Arrest History and Psychopathology Among Veterans at Risk for Suicide

Abstract

While preliminary evidence suggests an association between legal involvement and suicide risk among veterans, no research to date has explored the prevalence and/or correlates of legal involvement among veterans at high risk for suicide. The current study examined the relation of suicide attempt, suicidal ideation, and psychopathology to history of criminal arrest in a sample of 286 veterans at risk for suicide. Results indicated approximately half (47%) of at-risk veterans had a history of arrest. Inconsistent with hypotheses, arrest history was not associated with history of suicide attempt, current suicidal ideation, or severity of psychopathological symptoms. Arrest history was, however, associated with diagnoses of substance use disorder and antisocial personality disorder in this high-suicide risk sample. Further, likelihood of an antisocial personality disorder diagnosis was associated with higher frequency of past arrests. Taken together, results indicate that many veterans at risk for suicide have a history of arrest, and at-risk veterans with such history likely have a specific pattern of psychopathology, including antisocial personality traits and substance use. As such, legal status and history of justice involvement may be important considerations when assessing suicide risk and management of this high-risk population.

Rates of suicide within the United States have soared in recent years. Since the turn of the 21st century, national suicide rates have increased by approximately 33% at an average rate of 1.5% per year (Hedegaard, Curtin, & Warner, 2018; Stone et al., 2018). These rates are particularly notable for the nation’s military veterans. Although veterans account for only 8.5% of the population, veteran suicides account for approximately 18% of all suicide deaths in the United States (Office of Suicide Prevention in Veterans Health Administration, 2016). Mounting evidence suggests the psychosocial needs of veterans at risk for suicide are markedly complex and include legal involvement, homelessness, substance use, and psychopathology (Bohnert, Ilgen, Louzon, McCarthy, & Katz, 2017; Saxon et al., 2001; Tsai, Trevisan, Huang, & Pietrzak, 2018). As such, understanding these needs is likely key to suicide risk assessment and management of the nation’s veterans. The current study investigates a commonly overlooked factor associated with suicide risk—legal involvement (Fazel, Hayes, Bartellas, Clerici, & Trestman, 2016; Steele, Thrower, Noroian, & Saleh, 2018; Webb et al., 2011)—in a large sample of veterans at risk for suicide.

U.S. military veterans die by suicide at greater rates than the general population. An analysis of over 27,000 confirmed suicide deaths suggests veteran suicide rates to be approximately 58% greater than nonveteran suicide rates (Gibbons, Brown, & Hur, 2012). While some authors have suggested misclassification biases—such as misidentifying an individual as a veteran versus active-duty service member—may contribute to inflated estimates of veteran suicide (e.g., Carr, Hoge, Gardner, & Potter, 2004), research suggests any misclassification likely does not account for the large disparities in suicide rates between veteran and civilian populations (Huguet, Kaplan, & McFarland, 2014).

Legal Involvement as a Correlate of U.S. Veteran Suicide

Some evidence suggests veterans with a history of legal involvement may be at higher risk for suicide than veterans without such history. Epidemiological data reflect a modest, yet statistically significant, association between history of incarceration and history of suicide attempt in veterans at the national level (Edwards, Barnes, Govindarajulu, Geraci, & Tsai, 2020). Correspondingly, estimates suggest approximately 21% of currently incarcerated veterans have a history of suicide attempt (Saxon et al., 2001) compared to only 5% of veterans in the general population (Edwards et al., 2020). In a sample of over 15,000 veterans, recent incarceration was associated with a 4.45 times greater likelihood of attempting suicide and 1.75 times greater likelihood of dying by suicide (Barry et al., 2018). Historically, veterans have also been involved with the legal system at disproportionate rates; for example, throughout the 1980s, veterans accounted for approximately 20–25% of incarcerated persons (Bronson, Carson, Noonan, & Berzofsky, 2015). To date, the cause and directionality of this overrepresentation remains unclear (Timko et al., 2014). Nevertheless, if a disproportionate number of veterans have a history of legal involvement and legal involvement is associated with suicide, then veteran legal involvement may at least partially account for elevations in veteran suicide rates. Correspondingly, many authors have argued for a need to assess incarcerated veterans for potential suicide risk (e.g., Frisman & Griffin-Fennell, 2009; Wortzel, Binswanger, Anderson, & Adler, 2009).

Research on veteran legal involvement and suicide, however, has relied exclusively on either nationally representative or justice-involved samples. Given the low base rate of suicide attempts and deaths among these samples, the overall frequency of and risk for suicide among study participants has tended to be low. For example, the study of 15,000 veterans found less than 2% of recently incarcerated veterans attempted suicide within three years of release and less than 0.1% died by suicide during this time (compared to 0.4% and .05%, respectively, of veterans without recent legal involvement; Barry et al., 2018). Providing a more complete picture of the association between veteran legal involvement and suicide risk therefore necessitates investigations that specifically sample veterans at elevated risk for suicide. To date, however, no such research has been conducted.

Veteran Legal Involvement, Suicide, and Psychopathology

Various theories have been proposed to explain observed associations between legal involvement and suicide. For example, the interpersonal theory of suicide suggests a causal association wherein legal involvement elevates suicide risk by increasing perceived burdensomeness (e.g., on society because of criminal behavior) and/or decreasing social connectedness (e.g., through incarceration; Simlot, McFarland, & Lester, 2013). Consistent with this, veteran suicide risk tends to increase in the days and weeks immediately following arrest or imprisonment (Wortzel et al., 2009), and nearly 20% of persons who attempt suicide have been arrested in the year prior to their attempt (compared to only 3.5% of persons who do not attempt suicide; Cook, 2013).

By contrast, other theories suggest the association between legal involvement and suicide may be at least partially explained by shared risk factors, such as psychological stress and personality characteristics. Consistent with this latter suggestion, some forms of psychopathology—particularly those characterized by behavioral impulsivity, emotional agitation, and interpersonal disruption—appear similarly associated with both legal involvement and suicide. For example, estimates suggest up to 39% of justice-involved veterans and 19% of veterans who attempt and/or die by suicide meet criteria for posttraumatic stress disorder (PTSD; Blodgett et al., 2015; Bryan & Rudd, 2015; Ilgen et al., 2010). Among justice-involved veterans, a diagnosis of PTSD is also associated with higher rates of suicidal ideation and attempt (Saxon et al., 2001).

Rates of drug and alcohol use disorders also tend to be elevated among veterans who are justice involved (Blodgett et al., 2015; Kimbrel et al., 2014) and/or have a history of suicide attempt (Bachynski et al., 2012; Bohnert et al., 2017). Approximately 70% of veterans receiving Veterans Affairs (VA) substance use treatment services have a history of arrest, including 58% with a history of three or more arrests and 24% with a history of arrest for violent crime (Weaver, Trafton, Kimerling, Timko, & Moos, 2013). Among Veterans Health Administration–serviced veterans who die by suicide, approximately 17% have a documented diagnosis of substance use disorder (Bohnert et al., 2017).

Although preliminary, some evidence also suggests rates of personality disorders—particularly antisocial and borderline personality disorders—may be similarly elevated among justice-involved veterans and veterans at risk for suicide (Bachynski et al., 2012; Blodgett et al., 2015; Bryan & Rudd, 2015; Edwards et al., 2020; Tsai, Rosenheck, Kasprow, & McGuire, 2013a). A recent latent class analysis suggests veterans with a comorbid personality disorder–substance use disorder behavioral health pattern have the highest rates of suicide attempt and incarceration compared to other behavioral health patterns (Edwards et al., 2020). Similarly, in the general population, individuals with antisocial and borderline personality disorders account for only 2–6% of the general population (Samuels, 2011) yet over 30% of incarcerated persons (Black et al., 2007; Black, Gunter, Loveless, Allen, & Sieleni, 2010) and approximately 35% of persons who attempt suicide (Berk, Jeglic, Brown, Henriques, & Beck, 2007; Yilmaz, Kugu, Kavakci, & Dogan, 2018). Notably, diagnostic criteria for antisocial and borderline personality disorders specifically reference criminal and suicidal behavior, respectively (American Psychiatric Association, 2013), potentially inflating these estimates.

Current Study

Previous research provides evidence to suggest an association between legal involvement and suicide risk among veterans. However, studies that directly examine this association have tended to use either justice-involved or nationally representative samples, whereas no studies have approached this investigation by sampling veterans at risk for suicide. Such a methodological shift would provide greater representation of suicide risk within the study sample and thereby allow for novel insight into the complex needs of veterans at risk for suicide (as opposed to the needs of justice-involved veterans, per se).

To aid in addressing this gap in the literature, the current study examined the relationship of suicide attempt and suicidal ideation to history of criminal arrest in a sample of U.S. military veterans receiving care at James J. Peters VA Medical Center and identified as high risk for suicide. In line with previous suggestions that psychopathology may be a shared risk factor for both suicide and justice involvement (see Veteran Legal Involvement, Suicide, and Psychopathology), the study also examined relations of psychopathology to suicide attempt, suicidal ideation, and criminal arrest history in this high-risk sample. We hypothesized the following:

1. Compared to those without a history of criminal arrest, veterans with a history of at least one arrest would (a) be more likely to have a history of one or more suicide attempts; (b) experience more severe suicidal ideation; (c) be more likely to be diagnosed with PTSD, a substance use disorder, and antisocial and/or borderline personality disorders; and (d) experience more severe psychopathological symptoms.

2. Diagnosis of PTSD, substance use disorder, and antisocial and/or borderline personality disorders would each be associated with history of suicide attempts, suicidal ideation, and criminal arrest.

3. Frequency of arrests would be associated with (a) presence and higher frequency of past suicide attempts; (b) greater severity of suicidal ideation; (c) presence of PTSD, a substance use disorder, and antisocial and/or borderline personality disorders; and (d) greater psychopathological symptom severity.

Method

Participants and Procedure

Data were drawn from a larger sample of 310 U.S. military veterans identified as at risk for suicide by inpatient and outpatient providers at James J. Peters VA Medical Center between 2010 and 2015. Data from this sample have been previously reported by Goodman et al. (2016) and Hazlett et al. (2016); however, the current study is first to examine this sample through the lens of legal involvement.

All veterans were aged 18–57 and deemed “high risk” for suicide by inpatient and outpatient providers according to one or more of the following criteria: (a) recent suicide attempt or severe ideation resulting in psychiatric hospitalization or emergency room presentation within the past 3 months, (b) chronic suicidal ideation lasting more than 3 months, and (c) assignment to the “high-risk” suicide list maintained by the local VA Suicide Prevention Coordinator at James J. Peters VA Medical Center. Veterans were not eligible for inclusion if they (a) met DSM–IV diagnostic criteria for a schizophrenia-spectrum disorder (excluding schizophrenia-spectrum personality disorders) or (b) had a history of significant organic brain impairment. Demographically, 73% of participants were male, and 27% were female. Participant age ranged from 20 to 57 (M = 40.67, SD = 10.75), and most identified as Black/African American (43%), Hispanic (30%), or White (18%). Most participants were unmarried; 38% reported they were never married, 28% were divorced, 12% were separated, and 2% were widowed.

Participants completed study measures as part of a larger study on veteran suicidality (see Hazlett et al., 2016). All measures were administered by a clinical psychologist specifically trained in diagnostic assessments. Veterans provided written informed consent prior to participation, and this study was approved by the local VA institutional review board. Participants were provided monetary compensation for their time and travel in return for their participation. Of the 310 veterans in the original sample, 7 were diagnosed with a schizophrenia-spectrum disorder; therefore, their data were not suitable for inclusion. Arrest history was also unavailable for 17 veterans, resulting in a final sample size of 286.

Measures

To examine hypotheses, the current study assessed participants’ history of arrest, history of suicide attempt(s), suicidal ideation, psychiatric diagnosis, and psychopathological symptom severity. Assessment was completed using a combination of semistructured clinical interview and self-report measures.

History of arrest.

Participants’ history of arrest was assessed using the Addiction Severity Index (ASI; McLellan et al., 1992). The ASI is a semistructured clinical interview that assesses various potential problem areas commonly associated with substance abuse. For the purposes of the current study, frequency of prior arrests was drawn from the Legal Status subscale of the ASI. This subscale guides respondents through reporting past and current criminal justice involvement, including frequency of prior arrests defined according to 14 arrest types (e.g., “How many times in your life have you been arrested and charged with assault?”). From such information, arrests were categorized into four classes of offense: violent offense (e.g., assault), nonviolent offense (e.g., burglary), drug offense (e.g., drug possession), and technical violation offense (e.g., parole violation); arrests defined as “other arrest type” according to the ASI were not categorized according to offense class. Notably, although the ASI focuses on substance use, the Legal Status subscale assesses justice involvement both related and unrelated to substance use. The ASI has demonstrated strong reliability and validity across settings and populations (Leonhard, Mulvey, Gastfriend, & Shwartz, 2000). Previous research suggests also self-report to be a valid method of assessing arrest history in both psychiatric and general populations when self-report is compared to official records (Morris & Slocum, 2010; Nieves, Draine, & Solomon, 2000). See Table 1 for a summary of arrest types and corresponding classifications.

Table 1.

Veteran Arrest History

Arrest type	Number of participants reporting history	M (SD) arrest frequencya
Violent crimes	57	0.98 (1.80)
Weapons	14	0.16 (0.55)
Robbery	19	0.20 (0.55)
Assault	39	0.58 (1.45)
Arson	0	0 (0)
Rape	1	0.01 (0.09)
Homicide/manslaughter	3	0.03 (0.21)
Nonviolent crimes	45	0.56 (0.99)
Shoplifting/vandalism	22	0.27 (0.72)
Forgery	10	0.08 (0.30)
Burglary/larceny/B&E	22	0.21 (0.52)
Prostitution	0	0 (0)
Drug-related crimes	57	1.63 (3.43)
Drug charges	57	1.63 (3.43)
Technical violations	18	0.28 (0.89)
Parole/probation violation	16	0.23 (0.82)
Contempt of court	4	0.05 (0.30)
Other/unspecified	56	0.78 (1.38)

Note. B&E = breaking and entering.

^aAverage number of arrests among participants reporting any arrest history.

Suicide attempt and suicidal ideation.

Lifetime history of suicide attempt and suicidal ideation were assessed using the Columbia Suicide Severity Rating Scale (C-SSRS; Posner et al., 2008). The C-SSRS is a semistructured clinical interview that assesses various facets of suicidal behavior and ideation. For the purposes of the current study, the following measures from the C-SSRS were included in analyses: (a) presence or absence of any prior suicide attempt (including interrupted, aborted, and actual suicide attempts), (b) total frequency of prior suicide attempts; (c) presence or absence of any prior suicidal ideation, and (d) most severe prior suicidal ideation, rated on a scale of 1 (passive wish to be dead) to 5 (active suicidal ideation with intent and plan). Prior research suggests the C-SSRS demonstrates good concurrent and predictive validity in veteran samples (Katz, Barry, Cooper, Kasprow, & Hoff, 2020; Matarazzo et al., 2019).

The Beck Scale for Suicidal Ideation (BSS; Beck & Steer, 1991) was also used as a self-report measure of suicidal ideation. The BSS consists of 21 items rated on a scale of 0 to 2. Total scores range from 0 to 42 and reflect severity of suicidal ideation and are commonly considered a general indicator of suicide risk. The BSS has demonstrated strong psychometric properties across samples, including veterans (Gutierrez et al., 2019). In the current sample, α = .91.

Psychiatric diagnosis.

Participant psychopathology was determined using the Structured Clinical Interview for DSM–IV (SCID-IV; First, Spitzer, Gibbon, & Williams, 1996) and Structured Interview for DSM–IV Personality Disorders (SIDP-IV; Pfohl, Blum, & Zimmerman, 1997).¹ The SCID-IV and SIDP-IV are semistructured diagnostic interviews designed to assess current and lifetime diagnosis of DSM–IV psychiatric disorders. Weekly consensus and diagnostic meetings to confirm diagnoses were led by the interviewing clinical psychologist and a second clinical psychologist. Our team has achieved strong interrater reliability of κ = 0.81 for psychiatric diagnoses.

Psychopathological symptom severity.

Lastly, psychopathological symptom severity was assessed using the Beck Depression Inventory (BDI-II; Beck, Steer, & Brown, 1996), Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988), and Clinician-Administered PTSD Scale (CAPS; Blake et al., 1995). The BDI-II and BAI are 21-item, self-report measures of depressive and anxiety symptoms, respectively. Respondents complete each item using a 4-point Likert scale of 0 to 3; total scores range from 0 to 63 and reflect severity of symptoms. Previous research suggests the BDI-II and BAI to have strong psychometric properties in veteran samples (Palmer et al., 2014; Palmer, Happe, Paxson, Jurek, & Olson, 2016). In the current sample, both scales demonstrated excellent internal reliability, BDI-II (α = .92) and BAI (α = .93). The CAPS is a semistructured clinical interview assessing symptoms of PTSD. Although the CAPS yields scores of both current and lifetime PTSD symptom severity (possible range = 0–136), only the CAPS lifetime score (i.e., greatest severity of symptoms at any point in the lifetime) was included in analyses. This lifetime focus was consistent with the ASI and C-SSRS, which assessed lifetime arrest and suicidality, respectfully. The CAPS has demonstrated strong psychometric properties in veteran samples (Weathers et al., 2018).

Analysis Plan

To test Hypothesis A, univariate analyses of variance (for continuous variables) and Pearson’s chi-square tests of independence (for dichotomous variables) compared veterans with and without a history of arrest on measures of suicide attempt, suicidal ideation, psychiatric diagnosis, and psychopathological symptom severity. Follow-up post hoc comparisons using univariate analyses of variance (for continuous variables) and Pearson’s chi-square tests of independence (for dichotomous variables) were then used to further understand significant associations at the level of arrest type (i.e., comparing veterans with a history of arrest for violent crimes, nonviolent crimes, drug crimes, and technical offenses). To aid in accounting for multiple comparisons, a more conservative α = .01 was used to determine statistical significance of post hoc comparisons.

To test Hypothesis B, Pearson’s correlation analyses examined the extent to which likelihood of PTSD, substance use, antisocial personality, and borderline personality disorder diagnoses were associated with arrest frequency, suicide attempt, and suicidal ideation. Lastly, to test Hypothesis C, Pearson’s correlation analyses examined overall strength of association between arrest frequency and measures of suicide attempt, suicidal ideation, psychiatric diagnosis, and psychopathological symptom severity. To minimize the extent to which analyses associated with Hypothesis C reflected categorical differences in individuals with and without a history of arrest (as examined in Hypothesis A), these final analyses were limited to include only veterans reporting a history of arrest.

To minimize the potential biasing effects of outliers on analyses, individuals reporting a frequency of arrest or suicide attempt that was at least 3 SD greater than the mean were excluded from analyses. Within a normally distributed sample, less than 0.5% of cases lie more than 3 SD from the mean; it is therefore common practice to consider any instances of these extreme cases as outliers and to exclude them from analyses (Olewuezi, 2011). For all analyses (except aforementioned post hoc comparisons), α = .05 was used to determine statistical significance.

Results

Exclusion of Outliers and Missing Data

On average, participants reported an average of 2.55 prior arrests (SD = 8.03, range = 0–111). Using a 3-SD cutoff of 26.64, two veterans were excluded from statistical analyses; these subjects reported a history of 54 and 111 prior arrests, respectively. Consistent with the high suicide risk of this sample, participants also reported an average of 2.97 prior suicide attempts (SD = 7.43, range = 0–102). Using a standard deviation cutoff of 25.26, two subjects were excluded from statistical analyses; these individuals reported a history of 30 and 102 prior suicide attempts, respectively. After these exclusions, the final sample size was 282.

Missing data for the data set were as follows: 1 participant lacked data for all demographic information, 1 for age, 11 for suicide attempt history, 4 for suicidal ideation, 7 for BSS, 27 for BDI, 20 for BAI, 110 for CAPS, and 3 for SCID-IV diagnoses; there were no missing data for SIDP-IV diagnoses.

Descriptive Statistics

After exclusion of outliers, nearly half of veterans reported a history of one or more arrests (n = 133). Of those reporting at least one arrest, subjects reported an average of 4.23 total prior arrests (SD = 4.45, range = 1–20), including an average of 0.98 arrests for violent crimes (SD = 1.80, range = 0–10), 0.56 for nonviolent crimes (SD = 0.99, range = 0–5), 1.63 for drug-related crimes (SD = 3.43, range = 0–20), and 0.28 for technical violations (SD = 0.89, range = 0–6). See Table 1 for a summary of veteran arrest history.

Of the 271 veterans who provided information about suicide history, approximately 59% reported a history of at least one suicide attempt. Of those reporting at least one attempt, participants reported an average of 4.04 prior suicide attempts (SD = 4.18, range = 1–25). Scores on the BSS similarly suggested elevated suicidal ideation within this specialized sample (M = 7.40, SD = 9.12, range = 0–35).

Through diagnostic interviews, the most common diagnoses included substance use disorder (74%), major depressive disorder (57%), borderline personality disorder (48%), and PTSD (48%), whereas the least common diagnoses included histrionic personality disorder (0.35%) and bipolar disorder (11%). On measures of psychopathology, veterans reported experiencing, on average, moderate depressive symptoms on the BDI (M = 22.61, SD = 12.97), moderate anxiety symptoms on the BAI (M = 20.35, SD = 13.09), and severe posttraumatic symptoms on the CAPS (M = 72.24, SD = 35.01). See Table 2 for more detailed descriptive statistics.

Table 2.

Descriptive Statistics

Variable	No arrests, M (SD) or % (n = 149)	Any arrests M (SD) or % (n = 133)	Statistical comparison
Suicidality
Suicide attempt frequencya	2.21 (3.21)	2.55 (4.32)	t(269) = −0.73, p = .47
Suicide attempt prevalencea	60.00%	57.14%	χ2(1) = .23, p = .63
Suicidal ideation severitya	1.55 (0.77)	1.60 (0.72)	t(276) = −0.59, p = .56
Suicidal ideation prevalencea	83.11%	86.15%	χ2(1) = .49, p = .48
Beck Scale for Suicidal Ideation	7.53 (9.19)	7.26 (9.07)	t(273) = .24, p = .81
Psychiatric diagnosis
Depressive disorderb	57.14%	58.33%	χ2(1) = .04, p = .84
Bipolar disorderb	10.88%	12.12%	χ2(1) = .11, p = .75
Anxiety disorderb	17.69%	20.45%	χ2(1) = .35, p = .56
Posttraumatic stress disorderb	49.00%	46.21%	χ2(1) = .21, p = .64
Substance use disorderb	60.54%	88.64%	χ2(1) = 28.41, p < .01
Antisocial personality disorderc	6.08%	33.08%	χ2(1) = 33.37, p < .01
Borderline personality disorderc	44.97%	50.38%	χ2(1) = .83, p = .36
Histrionic personality disorderc	0%	0.75%	χ2(1) = 1.12, p = .29
Narcissistic personality disorderc	8.72%	15.04%	χ2(1) = 2.71, p = .10
Psychopathological symptom severity
Beck Depression Inventory	22.71 (13.78)	22.49 (12.00)	t(253) = .14, p = .89
Beck Anxiety Inventory	20.63 (12.97)	20.03 (13.27)	t(260) = .37, p = .72
Clinician-Administered PTSD Scale	70.00 (36.00)	75.00 (34.00)	t(170) = −0.86, p = .39
Demographics
Age	38.03 (11.13)	43.90 (9.64)	t(278) = −4.69, p < .01
Sex (% male)	62.16%	85.71%	χ2(1) = 19.86, p < .01
Race (% White/Caucasian)	17.57%	18.80%	χ2(1) = .07, p = .79
Marital status (% never married)	45.27%	29.32%	χ2(1) = 7.58, p < .01

Note. PTSD = posttraumatic stress disorder.

^aAs assessed by the Columbia Suicide Severity Rating Scale.

^bAs assessed by the Structured Clinical Interview for DSM–IV.

^cAs assessed by the Structured Interview for DSM–IV Personality Disorders.

Relation of Arrest History, Suicide Risk, and Psychopathology

Inconsistent with Hypothesis A, the presence of an arrest history was not associated with a history of suicide attempt, current suicidal ideation, or severity of psychopathological symptoms in categorical analyses (all ps ≥ .39). However, consistent with hypotheses, presence of an arrest history was associated with diagnosis of substance use disorder, χ²(1) = 28.41, p < .01, and antisocial personality disorder, χ²(1) = 33.37, p < .01. Arrest history was also associated with participant age, sex, and marital status, such that individuals reporting a history of arrest tended to be older, were more likely to be male, and were more likely to have been previously married (all ps < .01). See Table 2 for further information. Follow-up post hoc analyses suggested no significant differences in rates of substance use disorder, antisocial personality disorder, age, male sex, or marital status across arrest types. See Table 3 for further information about post hoc analyses.

Table 3.

Post Hoc Comparisons by Arrest Type

Variable	Violent crime arrest, M (SD) or % (n = 57)	Nonviolent crime arrest, M (SD) or % (n = 45)	Drug crime arrest, M (SD) or % (n = 57)	Technical crime arrest, M (SD) or % (n = 18)	Statistical comparison
Substance use	94.74%	91.11%	94.64%	100%	χ2(3) = 1.35, p = .72
Antisocial personality	36.84%	40.00%	42.11%	50.00%	χ2(3) = 1.05, p = .79
Age	43.32 (9.04)	43.89 (9.76)	45.21 (9.85)	48.89 (5.45)	F(3, 176) = 1.84, p = .14
Sex (% male)	82.46%	91.11%	87.72%	88.89%	χ2(3) = 1.80, p = .62
Marital status (% never married)	28.07%	31.11%	29.82%	38.89%	χ2(3) = .78, p = .86

Partially consistent with Hypothesis B, correlational analyses suggested the diagnosis of substance use disorder was similarly associated with both suicide and arrest history in this high-risk sample. Specifically, veterans with a substance use disorder were more likely to have a history of at least one suicide attempt, to report more severe suicidal ideation, and to have a more extensive arrest history than veterans without a substance use disorder (all ps ≤ .03). By contrast, antisocial personality disorder appeared uniquely associated with a higher frequency of prior arrests (p < .01), whereas borderline personality disorder and PTSD were uniquely associated with measures of suicide risk (all ps ≤ .02). See Table 4.

Table 4.

Correlations Between Suicide Risk Factors, Arrest History, and Psychopathology

Variable	Suicide attempt frequency	Suicide attempt (categorical)	Suicidal ideation intensity	Suicidal ideation (categorical)	Beck Scale for Suicidal Ideation	Frequency of prior arrests
PTSDa	.12	.14*	.11	.11	.13*	−.10
Substance use disordera	.07	.13*	.18**	.19**	.06	.24**
Antisocial personality disorderb	.01	.01	.05	.05	.07	.42**
Borderline personality disorderb	.34**	.46**	.44**	.35**	.36**	−.02

Note. PTSD = posttraumatic stress disorder.

^aAs assessed by the Structured Clinical Interview for DSM–IV.

^bAs assessed by the Structured Interview for DSM–IV Personality Disorders.

^*p < .05.

^**p < .01.

Correlational analyses were also partially consistent with Hypothesis C. Narrowing analyses to include only veterans reporting a history of arrest, frequency of prior arrests correlated only with greater likelihood of antisocial personality disorder diagnosis, r = .35, p < .01; greater likelihood of being male, r = .18, p = .04; and increased age, r = .20, p = .03. Again, no significant associations were noted between frequency of prior arrests and frequency of suicide attempts, r = −.02, p = .84; severity of suicidal ideation, r = −.04, p = .66; measures of psychopathological symptom severity (all ps > .19); or likelihood of other psychiatric diagnoses (all ps > .08).

Discussion

The current study is first to examine the prevalence and psychopathological correlates of prior legal involvement among veterans at risk for suicide. Results indicate that approximately half of veterans identified as at risk for suicide also had a history of arrest, and these veterans were more likely to present with antisocial personality and substance use disorders than at-risk veterans without such legal history.

Prevalence of Arrest History Among At-Risk Veterans

Previous research suggests suicide risk is often elevated among persons who are currently and/or previously justice involved (e.g., Barry et al., 2018; Fazel et al., 2016; Saxon et al., 2001). Similarly, members of the general population who attempt suicide are more likely to have a history of recent arrest than those who do not attempt suicide (Cook, 2013). To date, however, no research has examined whether this association between suicide risk and justice involvement is also salient among veterans at risk for suicide. The current study represents the first such investigation.

Broadly, results suggest high rates of arrest history among veterans at risk for suicide. Although suicidal ideation and attempts were not statistically associated with arrest history in the current sample, 47% of at-risk veterans had a history of arrest, including 20% with a history of arrest for drug-related crimes, 19% for violent crimes, 16% for nonviolent crimes, and 6% for technical offenses (20% of veterans also reported an arrest for unspecified crimes). Among those with a history of arrest, veterans reported an average of 4 arrests, with a range of 1–111 arrests total. These rates are in stark contrast to epidemiological estimates of approximately 22% veterans in the general population having a history of justice involvement (Edwards et al., 2020). Future research may wish to compare these arrest rates to those of nonveterans at risk for suicide; such investigation could aid in determining whether these elevated arrest rates are unique to veterans or reflective of broader social patterns.

Because nearly half of at-risk veterans had a history of arrest, legal status and justice involvement history may be necessary considerations when working with veterans at risk for suicide. History of legal involvement is associated with a wide range of difficulties, including barriers to employment (Apel & Sweeten, 2010), challenges in obtaining housing (Geller & Curtis, 2011), difficulties in relationships (Harman, Smith, & Egan, 2007; La Vigne, Naser, Brooks, & Castro, 2005), and disturbances in self-image (Oser, 2006), each of which may fuel elevations in suicide risk (Bhar, Ghahramanlou-Holloway, Brown, & Beck, 2008; Chu et al., 2017; Eynan et al., 2002; Schneider et al., 2011). Assessing legal status and history when working with veterans at risk for suicide would therefore provide opportunity for more thorough conceptualizations of suicide risk as well as more thorough understanding of potential barriers to case management and suicide prevention.

Mental Health of At-Risk Veterans With an Arrest History

Inconsistent with hypotheses, arrest history was not statistically related to the presence or absence of prior suicide attempts or ideation, frequency of prior suicide attempts, severity of suicidal ideation, or psychopathological symptom severity within this specialized sample of veterans at risk for suicide. These findings are largely inconsistent with previous research suggesting suicide risk to be elevated among justice-involved veterans and suggest the association between suicide risk and justice involvement is likely more complex than previously suggested. Without further research, however, the cause of this discrepancy in findings remains unclear. It is possible, for example, a ceiling effect driven by the elevated suicide risk of all participants in this unique sample limited ability to identify variance uniquely associated with arrest history. Indeed, among veterans with a history of arrest in this sample, 57% reported a history of suicide attempt compared to only 5% of veterans in the general population (Edwards et al., 2020). Alternatively, it is possible that while suicide risk is elevated among justice-involved veterans, such risk is not greater than is typically observed among veterans at risk for suicide. Continued research is needed to clarify these complex associations between legal involvement and suicidality.

Consistent with theories suggesting psychopathology may act as a shared risk factor for both suicide and legal involvement, the presence of a substance use disorder was similarly associated with both suicide risk factors and arrest history in this sample of veterans. Antisocial personality disorder, while not statistically associated with suicide attempt history or suicidal ideation, was also associated with a more extensive history of arrest and overrepresented in this sample of high-risk veterans (19% in the current sample vs. 6% in the broader veteran population; Edwards et al., 2020). These clinical presentations also appeared across arrest type (i.e., for violent, nonviolent, drug, and technical offenses). Such findings are consistent with previous research suggesting robust associations between antisocial personality disorder, substance use disorders, and criminal behavior (Bebbington et al., 2017; Trull et al., 2016) and lend further support to theories that implicate substance use disorders and/or antisocial personality disorder as potential shared risk factors for suicide and legal involvement. Given the high rate of comorbidity between substance use disorders and antisocial personality disorder (Trull et al., 2016), future research may aim to examine the independent contributions of these disorders on predictions of suicide risk and/or legal involvement.

Unfortunately, few (if any) suicide prevention efforts specifically address criminogenic factors, such as antisocial personality and substance use disorders. Nevertheless, these factors appear particularly relevant for the 47% of veterans at risk for suicide who also have a history of legal involvement. Further, although substance use disorders are commonly explored within the context of veteran suicide (e.g., Bohnert et al., 2017), antisocial personality disorder and criminal behavior are rarely examined within the context of suicide research and prevention. Particularly considering the substantial research suggesting these mental health conditions commonly interfere with treatment effectiveness when not adequately addressed (Marshall et al., 2010; Najt, Fusar-Poli, & Brambilla, 2011; Westermeyer & Thuras, 2005), assessing and treating these clinical needs is of utmost importance in working with veterans at risk for suicide. For example, suicide prevention efforts may consider integrating elements from offender rehabilitation programs that have proven promising for veteran populations, such as those based in the risk-need-responsivity model (Timko et al., 2014). For example, suicide prevention programming may integrate modules on prosocial problem-solving strategies, building prosocial relationships, or challenging antisocial thinking patterns. The structured, skills-focused format of standard offender rehabilitation programs is often compatible with leading suicide prevention programs that target problem solving, interpersonal skills, and managing unhelpful thoughts (e.g., dialectical behavior therapy [Linehan, 1993]; Safety Planning Intervention [Stanley & Brown, 2012]), providing opportunity for seamless integration.

Notably, however, PTSD and borderline personality disorder were not associated with arrest history in the current sample and instead appeared uniquely associated with suicide risk factors. Such findings raise questions about whether PTSD and borderline personality disorder also act as shared risk factors for suicide and legal involvement. However, in light of previous research suggesting prevalence of these disorders to be routinely overrepresented in forensic populations, continued research is necessary.

Demographics of At-Risk Veterans With an Arrest History

Observed associations between veteran arrest history and age suggest older veterans at risk for suicide may be more likely to have a history of legal involvement. These generational differences are consistent with historical overrepresentation of veterans within the correctional system (Bronson et al., 2015). Throughout the 1980s, veterans were incarcerated at rates 33–46% higher than that of nonveterans; however, by the early 2010s, veterans were incarcerated at 5–17% less than the rate of nonveterans (Bronson et al., 2015). Although the recency of participant arrest histories was not collected through the current study, estimates suggest most justice-involved veterans are incarcerated between the ages of 35 and 55 (Bronson et al., 2015). Further, many effects of legal involvement are lifelong (e.g., restrictions on employment, housing, benefits; Finzen, 2005; Schnittker, 2014). As such, although younger veterans are not currently overrepresented in the correctional system (Bronson et al., 2015; Tsai, Rosenheck, Kasprow, & McGuire, 2013b), many older veterans are likely to be suffering continued effects of past legal involvement.

Associations between veteran arrest history and sex are also consistent with extensive literature, suggesting legal involvement to be more common among male as compared to female samples (Bronson et al., 2015; Tsai et al., 2013b). Notably, however, female veterans accounted for approximately 15% of at-risk veterans with a history of arrest in the current sample—a rate much higher than the 1–2% rate observed in incarcerated samples (Bronson et al., 2015; Tsai et al., 2013b). These findings suggest that although an arrest history is more likely among male veterans at risk for suicide, a substantial proportion of at-risk female veterans may also have this history. As such, clinicians should not rely on heuristics of legal involvement and antisocial behavior as being restricted to male clients when working with at-risk veterans.

Limitations and Future Directions

The current study is characterized by several methodological strengths, most notably the inclusion of a well-defined and highly unique veteran sample. However, results should also be understood within the context of a few methodological limitations. First, whereas previous research in this area has tended to include samples with overall low risk for suicide, the current study is characterized by the opposing limitation. Lack of a low-suicide-risk comparison group limited the extent to which descriptive statistics could be interpreted and may have contributed to potential ceiling effects surrounding factors related to suicide risk. This limitation has especially notable implications for interpretation of observed associations that were not statistically significant (e.g., between suicide risk factors and arrest history) and of variables that, while overrepresented in this high-suicide-risk sample relative to the broader veteran population, were not statistically associated with suicide risk factors (e.g., antisocial personality disorder). Future research that includes demographically matched, low-risk veterans (e.g., recruited from the community) together with high-risk veterans is needed to provide more context for understanding and interpreting results.

Second, the cross-sectional nature of data collection did not allow for inferences of directionality. It therefore remains unclear to what extent legal history may precede and/or increase factors related to suicide risk (as suggested by the interpersonal theory of suicide; Simlot et al., 2013) or, conversely, the extent to which suicide risk may precede and/or increase factors related to legal involvement. Relatedly, dates of military service were not collected as a part of data collection, raising questions about the potential effects of service and service-related variables on suicide, suicide risk, and justice involvement. Use of longitudinal research designs and/or collection of dates at which suicide attempts, arrests, and military service occurred could help in clarifying these issues of directionality.

Third, data collection instruments were also characterized by a few limitations. A heavy reliance on self-report data (e.g., of arrest history, suicide attempt history), for example, may have introduced biases in participant responding, thereby limiting the validity of findings. Although previous research suggests self-reported legal and suicide histories are generally valid (Morris & Slocum, 2010; Nieves et al., 2000; Plöderl, Kralovec, Yazdi, & Fartacek, 2011), future research may aim to validate self-reported histories through review of official data, such as criminal records or hospital records. Also, the study’s use of the SCID-IV to make psychiatric diagnoses raises questions of diagnostic validity, particularly regarding disorders for which diagnostic criteria changed with publication of the DSM–5 (e.g., PTSD). Continued research using updated diagnostic instruments is therefore needed.

Fourth, diagnoses of antisocial and borderline personality disorders specifically reference engagement in illegal and suicidal behavior, respectively. Although these criteria are not necessary for diagnosis, this overlap between diagnostic criteria and assessed outcomes (i.e., arrest and suicide risk) may have inflated estimates of association.

Finally, the study’s sample size, while considerable in comparison to other research on veteran suicide, may have limited statistical power to identify differences at the level of arrest type. In the current sample, veterans reported comparatively few technical offenses (e.g., parole violation, contempt of court) and many drug offenses (e.g., possession, sale). Inclusion of a larger sample size may help to mitigate these differences in rate of arrest type in future research.

Conclusions

The current study investigated mental health and suicide correlates of arrest history within a sample of veterans at risk for suicide. Results suggested approximately half of at-risk veterans had a history of arrest, and these veterans were more likely to be older, to be male, and to present with antisocial personality and substance use disorders than at-risk veterans without a history of arrest. Such findings attest to the importance of assessing legal status and history when working with veterans at risk for suicide and integrating such assessment into conceptualizations and treatment planning.

Impact Statement.

In a sample of Veterans at risk for suicide, nearly half had a history of arrest. These Veterans were more likely to be male, older, previously married, and diagnosed with antisocial personality and/or substance use disorders than Veterans without such history. Findings attest to the importance of assessing legal status and history when working with Veterans at risk for suicide.