Skip to main content
NCBI home page
VA Author Manuscripts logoLink to VA Author Manuscripts
. Author manuscript; available in PMC: 2022 Feb 1.
Published in final edited form as: Psychiatry Res. 2020 Dec 28;296:113672. doi: 10.1016/j.psychres.2020.113672

Diagnostic Correlates of Nonsuicidal Self-Injury Disorder among Veterans with Psychiatric Disorders

Tapan A Patel 1,2,*, Adam J D Mann 1,4, Shannon M Blakey 1,2, Frances M Aunon 1, Patrick S Calhoun 1,2,3,4, Jean C Beckham 1,2,3,4, Nathan A Kimbrel 1,2,3,4
PMCID: PMC7855413  NIHMSID: NIHMS1661432  PMID: 33383275

1. Introduction

Nonsuicidal self-injury (NSSI) refers to the intentional destruction of one’s own body tissue without suicidal intent (e.g., cutting, burning, or hitting self; American Psychiatric Association [APA], 2013; Nock, 2009). NSSI represents a serious public health issue, with 4–6% of the general population estimated to have engaged in NSSI in their lifetime (Briere & Gil, 1998; Klonsky, 2011). In the most recent version of the Diagnostic and Statistical Manual (DSM-5; APA, 2013), NSSI disorder (NSSID) was introduced as a condition for further study. To be diagnosed with NSSID, NSSI behavior must occur on five or more occasions on five different days in a given year, be for the purpose of emotion regulation or the resolution of one’s interpersonal difficulties and cause clinically significant distress and/or impairment (APA, 2013). The long-term consequences of NSSID and NSSI behavior are sobering (Glenn & Klonsky, 2013). A recent meta-analysis found that NSSI was the strongest overall prospective predictor of suicide attempts, evidencing a larger effect size on risk for future suicide attempts than history of prior suicide attempts, suicide risk screening instruments, personality disorders, and history of psychiatric hospitalization (Franklin et al., 2017).

NSSI behaviors are prominent among people with borderline personality disorder (Gratz et al., 2015), and self-injurious thoughts and behaviors (including NSSI as well as recurrent suicidal behavior, gestures, and threats) are a symptoms of borderline personality disorder in DSM-5 (APA, 2013). As a result, many empirical studies of NSSI behavior have utilized samples of adolescent girls and women with borderline personality disorder. Moreover, a meta-analysis of risk factors for NSSI identified cluster B personality disorders (including borderline personality disorder; weighted OR = 5.93, p < .01), depression (weighted OR = 1.98, p < .01), and eating disorder pathology (weighted OR = 1.81, p < .01) as the strongest overall diagnostic predictors of NSSI (Fox et al., 2015). NSSI has, however, been associated with a much wider array of psychopathology (Bentley et al, 2015; Briere & Gil, 1998), including anxiety disorders (Bentley et al., 2015), substance use disorders (Gratz et al., 2015), and posttraumatic stress disorder (PTSD; Bentley et al., 2015; Briere & Gil, 1998; Glenn & Klonsky, 2013; Gratz et al., 2015). For example, Bentley and colleagues (2015) conducted a meta-analysis of the association between emotional disorders (i.e., mood and anxiety disorders) and NSSI and found that, broadly speaking, individuals with an emotional disorder were significantly more likely than those without an emotional disorder to engage in NSSI, OR = 1.75, p < .001. Indeed, with the exception of bipolar disorder, OR = 1.05, p = .849, and social anxiety disorder, OR = 1.44, p = .086, all other mood and anxiety disorders examined in this analysis were associated with significantly increased risk for NSSI. The emotional disorders with the largest pooled odds ratios in relation to NSSI included in this meta-analysis were panic disorder, OR = 2.67, p < .001, PTSD, OR = 2.06, p < .001, generalized anxiety disorder, OR = 1.94, p < .001, obsessive-compulsive disorder (OCD), OR = 1.94, p = 0.04, and depression, OR = 1.90, p < .001.

The vast majority of studies in this area of literature have examined the relationship between psychiatric disorders and NSSI behaviors rather than NSSID per se; however, growing research suggests that NSSID is also frequently comorbid with other psychiatric disorders. For example, Kiekens and colleagues (2018) reported that undergraduate students meeting NSSID criteria were more likely to have multiple psychiatric disorders, particularly major depressive disorder and alcohol dependence, compared to students without NSSID. Similarly, Gratz and colleagues (2015) found that adults with NSSID were more likely to have co-occurring borderline personality disorder and depression compared to their counterparts without NSSID. Taken together, prior research suggests that individuals with NSSID are more likely to meet diagnostic criteria for multiple psychiatric disorders compared with individuals without NSSID. In addition, as with NSSI behaviors, borderline personality disorder and depression are among the most consistent diagnostic predictors of NSSID. Thus, it is perhaps not surprising that much of the research on NSSI and NSSID to date has focused on individuals with these disorders.

At the same time, there has also been growing recognition in recent years of the relevance of NSSI to a variety of previously understudied populations, particularly male veterans with PTSD (e.g., Kimbrel et al., 2017, 2018). As noted by Kimbrel et al. (2017), this oversight is partly due to the historical viewpoint that NSSI was much more common among women than men. Though some studies have found that NSSI prevalence rates may differ by sex, these studies primarily observe these findings in clinical samples (Bresin & Schoenloeber, 2015). In contrast, population-based studies of NSSI have found rates of NSSI do not differ between males and females (Briere & Gil, 1998; Klonsky et al, 2014). Moreover, high rates of both PTSD (Fulton et al., 2015) and suicide (Villatte et al., 2015) among veterans, coupled with research linking NSSI to PTSD (Sami & Hallaq, 2018) and suicide (Kiekens et al., 2018) have prompted investigators to study NSSI among veterans. For instance, Kimbrel and colleagues (2018) recently reported that 82% of veterans seeking treatment for PTSD (96% of whom were men) reported a lifetime history of engaging in NSSI. Furthermore, nearly two in three (64%) reported that they had engaged in one or more forms of NSSI during the past two weeks. Beyond PTSD, veterans also have higher rates of other psychiatric disorders (Trivedi et al., 2015; Williamson et al., 2018), which is likely due in part to the broad effects of combat exposure on risk for psychiatric symptomatology (Kimbrel et al., 2015). For example, Kimbrel and colleagues (2015) found that nearly three-fourths (73.6%) of Iraq/Afghanistan veterans screened positive for one or more psychiatric conditions, half (51.2%) screened positive for three or more, and one in three (34.9% or 45/129) screened positive for five or more psychiatric conditions on a well-validated psychiatric screening measure. Similarly, Fiedler and colleagues (2006) found that veterans deployed in the Gulf War had rates of depression and generalized anxiety disorder that were approximately double those of men in the general population.

1.1. Study Objective and Hypotheses

Given the dearth of research on NSSID in general to date, particularly among veterans, the objective of the present research was to identify diagnostic predictors of NSSID in a large, well-characterized sample of veterans. Based on prior research (e.g., Gratz et al., 2015; Kiekens et al., 2018), we hypothesized that: (H1) veterans with lifetime NSSID would meet criteria for a significantly greater number of lifetime psychiatric disorders (excluding NSSID) compared with veteran participants diagnosed with other psychiatric disorders (i.e., psychiatric controls). While prior research has indicated that individuals with NSSID are more likely than individuals without NSSID to have additional psychiatric disorders, this work has been limited by the fact that prior studies have not examined this question among veteran and only one prior study (Gratz et al., 2015) to our knowledge has utilized psychiatric controls for comparison. In addition, based on prior meta-analyses of diagnostic predictors of NSSI behaviors in civilians (e.g., Bentley et al., 2016; Fox et al., 2015), we further hypothesized that: (H2) borderline personality disorder, depression, PTSD, eating disorders, substance use disorders, and anxiety disorders would all be associated with NSSID at the bivariate level. Finally, based on the magnitude and consistency of their associations with NSSI and NSSID in prior research (e.g., Bentley et al., 2016; Fox et al., 2015; Gratz et al., 2015; Kiekens et al., 2018), we hypothesized that: (H3) borderline personality disorder and depression would have the strongest bivariate associations with NSSID (Bentley et al., 2016; Fox et al., 2015; Kimbrel et al., 2018) and would remain unique predictors of NSSID among veterans with psychiatric disorders in the multivariable logistic regression model.

2. Method

2.1. Participants and Procedures

Study participants included 124 veterans who consented to participate in a VA-sponsored study (#I01CX001486) aimed at identifying the functional consequences of NSSID in veterans relative to other psychiatric conditions. The study received approval from the Durham Veterans Affairs Health Care System (VAHCS) Institutional Review Board prior to data collection. Participants were primarily recruited through letters and calls to potential participants who: (1) had previously sought care for PTSD at the Durham VAHCS; (2) had previously agreed to have their names included in one or more research recruitment databases; or (3) were referred to the study by clinicians at the Durham VAHCS. All participants were initially screened by phone.

Note that veterans with PTSD were specifically targeted because: (1) NSSI is substantially underreported among veterans (Kimbrel et al., 2017); and (2) prior research suggests that veterans seeking treatment for PTSD have particularly high rates of NSSI (e.g., Kimbrel et al., 2018). Participants that appeared to meet basic eligibility criteria were then invited for an in-person screening to determine final study eligibility. To be eligible to participate in the study, participants had to have served in the U.S. military, be over the age of 18, be willing and able to complete the study and procedures, and had to have one or more current psychiatric disorders; however, participants with diagnoses of bipolar disorder, schizophrenia, and/or schizoaffective disorder were excluded from the study, as functional disability in relation to NSSID was a primary outcome and there was concern that inclusion of participants with severe mental illness in the study could potentially confound the association between NSSID and functioning, given the robust association between severe mental illness and functional disability.

Veterans with NSSID and women veterans were oversampled through the screening process to ensure sufficient representation in the statistical analyses. As a result, nearly half of the sample (n = 59; 47.6%) met full diagnostic criteria for a lifetime diagnosis of NSSID, and a third of the sample (n = 41; 33.1%) met criteria for current diagnosis of NSSID. The rest of the sample (n = 65; 52.4%) comprised veterans who met criteria for one or more lifetime psychiatric disorders but did not meet full criteria for lifetime NSSID. Note that 35.4% (n = 23) of the veterans who did not meet lifetime criteria for NSSID did report a history of lifetime NSSI behavior; however, their NSSI symptoms were insufficient to meet full criteria for lifetime NSSID. See Table 1 for a list of NSSI behaviors endorsed by the n = 89 participants with lifetime NSSI.

Table 1.

Frequency of NSSI Behaviors Among Veterans with Lifetime History of NSSI (n = 82)

NSSI Behavior n (%) Lifetime NSSI n ( %) Lifetime NSSI Disorder X2 φ p
Cutting 5.48 0.26 .019
 No 23 (100.0%) 47 (79.7%)
 Yes 0 (0.0%) 12 (20.3%)
Burning with Cigarette 2.08 0.16 .150
 No 23 (100.0%) 54 (91.5%)
 Yes 0 (0.0%) 5 (8.5%)
Burning with Lighter or Match 1.64 0.14 .200
 No 23 (100.0%) 55 (93.2%)
 Yes 0 (0.0%) 4 (6.8%)
Carving words 0.39 0.07 .530
 No 23 (100.0%) 58 (98.3%)
 Yes 0 (0.0%) 1 (1.7%)
Carving pictures or designs 1.67 0.14 .196
 No 23 (100.0%) 55 (93.2%)
 Yes 0 (0.0%) 4 (6.8%)
Scratching 4.80 0.57 .028
 No 21 (91.3%) 40 (67.8%)
 Yes 2 (8.7%) 19 (32.2%)
Biting 2.65 0.18 .104
 No 20 (87.0%) 41 (69.5%)
 Yes 3 (13.0%) 18 (30.5%)
Rubbing sandpaper 0.04 0.02 .836
 No 22 (95.7%) 57 (96.6%)
 Yes 1 (4.3%) 2 (3.4%)
Sticking Sharp Objects 4.68 0.24 .031
 No 22 (95.7%) 44 (74.6%)
 Yes 1 (4.3%) 15 (25.4%)
Rubbing Glass 0.40 0.07 .530
 No 23 (100.0%) 58 (98.3%)
 Yes 0 (0.0%) 1 (1.7%)
Banging Head 4.80 0.57 .028
 No 21 (91.3%) 40 (67.8%)
 Yes 2 (8.7%) 19 (32.2%)
Punching Self 3.57 0.21 .059
 No 20 (87.0%) 39 (66.1%)
 Yes 3 (13.0%) 20 (33.9%)
Preventing Wounds from Healing 1.97 0.15 .160
 No 21 (91.3%) 46 (78.0%)
 Yes 2 (8.7%) 13 (22.0%)
Punching Walls 6.41 0.28 .011
 No 14 (60.9%) 18 (30.5%)
 Yes 9 (39.1%) 41 (69.5%)
Open in a new tab

Note. NSSI = Nonsuicidal self-injury; Degrees of freedom for all tests = 1; φ = Phi effect size coefficient, where 0.1, 0.3, and 0.5 signify effects of small, medium, and large magnitude, respectively.

Approximately one-fourth of the sample identified as women (n = 32; 25.8%). With respect to race, 51.6% (n = 64) veterans identified as Black, 41.9% (n=52) identified as White, 4.0% (n = 5) identified as Other or else declined to answer, 1.6% (n = 2) identified as more than one race, and 0.8% (n = 1) of the sample identified as Asian. With respect to ethnicity, two veterans (1.6%) identified as Hispanic. The mean age of the sample was 48.7 years (SD = 13.0; range: 23–77).

2.2. Measures

The Structured Clinician Interview for DSM-5 (SCID-5; First et al., 2015) was used to assess mood disorders, anxiety disorders, and substance-use disorders. It has demonstrated excellent overall internal consistency and test-retest reliability in previous work (Shankman et al., 2018). Master’s level clinicians administered the SCID-5 under the supervision of licensed clinical psychologists. Reliability among interviewers for SCID-based diagnoses was excellent (Fleiss’ kappa = 0.92 for lifetime psychiatric disorders on fidelity training videos). The Clinician-Administered Nonsuicidal Self-injury Disorder Index (CANDI; Gratz et al., 2015) was used to diagnose NSSID. The CANDI is a diagnostic interview for NSSID that has demonstrated good interrater reliability (ᴋ = .83) and adequate internal consistency (α = .71) in prior research (Gratz et al., 2015). The Deliberate Self-Harm Inventory (DSHI), which lists various NSSI behaviors (e. g., cutting, burning, carving in skin, scratching, biting, acid or cleaner burns, sticking sharp objects, banging head, and punching walls or objects) and has the participant rate if they have engaged in the behavior and the frequency in the past year, was used as part of the CANDI interview. Master’s level clinicians administered the CANDI under the supervision of licensed clinical psychologists. Because NSSID was the focus of the present study and is currently listed as condition for further study in DSM-5, each CANDI was reviewed and discussed in diagnostic review groups led by a licensed clinical psychologist until diagnostic consensus was reached.

2.3. Data Analysis Plan

All analyses were conducted with SPSS Version 25. A total number of psychiatric disorders sum variable was created by summing all twenty diagnostic predictors considered in the analyses (see Table 2). Note that NSSID was excluded from the calculation of this score. Chi-square tests and t-tests were then used to evaluate bivariate associations between NSSID, demographic characteristics, and psychiatric disorders. Logistic regression was used to identify psychiatric diagnostic predictors of NSSID in veterans.

Table 2.

Bivariate Associations between Demographic Variables, Psychiatric Diagnoses, and Lifetime Nonsuicidal Self-Injury Disorder among Veterans with Psychiatric Disorders (N=124).

Variable % Lifetime NSSI Disorder (n=59) % Without Lifetime NSSI Disorder (n=65) Test Statistic p-value
Age t(122) = 2.35 0.02
Biological Sex at Birth X2(1) = 1.30 0.25
Male Sex at Birth 69.5% 78.5%
Female Sex at Birth 30.5% 21.5%
Race X2(4) = 1.67 0.80
Black 52.5% 50.8%
White 39.0% 44.6%
Other/Unknown 5.1% 3.1%
More than One Race 1.7% 1.5%
Asian 1.7% 0.0%
Ethnicity X2(1) = 2.24 0.14
Non-Hispanic 96.6% 100.0%
Hispanic 3.4% 0.0%
Psychiatric Disorders (Lifetime)
Borderline Personality Disorder 66.1% 15.4% X2(1) = 33.28 <0.001
Posttraumatic Stress Disorder 96.6% 89.2% X2(1) = 2.50 0.11
Major Depressive Disorder 94.9% 78.5% X2(1) = 7.08 0.008
Obsessive-Compulsive Disorder 50.8% 21.5% X2(1) = 11.61 0.001
Panic Disorder 27.1% 15.4% X2(1) = 2.57 0.11
Social Anxiety Disorder 16.9% 6.2% X2(1) = 3.60 0.06
Specific Phobia 3.4% 1.5% X2(1) = 0.45 0.50
Generalized Anxiety Disorder 27.1% 10.8% X2(1) = 5.47 0.02
Alcohol Use Disorder 71.2% 70.8% X2(1) = 0.003 0.96
Sedative Use Disorder 5.1% 1.5% X2(1) = 1.25 0.26
Cannabis Use Disorder 22.0% 9.2% X2(1) = 3.91 0.048
Stimulant Use Disorder 18.6% 16.9% X2(1) = 0.06 0.80
Opioid Use Disorder 10.2% 3.1% X2(1) = 2.58 0.11
Inhalant Use Disorder 0.0% 1.5% X2(1) = 0.92 0.34
Hallucinogen Use Disorder 0.0% 1.5% X2(1) = 0.92 0.34
Other Drug Use Disorder 1.7% 0.0% X2(1) = 1.11 0.29
Anorexia Nervosa 1.7% 0.0% X2(1) = 1.11 0.29
Bulimia Nervosa 3.4% 0.0% X2(1) = 2.24 0.14
Binge Eating Disorder 8.5% 6.2% X2(1) = 0.25 0.62
Any Eating Disorder 18.6% 7.7% X2(1) = 3.30 0.07
Any Drug Use Disorder 35.6% 26.2% X2(1) = 1.30 0.26
Open in a new tab

Note: Variables with bivariate associations significant at p < 0.05 are shown in bold.

3. Results

Sex, race, and ethnicity were unrelated to NSSID at the bivariate level (all p’s ≥ .14); however, a significant association was observed between age and NSSID, such that veterans who met lifetime criteria for NSSID (M = 45.9, SD = 12.4) were significantly younger than veterans who did not [M = 51.3, SD = 12.9, t(122) = 2.35, p = 0.02]. Notably, the total number of psychiatric disorders variable was normally distributed in the present sample (M = 4.4, SD = 1.8, range: 1–10, skewness = 0.22, kurtosis = 0.43). Moreover, veterans with NSSID (n = 59, M = 5.3, SD = 1.9) met criteria for a significantly greater number of disorders than veterans with other psychiatric disorders [n= 65, M = 3.5, SD = 1.2, t(122) = 2.35, p = 0.02]. As can be seen in Figure 1, a clear linear relationship between number of disorders and rate of NSSID was observed. Consistent with H1, a logistic regression model [Table 2: model 1; X2(1) = 35.99, p < 0.001, Nagelkerke pseudo R2 = 0.34, overall classification rate = 72.6%; f2 = 0.59, large effect size] found that each additional psychiatric diagnosis more than doubled veterans’ odds of meeting criteria for NSSID, OR = 2.13, p < 0.001.

Figure 1.

Figure 1.

Open in a new tab

Rates of NSSID as a Function of Total Number of Psychiatric Disorders (N=124).

Next, we examined bivariate associations between NSSID and psychiatric disorders. Consistent with H3, we observed statistically significant bivariate associations between lifetime NSSID and borderline personality disorder, X2(1) = 33.28, p < 0.001, and depression, X2(1) = 7.08, p = 0.008, such that veterans with these conditions were significantly more likely to also meet criteria for NSSID. We observed similar statistically significant associations between NSSID and OCD, X2(1) = 11.61, p = 0.001, generalized anxiety disorder, X2(1) = 5.47, p = 0.02, and cannabis use disorder, X2(1) = 3.91, p = 0.048, such that veterans with these disorders were also more likely to meet diagnostic criteria for NSSID (Figure 2). However, in contrast with H2, we did not observe significant bivariate associations between NSSID and PTSD, panic disorder, social anxiety disorder, specific phobia, alcohol use disorder, sedative use disorder, stimulant use disorder, opioid use disorder, inhalant use disorder, hallucinogen use disorder, other use disorder, anorexia nervosa, bulimia nervosa, binge eating disorder, and other eating disorder (all p’s > 0.05), though it should be noted that most of the drug use and eating disorders had too few cases to analyze individually in a meaningful way. Accordingly, we also constructed a composite variable for any drug use disorder and for any eating disorder; however, as can be seen in Table 2, these composite variables were also unrelated to NSSID.

Figure 2.

Figure 2.

Open in a new tab

Rates of NSSID by Diagnosis among Veterans with Psychiatric Disorders (N=124).

The demographic variables and psychiatric diagnoses observed to have significant bivariate associations with NSSID at α =0.05 (i.e., age, borderline personality disorder, depression, OCD, generalized anxiety disorder, and cannabis use disorder) were subsequently entered into a logistic regression model (model 2; simultaneous entry) in order to identify variables that uniquely contributed to the prediction of NSSID among veterans with psychiatric disorders. As can be seen in Table 3, the overall model was statistically significant, X2(6) = 50.88, p < 0.001, Nagelkerke pseudo R2 = 0.45, overall correct classification rate = 75.8%, f2 = 0.82, large effect size; however, among the six variables considered, only borderline personality disorder, adjusted odds ratio (AOR) = 7.67, p < 0.001, and OCD, AOR = 3.23, p = 0.02, continued to have statistically significant associations with NSSID in this model. Accordingly, a third logistic regression model was conducted in which backward stepwise selection was used to systematically eliminate non-significant (i.e., p ≥ 0.05) predictors from the model. As can be seen in Table 3, age was removed from the model in step 2, generalized anxiety disorder was removed in step 3, cannabis use disorder was removed in step 4, and depression was removed in step 5, leaving only borderline personality disorder, AOR = 10.1, p < 0.001, and OCD, AOR = 3.4, p = 0.008, in the final model. The final, more parsimonious model (i.e., step 5) resulting from the backwards stepwise selection procedure remained statistically significant, X2(2) = 42.43, p < 0.001, Nagelkerke pseudo R2 = 0.39, f2 = 0.64, large effect size. Moreover, the overall classification rate (75.8%) for this more parsimonious model containing only borderline personality disorder and OCD was identical to the overall classification rate for model 2 (i.e., 75.8%; simultaneous entry model) and higher than that of model 1 (72.6%; total number of psychiatric disorders model), suggesting that borderline personality disorder and obsessive-compulsive are likely responsible for much of the robust association observed between total number of psychiatric disorders and lifetime NSSID (Figure 1)

Table 3.

Summary of Logistic Regression Models Predicting Lifetime NSSI Disorder (N=124).

Model 1: Total number of psychiatric disorders as a predictor of lifetime NSSI Disorder (Nagelkerke R2 = 0.37)
Variable Name B SE OR 95% CI p-value
Total Number of Psychiatric Disorders 0.74 0.15 2.1 1.56 – 2.84 < 0.001
Age −0.03 0.02 0.97 0.94 – 1.00 0.048
Constant −1.68 1.04 0.19 0.11
Model 2: All variables with significant bivariate associations with lifetime NSSI Disorder (Nagelkerke R2 = 0.45)
Variable Name B SE OR 95% CI p-value
Borderline Personality Disorder 2.04 0.48 7.67 2.98 – 19.7 < 0.001
Obsessive-Compulsive Disorder 1.17 0.47 3.23 1.28 – 8.16 0.01
Major Depressive Disorder 1.32 0.75 3.74 0.87 – 16.1 0.08
Cannabis Use Disorder 0.89 0.67 2.43 0.66 – 8.96 0.18
Generalized Anxiety Disorder 0.75 0.62 2.12 0.63 – 7.09 0.22
Age −0.02 0.02 0.98 0.95 – 1.02 0.28
Constant −1.75 1.21 0.17 0.15
Model 3: Backward stepwise selection model (Step 5 Nagelkerke Pseudo R2 = 0.39)
Variable Name B SE OR 95% CI p-value
Step 1 Borderline Personality Disorder 2.04 0.48 7.67 2.98 – 19.7 < 0.001
Obsessive-Compulsive Disorder 1.17 0.47 3.23 1.28 – 8.16 0.01
Major Depressive Disorder 1.32 0.75 3.74 0.87 – 16.1 0.08
Cannabis Use Disorder 0.89 0.67 2.43 0.66 – 8.96 0.18
Generalized Anxiety Disorder 0.75 0.62 2.12 0.63 – 7.09 0.22
Age −0.02 0.02 0.98 0.95 – 1.02 0.28
Constant −1.75 1.21 0.17 0.15
Step 2 Borderline Personality Disorder 2.14 0.47 8.50 3.36 – 21.5 < 0.001
Obsessive-Compulsive Disorder 1.16 0.48 3.18 1.25 – 8.06 0.02
Major Depressive Disorder 1.39 0.75 4.03 0.92 – 17.6 0.06
Cannabis Use Disorder 0.85 0.65 2.34 0.66 – 8.30 0.19
Generalized Anxiety Disorder 0.77 0.62 2.17 0.65 – 7.28 0.21
Constant −2.81 0.78 0.06 < 0.001
Step 3 Borderline Personality Disorder 2.24 0.47 9.35 3.73 – 23.4 < 0.001
Obsessive-Compulsive Disorder 1.2 0.47 3.31 1.31 – 8.33 0.01
Major Depressive Disorder 1.24 0.72 3.45 0.84 – 14.2 0.09
Cannabis Use Disorder 0.95 0.64 2.59 0.73 – 9.14 0.14
Constant −2.6 0.73 0.08 < 0.001
Step 4 Borderline Personality Disorder 2.25 0.46 9.49 3.83 – 23.5 < 0.001
Obsessive-Compulsive Disorder 1.2 0.47 3.32 1.33 – 8.29 0.01
Major Depressive Disorder 1.24 0.72 3.45 0.84 – 14.2 0.09
Constant −2.47 0.72 0.09 < 0.001
Step 5 Borderline Personality Disorder 2.32 0.46 10.1 4.15 – 24.7 < 0.001
Obsessive-Compulsive Disorder 1.22 0.46 3.4 1.38 – 8.31 0.008
Constant −1.42 0.32 0.24 < 0.001
Open in a new tab

4. Discussion

Though previous studies have explored the link between psychiatric disorders and NSSI behaviors broadly (e.g., Bentley et al., 2016; Fox et al., 205), the vast majority of research in this area has focused on civilian populations and NSSI behaviors (as opposed to NSSID). To our knowledge, the present research represents the first and only study to examine the association between NSSID and a broad array of psychiatric disorders in a sample of veterans.

Consistent with our first hypothesis, we found that veterans with NSSID met criteria for a significantly greater number of disorders than veterans with other psychiatric disorders. Indeed, on average, we found that veterans with NSSID met full diagnostic criteria for five psychiatric disorders over the course of their lifetimes. Remarkably, 100% of the 17 veterans who met criteria for seven or more psychiatric disorders in this study also met criteria for NSSID (Figure 1). Moreover, each additional disorder more than doubled participants’ odds of meeting criteria for NSSID.

Our second hypothesis was only partially supported. As expected, we did observe statistically significant bivariate associations between NSSID and depression, OCD, generalized anxiety disorder, and cannabis use disorder, such that veterans with these disorders were also more likely to meet diagnostic criteria for NSSID; however, we did not observe the expected association between NSSID and PTSD. This finding is likely due to the large number of veterans in the sample with PTSD (n = 115). NSSID was also unrelated to the majority of anxiety disorders, substance use disorders, and eating disorders considered. Given the robust associations identified between many of these disorders and NSSID in civilians, we were surprised that so few of these conditions were associated with NSSID. In some cases (e.g., specific drug use disorders and eating disorders), there were too few cases to meaningfully examine these associations; however, even when we grouped drug use disorders and eating disorders together, we failed to find significant associations between these classes of disorders and NSSID in this sample of veterans (all p’s > 0.05).

Finally, consistent with our third hypothesis, we found that borderline personality disorder was the psychiatric diagnosis most strongly associated with NSSID at both the bivariate and multivariate level. Depression was also associated with NSSID at the bivariate level; however, it was not selected for inclusion in the final step of the stepwise regression model. Instead, OCD was selected for inclusion in the final regression model along with borderline personality disorder. Notably, this final model had a similar overall classification rate as the simultaneous regression model that included all six of the variables that had significant bivariate associations with NSSID (i.e., model 2) and a better classification rate than the model that utilized total number of disorders (i.e. model 1), which suggests that borderline personality disorder and OCD were largely responsible for the robust association observed between number of psychiatric disorders and lifetime NSSID observed in the present study (Figure 1).

While unexpected, the strong association between OCD and NSSID observed in the present study is in many ways consistent with Bentley and colleagues’ (2015) meta-analysis of the relationship between NSSI and emotional disorders. For instance, of the eight different emotional disorders considered in this meta-analysis, the least studied disorder in relation to NSSI was OCD, which had only nine effect sizes in the literature at the time of the meta-analysis. However, despite the relatively limited amount of attention given to the association between OCD and NSSI behaviors, Bentley et al. (2015) found that OCD had the third largest meta-analytic association with NSSI, OR = 1.94, p = .036, ranking behind only panic disorder, OR = 2.67, p<.001, and PTSD, OR = 2.06, p < .001. OCD’s meta-analytic odds ratio was similar to the effect size of depression, OR = 1.90, p < .001, one of the top two diagnostic predictors of NSSI identified by Fox et al. (2015).

Strengths of the present study include our use of structured interviews to systematically diagnose a broad array of psychiatric disorders as well as an analytical approach that allowed us to examine unique contributions of each disorder to the prediction of NSSID; however, it is likely that these same strengths are at least partly responsible for our failure to find support for the hypothesis that depression would be a unique predictor of NSSID among veterans. As noted above, depression had a significant association with NSSID at the bivariate level but failed to attain statistical significance in the simultaneous logistic regression model. A follow-up analysis in which we excluded borderline personality disorder from a simultaneous logistic regression model similar to model 2 resulted in OCD, OR = 3.61, p = .003, depression, OR = 5.59, p = 0.02, and age, OR = 0.97, p = .038, all remaining statistically significant predictors of NSSID (data not shown; cannabis use disorder, OR = 2.62, p = .11, and generalized anxiety disorder, OR = 2.86, p = .06, remained non-significant predictors in this model). Thus, it is likely that the high rate of co-occurrence between borderline personality disorder, depression, and NSSID resulted in the model selecting borderline personality disorder over depression. For example, the rate of NSSID was much higher among individuals diagnosed with borderline personality disorder than it was among individuals diagnosed with depression (79.6% vs. 52.3%). Moreover, 93.9% of veterans with borderline personality disorder in the present study also met criteria for depression. In contrast, fewer than half of veterans with borderline personality disorder in the present study met criteria for OCD (46.9%). Thus, the robust association between NSSID and OCD in the present study likely represented unique variance in the prediction of NSSID, whereas the association between depression and NSSID was largely accounted for by borderline personality disorder.

4.1. Clinical Implications

There are a number of important clinical implications that emerge from the present study. First, the present study adds to the literature on NSSI in veterans by demonstrating for the first time that veterans with NSSID are likely to be highly complex and challenging patients to treat, as veterans with NSSID were found to meet full diagnostic criteria for five additional psychiatric disorders, on average. As has been noted previously (e.g., Kimbrel et al., 2018), NSSI and NSSID are frequently overlooked conditions in veterans, despite the fact that NSSI is likely the single strongest prospective predictor of suicide attempts identified to date (Franklin et al, 2017) and NSSID is associated with substantial distress and impairment (Gratz et al., 2015). Similarly, despite the high levels of distress and impairment associated with both borderline personality disorder (Gratz et al., 2015) and OCD (McIngvale et al., 2019), both of these conditions are likely to be underdiagnosed among veterans (Barrera et al., 2019; Cunningham et al., 2019). For example, Cunningham et al. (2019) recently observed that nearly one in four (23.5%) male veterans seeking treatment for PTSD screened positive for severe borderline personality features. This same study found that veterans with PTSD, NSSI, and severe borderline personality features were significantly more likely to report suicidal ideation compared with veterans with PTSD only (OR = 5.68).

Studies utilizing structured diagnostic interviews to estimate the 12-month prevalence of OCD among veterans suggest that the rate of OCD may be substantially higher than the 12-month estimates observed among civilians in community settings (e.g., 1.2%; Barrera et al., 2019; McIngvale et al., 2019). Moreover, Barrera and colleagues (2019) have argued that OCD is often underdiagnosed in the veteran population due to the complexity of the disorder, lack of knowledge by providers, and attribution of symptoms to PTSD, suggesting that the true prevalence of OCD may be much higher than the estimates that have been previously reported. It could also be argued that veterans may be especially prone to engage in repetitive behaviors meant to reduce distress and gain control over their environment due to military training learning histories and/or comorbid disorders associated with increased need for control and decreased tolerance for unpleasant emotions (Tuerk et al., 2011). Future research investigating the motivations for NSSI behaviors among veterans with OCD would help to understand, assess, and target NSSID in this population. For example, substantial research indicates that people with OCD have difficulty tolerating aversive internal experiences (e.g., unpleasant emotions, intrusive thoughts), which might result in performing compulsions and other repetitive behaviors to minimize or escape their unpleasant internal states (Robinson & Freeston, 2014). It is therefore possible that there may be a subtype of people with OCD who frequently turn to NSSI to regulate or better control their aversive internal experiences (Starcevic et al., 2011).

4.2. Study Limitations and Future Directions

The present research had several limitations which should be considered when interpreting these findings. First, this study was cross-sectional in nature, which precludes us from being able to determine if some disorders precede the occurrence of other disorders. Thus, we are unable to determine if borderline personality disorder and OCD typically precede the onset of NSSID or, if instead, NSSID typically precedes the onset of these disorders. Second, the sample was composed entirely of veterans with psychiatric disorders, the vast majority of whom had either PTSD or depression but excluded those with severe mental illness. Thus, the degree to which the present findings might generalize to other samples of veterans and civilians is not known. Third, the sample was relatively small and did not allow for more refined analytical approaches.

Despite these limitations, these findings have important implications for both future research and clinical practice. Future research could build on the present study’s limitations and focus on discovering why OCD and NSSID co-occur. It is important to understand the mechanism of how and why these two disorders are related. Furthermore, the findings of this study provide broader implications to the NSSID body of research. It will be important in future research in NSSID to assess for OCD in order to further explore whether OCD is a risk factor for engaging in NSSID. Additionally, in future research, it would be helpful to investigate the reason for this association and how it affects the severity of the individual diagnoses. Though the findings from the present study cannot be generalized to the general population, it does provide specific implications for veterans. The study results suggest that it would be useful in future research with veterans to assess for NSSI behaviors and NSSID to further identify risk factors and determine clinical implications. Mental health and medical practitioners need to be aware of the risk of veterans with OCD having co-occurring NSSID. Both phenomena can be debilitating and having both could prove to be difficult given there are currently no current evidence-based treatments for NSSI in the veteran population. When working with veterans, it will be important to assess for both NSSI and OCD to maximize treatment outcomes. The results suggest that the highest levels of psychiatric comorbidity are accompanied by NSSID. Developing and evaluating effective treatments for NSSID will be important in addressing the mental health needs of veterans who have the highest prevalence of multiple psychiatric disorders.

Role of Funding source

Funding: This research was supported by a Merit Award to Dr. Kimbrel from the Clinical Sciences Research and Development Service of the Department of Veterans Affairs’ Office of Research and Development (#I01CX001486). Dr. Blakey was supported by the VA Office of Academic Affiliations Advanced Fellowship in Mental Illness Research and Treatment. Dr. Beckham was funded by a Senior Research Career Scientist award from VA Clinical Sciences Research and Development (IK6BX00377).

References

  1. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: Author. [Google Scholar]
  2. Barrera TL, McIngvale E, Lindsay JA, Walder AM, Kauth MR, Smith TL, Van Kirk N, Teng EJ, & Stanley MA (2019). Obsessive-compulsive disorder in the Veterans Health Administration. Psychological Services, 16(4), 605–611 [DOI] [PubMed] [Google Scholar]
  3. Bentley KH, Cassiello-Robbins CF, Vittorio L, Sauer-Zavala S, Barlow DH, 2015. The association between nonsuicidal self-injury and the emotional disorders: A meta-analytic review. Clin Psychol Rev, 37, 72–88. [DOI] [PubMed] [Google Scholar]
  4. Bresin K, & Schoenleber M (2015). Gender differences in the prevalence of nonsuicidal self-injury: A meta-analysis. Clinical psychology review, 38, 55–64. 10.1016/j.cpr.2015.02.009 [DOI] [PubMed] [Google Scholar]
  5. Briere J, & Gil E (1998). Self-mutilation in clinical and general population samples: Prevalence, correlates, and functions. American Journal of Orthopsychiatry, 68(4), 609–620. [DOI] [PubMed] [Google Scholar]
  6. Cunningham KC, Grossmann JL, Seay KB, Dennis PA, Clancy CP, Hertzberg MA, Berlin K, Ruffin RA, Dedert EA, Gratz KL, Calhoun PS, Beckham JC, & Kimbrel NA (2019). Nonsuicidal self-injury and borderline personality features as risk factors for suicidal ideation among male veterans with posttraumatic stress disorder. Journal of Traumatic Stress, 32(1), 141–147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Fiedler N, Ozakinci G, Hallman W, Wartenberg D, Brewer NT, Barrett DH, & Kipen HM (2006). Military deployment to the Gulf War as a risk factor for psychiatric illness among US troops. The British Journal of Psychiatry : The Journal of Mental Science, 188, 453–459. [DOI] [PubMed] [Google Scholar]
  8. First MB, Williams JBW, Karg RS, Spitzer RL (2015). Structured Clinical Interview for DSM-5—Research Version. Arlington, VA: American Psychiatric Association [Google Scholar]
  9. Fox KR, Franklin JC, Ribeiro JD, Kleiman EM, Bentley KH, & Nock MK (2015). Meta-analysis of risk factors for nonsuicidal self-injury. Clinical psychology review, 42, 156–167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Franklin JC, Ribeiro JD, Fox KR, Bentley KH, Kleiman EM, Huang X, Musacchio KM, Jaroszewski AC, Chang BP, & Nock MK (2017). Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychological Bulletin, 143(2), 187–232 [DOI] [PubMed] [Google Scholar]
  11. Fulton J, Calhoun PS, Wagner HR, Schry AR, Hair LP, Feeling N, Elbogen E Beckham JC, 2015. The prevalence of PTSD in Operation Enduring Freedom/Operation Iraqi Freedom veterans: A meta-analysis. J Anxiety Disord, 31, 98–107. [DOI] [PubMed] [Google Scholar]
  12. Glenn CR, & Klonsky ED (2013). Nonsuicidal self-injury disorder: An empirical investigation in adolescent psychiatric patients. Journal of Clinical Child and Adolescent Psychology, 42(4), 496–507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gratz KL, Dixon-Gordon KL, Chapman AL, & Tull MT (2015). Diagnosis and characterization of DSM-5 nonsuicidal self-injury disorder using the Clinician-Administered Nonsuicidal Self-Injury Disorder Index. Assessment, 22(5), 527–539. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kiekens G, Hasking P, Claes L, Mortier P, Auerbach RP, Boyes M, … & Nock MK (2018). The DSM-5 nonsuicidal self-injury disorder among incoming college students: Prevalence and associations with 12-month mental disorders and suicidal thoughts and behaviors. Depression and anxiety, 35(7), 629–637. [DOI] [PubMed] [Google Scholar]
  15. Kimbrel NA, Calhoun PS, & Beckham JC (2017). Nonsuicidal self-injury in men: A serious problem that has been overlooked for too long. World Psychiatry, 16(1), 108–109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kimbrel NA, Gratz KL, Tull MT, Morissette SB, Meyer EC, DeBeer BB, … Beckham JC (2015). Kimbrel, N. A., Gratz, K. L., Tull, M. T., Morissette, S. B., Meyer, E. C., DeBeer, B. B., Silvia, P. J., Calhoun, P. C., & Beckham, J. C. (2015). Non-suicidal self-injury as a predictor of active and passive suicidal ideation among Iraq/Afghanistan war veterans. Psychiatry research, 227(2–3), 360–362. 10.1016/j.psychres.2015.03.026 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kimbrel NA, Johnson ME, Clancy C, Hertzberg M, Collie C, Van Voorhees EE, … Beckham JC (2014). Deliberate Self-Harm and Suicidal Ideation Among Male Iraq/Afghanistan-Era Veterans Seeking Treatment for PTSD. Journal of Traumatic Stress, 27(4), 474–477. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kimbrel NA, Thomas SP, Hicks TA, Hertzberg MA, Clancy CP, Elbogen EB, Meyer EC, DeBeer BB, Gross GM, Silvia PJ, Morissette SB, Gratz KL, Calhoun PS, Beckham JC, (2018). Wall/Object Punching: An important but under-recognized form of nonsuicidal self-injury. Suicide Life Threat Behav, 48(5), 501–511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Klonsky ED (2011). Non-suicidal self injury in United States adults: Prevalence, sociodemographics, topography and functions. Psychological Medicine, 41(9), 1981–1986. [DOI] [PubMed] [Google Scholar]
  20. Klonsky ED, Victor SE, & Saffer BY (2014). Nonsuicidal self-injury: What we know, and what we need to know. The Canadian Journal of Psychiatry / La Revue Canadienne de Psychiatrie, 59(11), 565–568 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. McIngvale E, Van Kirk N, Amspoker AB, Stanley MA, & Fletcher TL (2019). Prevalence and treatment of obsessive-compulsive disorder in veterans and active-duty service members: A systematic review. Journal of Cognitive Psychotherapy, 33(1), 11–22. [DOI] [PubMed] [Google Scholar]
  22. Nock MK (2009). Why do people hurt themselves? Current directions in psychological science, 18(2), 78–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Robinson LJ, & Freeston MH (2014). Emotion and internal experience in Obsessive Compulsive Disorder: Reviewing the role of alexithymia, anxiety sensitivity and distress tolerance. Clinical Psychology Review, 34(3), 256–271 [DOI] [PubMed] [Google Scholar]
  24. Sami H, & Hallaq E (2018). Nonsuicidal self-injury among adolescents and young adults with prolonged exposure to violence: The effect of post-traumatic stress symptoms. Psychiatry research, 270, 510–516. 10.1016/j.psychres.2018.10.028 [DOI] [PubMed] [Google Scholar]
  25. Starcevic V, Berle D, Brakoulias V, Sammut P, Moses K, Milicevic D, & Hannan A (2011). Functions of compulsions in obsessive--compulsive disorder. Australian & New Zealand Journal of Psychiatry, 45(6), 449–457. [DOI] [PubMed] [Google Scholar]
  26. Trivedi RB, Post EP, Sun H, Pomerantz A, Saxon AJ, Piette JD, Maynard C, Arnow B, Curtis I, Fihn SD, & Nelson K (2015). Prevalence, comorbidity, and prognosis of mental health among US veterans. American Journal of Public Health, 105(12), 2564–2569. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Tuerk PW, Grubaugh AL, Hamner MB, & Foa EB (2009). Diagnosis and treatment of PTSD-related compulsive checking behaviors in veterans of the Iraq war: The influence of military context on the expression of PTSD symptoms. The American Journal of Psychiatry, 166(7), 762–767. [DOI] [PubMed] [Google Scholar]
  28. Villatte JL, O’Connor SS, Leitner R, Kerbrat AH, Johnson LL, & Gutierrez PM (2015). Suicide Attempt Characteristics Among Veterans and Active-Duty Service Members Receiving Mental Health Services: A Pooled Data Analysis. Military Behavioral Health, 3(4), 316–327. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Williamson V, Stevelink SAM, Greenberg K, & Greenberg N (2018). Prevalence of mental health disorders in elderly US military veterans: A meta-analysis and systematic review. The American Journal of Geriatric Psychiatry, 26(5), 534–545. [DOI] [PubMed] [Google Scholar]

RESOURCES