dixon-Gordon, 2007; Nock & Prinstein, 2004). It is unclear, however, to what extent that comfort extends beyond that specific behavior. On the other hand, individuals who repeatedly engage in a variety of methods of self-inflicted bodily harm may find reinforcement less from the sensations associated with a specific familiar behavior but rather from the general infliction of bodily harm. If this were the case, frequently engaging in NSSI may directly impact the pain tolerance component of the acquired capability regardless of versatility in methods, but versatility may enhance the degree to which engagement in NSSI impacts
HOW MANY TIMES/WAYS an individual’s comfort with the possibility of inflicting harm upon his or her body. We do not posit that other variables proposed to impact the relationship between NSSI and suicidal behavior are unfounded; however, we believe that comfort with bodily harm may represent an additional factor capable of providing incrementally valuable information. To our knowledge, no direct measure of this construct exists (measures of acquired capability only measure fear of death), thereby requiring indirect tests of our model. If our proposed explanation of the relation of NSSI to suicidal behavior were true, we would expect the relationship between how often an individual engages in NSSI and lifetime number of suicide attempts to increase in strength as the number of NSSI methods increases. The frequency of the individual’s engagement in the behavior would not be seen as irrelevant to suicidal behavior—as it likely increases pain tolerance, facilitating only one component of the acquired capability for suicide—but its relevance would be more pronounced in individuals who engage in a greater variety of NSSI methods. Such individuals would likely exhibit not only a familiarity with harming themselves in a limited number of specific ways, but likely also a comfort with the notion of bodily harm in general. The primary aim of the current study was to examine how the number of methods used for NSSI impacts the relationship between NSSI frequency and suicidal behavior. Specifically, we wanted to determine whether or not the amount of NSSI methods an individual has utilized moderates the relationship between lifetime NSSI frequency and lifetime number of suicide attempts. In this sense, our study builds on extant research by looking specifically at suicidal behavior rather than composite suicide risk variables that include a range of disparate outcomes (e.g., suicidal ideation, talking about suicide; e.g., Turner et al., 2013). We first aimed to test our hypothesis in a large sample of emerging adults to determine the extent to which this model is supported in a sample with varying levels of experience with
ANESTIS
ET AL.
167
both NSSI and suicidal behavior. We then examined the same model only within individuals with a past history of suicidal behavior to clarify the extent to which NSSI methods would distinguish individuals who have experienced various levels of suicidal behaviors (as opposed to largely differentiating those with and without a past history of suicidal behavior). Considering the broad range of theoretically plausible mechanisms past research has suggested might influence suicidal behaviors (e.g., depression, emotion dysregulation, acquired capability, perceived burdensomeness, thwarted belongingness, and fear of death), we tested our hypothesized model with and without controlling for the aforementioned risk factors to demonstrate a clear, robust fit for our model.
METHOD
Participants Participants in the full sample were 1,317 adults (78.8% female) ranging in age from 18 to 69 (M = 21.11; SD = 5.46). The sample was drawn from undergraduates enrolled in an introductory psychology course who registered for and completed
the study online through a secure link. Procedures were approved by the institutional review board of the University of Southern Mississippi and all participants provided informed consent prior to participation. Participants in the subsample of attempters were 143 adults (79.0% female) ranging in age from 18 to 45 (M = 21.38; SD = 4.87). Participants in this subsample were selected from the larger sample based on their self-reported history of suicidal behavior. All participants received course credit for their participation (Table 1). Measures Predictor/Moderator. The Deliberate Self-Harm Inventory (DSHI; Gratz, 2001) is a 17-item self-report questionnaire assessing the lifetime frequency, duration, and severity of NSSI behaviors across a range of methods. A total score equivalent to the number of lifetime NSSI episodes across methods was computed for each participant (referred to as NSSI Frequency). Additionally, a total score equivalent to the number of methods each participant endorsed utilizing at least once was computed (referred to as NSSI Methods).
TABLE 1
Demographic Information for the Full Sample (n = 1317) and the Subsample of Attempters (n = 143) Full sample Sex Male Female Family Income $0–$10,000 $10,001–$25,000 $25,001–$50,000 $50,001–$75,000 $75,001–$100,00 $100,000+ Race/Ethnicity White African American Other
N (%) 274 (20.8%) 1038 (78.8%) 123 189 324 296 199 182
(9.3%) (14.4%) (24.6%) (22.5%) (15.1%) (13.8%)
727 (55.2%) 507 (38.5%) 75 (5.7%)
Attempters Sex Male Female Family Income $0–$10,000 $10,001–$25,000 $25,001–$50,000 $50,001–$75,000 $75,001–$100,00 $100,000+ Race/Ethnicity White African American Other
N (%) 28 (19.6%) 113 (79.0%) 20 26 37 29 14 16
(14.0%) (18.2%) (25.9%) (20.3%) (9.8%) (11.2%)
73 (60.4%) 42 (34.7%) 5 (4.1%)
168 Covariates. Covariates were chosen based on variables that have previously been proposed as suicide risk factors and which could represent alternative variables that might better account for suicidal behavior than would our proposed independent variable and moderator. The Depression Anxiety Stress Scale (DASS; Lovibond & Lovibond, 1995) is a 21-item self-report questionnaire assessing severity of depression, anxiety, and stress symptoms over the course of the previous 7 days. The Depression subscale was utilized as a covariate in the current study and Cronbach’s alpha for this subscale in the present sample was .88. The Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) is a 36-item self-report measure that assesses individuals’ general capacity for identifying, understanding, accepting, and adaptively responding to negative affective states. The scale yields six subscale scores as well as a total score indicative of overall emotion dysregulation. Items are scored on a Likert-type scale ranging from 1 (Almost never [0–10%]) to 5 (Almost always [91– 100%]), with higher scores indicating greater emotion dysregulation. Total score was utilized as a covariate, and the alpha coefficient in this sample was .94. The Interpersonal Needs Questionnaire-15 (INQ-15; Van Orden et al., 2008; Van Orden, Witte, Cukrowicz, & Joiner, 2012) is a questionnaire that was developed to measure thwarted belongingness (nine items) and perceived burdensomeness (six items). Items are rated on a scale from 1 (Not at all true for me) to 7 (Very true for me). The alpha coefficients for burdensomeness and belongingness in this sample were .94 and .91, respectively. The Acquired Capability for Suicide Scale–Fear of Death Scale (ACSS-FDS; Ribeiro et al., 2014) is a 7-item version of the original 20-item ACSS (Bender, Gordon, Bresin, & Joiner, 2011) used to measure an individual’s fearlessness about lethal selfinjury. The 7-item version was developed through factor analysis and has been utilized
HOW MANY TIMES/WAYS previously in undergraduate samples (Witte, Gordon, Smith, & Van Orden, 2012). Ratings for each item ranged from 1 (Not at all like me) to 5 (very much like me). The alpha coefficient in this sample was .76. Outcome. The Measure of Episodic Planning of Suicide (MEPOS; Anestis, Pennings, & Williams, 2014) is a brief selfreport questionnaire based closely on the Lifetime Suicide Attempt Self-Injury Interview (L-SASI; Linehan & Comtois, 1996). First, participants are asked, “Throughout the course of your life, have you ever intentionally harmed yourself with at least some intent to cause your own death?” Participants who indicate that they have done so at least once are then asked a series of follow-up questions. Items assess total number of past attempts (Over the course of your entire life, how many times have you intentionally harmed yourself with at least some intention of causing your own death?), intent to die for past attempts (Of those events, how many times were you certain you wanted to die? Of those events, how many times were you uncertain if you truly wanted to die?), level of medical attention received in response to past attempts (Of those events, how many times did your behavior require medical attention [e.g., visit to hospital or other medical facility]?), method used for the most recent attempt (What method did you use in your most recent suicide attempt [e.g., firearms, prescription medication]?), level of intent for the most recent attempt (During your most recent attempt, to what extent did you want to die?), and level of planning for most recent attempt (Prior to your most recent attempt, how much had you thought about trying to die by suicide using the method you ultimately chose?). For items asking the number of times an event had occurred, participants entered a number (software prevented participants from entering text). In this project, participants were included from the larger sample if they endorsed at least one prior incident of suicidal behavior with clear or ambivalent intent to die. The number of such instances (total attempts with clear intent plus total attempts with ambivalent intent) was then computed for each participant and served as the outcome variable.
ANESTIS
ET AL.
Data Analytic Procedure. Following transformations (described below), we conducted a hierarchical multiple regression analysis to examine the potential moderating effect of NSSI Methods on the relationship between NSSI Frequency and lifetime number of suicide attempts. Both the predictor (NSSI Frequency) and the moderator (NSSI Methods) were grand-mean centered to facilitate the interpretation of a significant two-way interaction. In step one of the equation, the covariates and the main effects of the predictor and moderator were entered. In step two, the interaction of the predictor and moderator were entered. Lifetime number of suicide attempts (with ambivalent or clear intent to die) served as the outcome variable. Simple slopes analyses were used to test the nature and direction of the two-way interaction.
RESULTS
Descriptive data and intercorrelations for the variables utilized in the primary analyses are provided in Table 2.
Variable Transformations NSSI Frequency (Full sample skew = 29.03; Full sample kurtosis = 933.95; Attempter skew = 7.12; Attempter kurtosis = 56.78), NSSI Methods (Full sample skew = 2.94; Full sample kurtosis = 11.00; Attempter skew = 1.04; Attempter kurtosis = 1.14), and suicide attempts (Full sample skew = 8.36; Full sample kurtosis = 102.87; Attempter skew = 3.89; Attempter kurtosis = 20.25) exhibited significant skew and/or kurtosis. In an effort to approximate normality within distributions, we performed rank transformations using Blom’s formula, which resulted in acceptable levels (Kline, 2005) of skewness (
