Introduction

The military-crime nexus is well-established in the criminological literature. Indeed, several studies show that veteran status is associated with a number of criminal justice outcomes, including an increased likelihood of substance abuse, arrest, incarceration, and recidivism (Greenberg & Rosenheck, 2009; Institute of Medicine, 2013; Van Dyke & Orrick, 2017; White, Mulvey, Fox, & Choate, 2012; White, 2013). Regarding incarceration, recent estimates by the United States Department of Justice indicate that over 180,000 veterans are serving time in correctional facilities (Bronson, Carson, Noonan, & Berzofsky, 2015). However, explicit research on incarcerated veterans has been peripheral in the criminological literature. While some studies have designated prior military service as a control variable in their analyses (Goetting & Howsen, 1986; Maguire, Flanagan, & Thornberry, 1988), a paucity of empirical research exists regarding the actual institutional experiences of veterans who end up in prison (Logan & Pare, 2016; May, Stives, Wells, & Wood, 2017). Accordingly, despite what is known about the link between veteran status, criminality, and the likelihood of imprisonment, this knowledge itself does not generally extend further into the prison environment.

This is problematic for two major reasons. First, a number of veterans return from tours of duty having experienced both physical and psychological trauma—the latter of which often manifests in the form mental disorder, including most notably post-traumatic stress disorder (PTSD), which has been linked to an increased risk of negative life outcomes such as victimization, incarceration, and institutional maladjustment (Accordino, Porter, & Morse, 2001; Bronson et al., 2015). Second, the extent to which individuals adjust to incarceration is contingent on a number of compositional factors, including their sociodemographic characteristics, their mental health histories, and their criminal histories—differences that may be especially pronounced between veteran and nonveteran inmates (Logan & Pare, 2016; Tsai, Rosenheck, Kasprow, & McGuire, 2013). For instance, reports by Armor and Sackett (2004), Noonan and Mumola (2007), and White (2013) suggest that veteran inmates residing in state and federal correctional facilities are, on average, older and better educated than their nonveteran counterparts. They are also more likely to be steadily employed prior to their contact with the criminal justice system. Conversely, some reports indicate that veteran inmates are also (1) more likely to report having been diagnosed with a mental health disorder; (2) more likely to be serving time for serious, violent offenses; and (3) more likely to receive lengthier sentences, including life sentences and death sentences, relative to the nonveteran inmate population (Bronson et al., 2015). Importantly, past research on institutional adjustment indicates that these characteristics are linked to both prosocial and antisocial behavior within the prison context (DeLisi, Trulson, Marquart, Drury, & Kosloski, 2011).

In light of these differences, there is reason to suspect that veteran inmates differ from their nonveteran counterparts with respect to the prison experience. Indeed, there are two schools of thought on how incarcerated veterans respond to imprisonment. On the one hand, some scholars have argued that serving in the armed forces is a difficult but ultimately positive experience—one that serves to make ordinary people especially resilient when confronted with personal challenges (Cullen, Blevins, Trager, & Gendreau, 2005; Logan & Pare, 2016). Such resiliency may serve as a protective factor for incarcerated veterans, assuming that their training provided them with the necessary discipline, character, and skills to adapt to a regimented institution like prison (Sun, Sung, & Chu, 2007). On the other hand, some scholars maintain that prior military service may increase the risk of maladaptive prison behavior—especially for veterans who have sustained physical and psychological scars, or who have been “violentized” as a result of being indoctrinated into a military culture based on masculinity and emotional suppression (Athens & Ulmer, 2003), and for those who have been exposed to intense violence during combat (Hajjar, 2014; Lunasco, Goodwin, Ozanian, & Loflin, 2010; McCormick-Goodhart, 2012).

Guided by the broader rehabilitative movement in contemporary criminal justice policy, increasing attention has also been paid to the specific risks and needs of justice-involved veterans (Andrews & Bonta, 2010; Blodgett et al., 2015). For instance, compared to the general adult offender population, veterans appear to exhibit more mental health, trauma, and substance abuse issues, including associated comorbidities, that could potentially be ameliorated through the involvement of specialty veterans treatment courts and other outreach or re-entry programs to lower recidivism (Hartley & Baldwin, 2016; Schaffer, 2016). Naturally, the continued development of these initiatives rests on the assumption that veterans constitute a distinct group capable of benefiting from targeted treatments that addresses their unique criminogenic needs (Blonigen et al., 2017). Taking these perspectives into account, the present study provides an investigation into the prevalence of mental disorders and suicidal behaviors among veteran and nonveteran inmates and further examines how these issues affect psychological adjustment and the obtainment of mental health treatment within prison using nationally representative data from the Survey of Inmates in State and Federal Correctional Facilities, 2004.

Incarcerated Veterans

The idea that incarcerated veterans differ from their nonveteran counterparts in terms of their ability to adjust to imprisonment stems from the fact that they are substantially different regarding their social and demographic background characteristics; however, with the exception of a few recent studies, this perspective has received limited attention in the academic literature (May et al., 2017). Research by Logan and Pare (2016), for example, used self-report data to examine differences in the rates of institutional misconduct between veteran and nonveteran inmates. Results from their logistic regression analyses indicated that, relative to nonveteran inmates, veteran inmates were significantly less likely to (1) be found guilty of verbal misconduct toward staff or other inmates; (2) be found guilty of physical assault against other inmates; (3) be found guilty of drug-related infractions; (4) be found guilty of insubordination or other major infractions; and (5) be placed in solitary confinement. The authors partially attributed their findings to the fact that veteran and nonveteran inmates differed with respect to their average levels of age and education; specifically, veteran inmates were older and better educated.

Past research on institutional adjustment also shows that pre-prison variables such as age and socioeconomic status influence the prison experience, and studies indicate that younger, less-educated inmates tend to be at greater risk of experiencing victimization, institutional misconduct, and affective disorders (Cesaroni & Peterson-Badali, 2010; Sappington, 1996). For example, Wooldredge (1999) found that less-educated inmates were more likely to experience prison-related stressors, including signs of depression and anxiety, while Sappington (1996) observed a positive relationship between education and perceived control over the prison environment including the belief that (1) one might control one’s own behavior; (2) one’s actions might affect one’s treatment; and (3) one might enjoy oneself in prison. These findings support the well-known importation model of prison adjustment, which is based on the notion that inmates “import” with them a host characteristics that affect the extent to which they are able to adapt to incarceration (DeLisi et al., 2011). For the purposes of the current study, differences in perspective regarding whether and how veteran inmates adjust to incarceration represent variations of the importation model. Each perspective posits that the social, personal, and psychological characteristics of veteran inmates should affect how they fare in the correctional setting; however, they disagree on exactly how these effects should play out.

Veteran status outside of the prison context is synonymous with notions of maturity, intelligence, discipline, respect, and leadership—the likes of which may be easily transferred into the prison environment to serve as a buffer against the pains of imprisonment for veteran inmates (Logan & Pare, 2016; Sykes, 1958). As previously mentioned, research shows that veteran inmates are more likely to be older and better educated than other inmates (Armor & Sackett, 2004; Noonan & Mumola, 2007; White, 2013). Regarding the current study, then, it may be that veteran inmates are better than nonveteran inmates at avoiding conflicts with others (e.g., through verbal negotiation or other means). They may be more mature, less confrontational, and—through their military training—have had more time to develop and hone the social, mental, or physical skills necessary to navigate the difficulties associated with prison life. Moreover, to the extent that the military represents a “total institution,” as some scholars have suggested (Sun et al., 2007), veteran inmates may have an easier time adjusting to their imprisonment, as several characteristics of the prison environment—such as constant monitoring, lack of privacy, isolation from civilians, and highly structured routines—mirror the trappings associated with military life. Recent research by May et al. (2017) supports this logic. Using self-report data on more than 1100 inmates across six state prisons in a Midwestern state, the authors found that veteran inmates were more likely than their nonveteran counterparts to favor prison sentences over less punitive alternative sanctions, such as intensive supervision probation and community service. In explaining these differences, the authors posited that veteran inmates saw prison as a better, more expedient sanction than the requirements necessary to complete a community-based sanction—the criteria of which fostered feelings of impatience, anxiety, and irritability among an already vulnerable group.

Conversely, it is possible that veteran inmates might import with them into the prison context a number of characteristics which could serve to increase their institutional pains. Such attributes, as Hawkins (2009) remarked in his justification for specialized veteran courts, often revolve around concerns that veterans “may have been subjected, even repeatedly subjected, to life-threatening events the general public may never know” (p. 569). Reports by White (2013) and Bronson et al. (2015), for example, note that the frequency of mental disorder, including PTSD, is greater among veteran arrestees (see also Greenberg & Rosenheck, 2009; White et al., 2012). Similar observations were made by Logan and Pare (2016), who found that the prevalence of PTSD was significantly higher in the veteran inmate population, relative to the nonveteran population in their study of institutional misconduct rates of inmates housed in state and federal correctional facilities. Moreover, among recent veterans or those still in the military, the decision to be screened or treated for PTSD is likely constrained by concerns over confidentiality and the belief that they would be branded with “a scarlet P,” as Friedman (2004) notes:

[Those] who reported the greatest number or the most severe symptoms were the least likely to seek treatment for fear that it could harm their careers, cause difficulties with their peers and with unit leadership, and become an embarrassment in that they would be seen as weak. (p. 77)

According to estimates, as many as 1 in 8 soldiers who served during the 2003 Iraq War exhibited signs of PTSD upon returning from active duty (Hoge et al., 2004), and several empirical studies have positively linked PTSD to post-military antisocial behavior, including violent and drug-related offenses. For instance, Fontana and Rosenheck’s (2005) analysis of the data from the National Vietnam Readjustment Study indicated that the relationship between combat exposure and antisocial behavior—including both violent (e.g., stabbing) and nonviolent (e.g., drug use) offenses—was mediated by the onset of PTSD. They also noted that the extent to which veterans received either a positive or negative homecoming reception—from friends, family, and the wider community—influenced the likelihood of PTSD manifesting into future antisocial behaviors. Specifically, veterans who received “colder” receptions were more likely to feel as though their service had been undermined which, in turn, increased their chances of engaging in dysfunctional (i.e., violent) behaviors as a way to cope.

To the degree that these behaviors do result in incarceration, veteran inmates suffering from PTSD or other mental disorders may have greater difficulty in adapting to imprisonment. Specifically, they may be less likely than other inmates to seek out treatment which may be exacerbated by—and a manifestation of—a broader military culture that promotes identities based upon emotional suppression, increased pain tolerance, stoicism, and self-sacrifice at the expense of other important cultural orientations, such as that of a peacekeeper, diplomat, technician, or teacher (Hajjar, 2014). Indeed, both quantitative and qualitative research suggests that veterans’ perceptions of community mental health services are at least partially influenced by the extent to which they adhere to the “warrior code” (Lunasco et al., 2010). For example, Pietrzak, Johnson, Goldstein, Malley, and Southwick (2009) examined a sample of 272 veterans from Operation Enduring Freedom (OEF) and Operation Iraqi Freedom (OIF) regarding perceived social support and beliefs about mental health treatment and found that service members who embraced a warrior identity were significantly less likely to seek out treatment due to the stigma and vulnerability associated with it.

In the same way, Sayers, Farrow, Ross, and Oslin (2009) conducted open-ended interviews with 44 veterans applying for Veterans Affairs (VA) disability status with respect to why they did or did not seek out treatment. Of the reasons given for not seeking mental health treatment, pride in self-reliance was the most salient; specifically, veterans preferred to deal with their problems on their own out of fear that they would appear vulnerable to others. As one participant lamented, “I guess it was kind of a view that to see a psychologist was for people who were weak, couldn’t take care of themselves, couldn’t deal with day to day life” (Sayers et al., 2009, p. 245). In essence, the veterans in this study expressed the desire to address their mental health problems privately and viewed professional help as a last resort. Regarding the current study, if personal beliefs related to mental health serve as a barrier to treatment outcomes for veterans outside the prison context (Vogt, 2011), then there is also reason to suspect that incarcerated veterans would be less likely to receive or seek out treatment options—especially since (1) inmate social systems are comparable to those in the military in that they are based on perceptions of toughness, stoicism, and grit (Clemmer, 1958; Irwin & Cressey, 1962) and (2) many disorders in prison go unnoticed, unreported, or undiagnosed by correctional staff (Diamond, Wang, Holzer, & Thomas, 2001).

Methodology

The current study seeks to examine whether veteran inmates differ significantly from nonveteran inmates in their reporting of mental disorders and suicidal behaviors, adjustment to prison life, and their likelihood of seeking out or receiving mental health treatment while incarcerated. Based upon this research question and a review of the current literature, we formed two hypotheses:

  • H1: Veteran inmates will be significantly more likely than nonveteran inmates to report a history of mental disorder diagnoses, suicidal behaviors, and higher levels of psychological maladjustment.

  • H2: Veteran inmates will be significantly less likely than nonveteran inmates to seek out or receive mental health treatments such as medication, counseling, or hospitalization in prison.

Data Source and Sample

The current study utilizes data provided by the Department of Justice’s Survey of Inmates in State and Federal Correctional Facilities, 2004—a nationally representative sample based on the self-reports of 18,185 inmates housed in 287 state and 39 federal correctional facilities. Using computer-assisted personal interviewing, the individual assessments were conducted between October 2003 and May 2004. To collect this data, the survey employed a two-stage research design. In the first stage, prisons were selected using a stratified random sampling approach with probability proportional to size. Then, in the second stage, inmates were selected using random sampling within state prisons and stratified random sampling within federal prisons. Notably, the survey also includes a structured clinical interview using a multitude of items modified from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) which permits the examination of various concepts related to psychological adjustment.

As previously noted, the full sample consisted of 18,185 cases, but we reduced this to better address our research questions by removing all females, representing approximately 21% of the sample, since the number of incarcerated female veterans was too small to permit reliable analyses (N = 64). Thus, our final analytical sample consisted of 12,680 male nonveterans and 1598 male veterans clustered within 261 separate prisons. As is common in large-scale surveys, not all individuals responded to every item in the questionnaire, resulting in a small amount of missing data across most variables of interest. The largest amounts were found on the number of prior arrests and time spent in prison variables where 4.9% and 4.3% of the data were missing, respectively. As a result, we utilized hotdeck imputation in SPSS 21 with an algorithm designed by Myers (2011), since this amount was well within the acceptable imputation range even for data that is potentially missing not at random.

Dependent Variables

The current study extends the research on inmate adjustment and assesses the extent to which veteran inmate status serves as either a risk or protective factor in prison by comparing relative likelihoods on the following outcomes: mental disorder diagnoses, suicidal behaviors, psychological maladjustment, and mental health treatment in prison.

Mental disorder diagnoses is a 7-item index capturing whether inmates had ever been told by a mental health professional, such as psychiatrist or psychologist, that they suffered from any of the following six disorders: depression, bipolar (including manic-depression or mania), schizophrenia (including psychotic disorders), PTSD, anxiety (including panic disorders), personality (including antisocial or borderline personality), or any other type of mental or emotional condition (α = .714). Inmates who reported being diagnosed with a mental disorder were given one point for each disorder they acknowledged. This index is designed to capture potential comorbidity, which is preferable to separate dichotomous measures that do not necessarily account for such overlap. We recognize that these diagnoses may include the experiences of veterans prior to their current incarceration; however, this measure still serves to establish whether or not the prevalence of mental disorders is higher in the veteran inmate population which theoretically influences the extent to which they are able to adjust to their incarceration, including whether or not they seek out or receive mental health treatment.

Suicidal behavior is a three-category nominal variable that indicates whether inmates had ever: (1) considered suicide or (2) attempted suicide, compared to those who did not (the reference category). Again, it is likely that suicidal behavior captures experiences that may have occurred outside of prison; nonetheless, we are primarily interested in understanding if veteran inmates are more likely to report such behaviors and whether or not this contributes to variation in prison adjustment or treatment.

Psychological maladjustment is a 20-item measure constructed to tap a variety of symptomatic issues related to negative affect, depression, changes in functioning, anxiety, and delusional thinking experienced by the inmate within the last year (α = .864). For a complete listing of the items used, please consult the supplementary file. While diverse, these psychological symptoms converge to provide a valuable assessment of the inmate’s recent mental and behavioral health inside the prison environment. Since the items comprising this measure were originally binary indicators, traditional exploratory factor analysis techniques using Pearson correlations were inappropriate (Bonett & Price, 2005). Accordingly, to examine the extent to which these items clustered together, we generated a tetrachoric matrix using Stata 12.1 and evaluated it by means of the factor analysis command. The results showed that the measure loaded strongly onto a single component (λ = 9.65, average loading = .693, KMO = .937). Given this consistency, we summed the items into an index ranging from 0 to 20 where higher scores indicated a greater amount of mental health problems reported by the individual. Although these reported symptoms are unsurprisingly correlated with official diagnoses, the effect itself is modest (r = .420). Moreover, this measure assists in capturing previously undetected mental health issues among inmates—issues for which veterans are characteristically less likely to seek out treatment and who are thus more likely to remain undiagnosed.

Mental health treatment is comprised of three separate dichotomous variables indicating whether inmates, since their current prison admission, had ever received medication for a mental or emotional problem (0 = No, 1 = Yes), had ever received trained counseling, therapy, or other treatment services (0 = No, 1 = Yes), or had been admitted to a mental hospital, unit, or treatment program (0 = No, 1 = Yes). Finally, it is important to note that we control for the aforementioned mental disorder diagnoses and suicidal behaviors when presenting our models in order to examine the degree to which inmates report symptoms related to poor adjustment and whether or not they sought out or received treatment.

Independent Variables

The primary independent variable of interest, veteran inmate status, is a binary measure based on whether the inmate had ever served in any branch of the United States Armed Forces prior to their admission (1 = No, 2 = Yes). Additionally, we later disaggregate this measure to conduct within-group analyses according to whether or not inmates had experienced combat during their military service (1 = No, 2 = Yes) and the type of discharge they received (1 = Honorable discharge, 2 = General discharge or worse).Footnote 1

We also include an array of covariates that have been associated with prison adjustment and which pertain to the inmates’ demographic characteristics, criminal histories, and their prison experiences. Age and education are continuous measures coded in years, where education ranges from kindergarten or less to two or more years of graduate school. Race or ethnicity is measured using a set of mutually exclusive variables including black, Hispanic, and other race, with white serving as the reference category. Employment history is based on whether or not an inmate was employed or ran a business during the month before their current arrest (1 = No, 2 = Yes). Marital status indicates whether an inmate was married at the time of the survey (1 = No, 2 = Yes).

Criminal History

Criminal history is captured using three measures: arrest history, offense type, and correctional history. Arrest history is a continuous measure based on the number of prior arrests an inmate had before their most recent admission to prison. Offense type is a dichotomous measure indicating whether or not the inmate was currently serving time for a violent offense, such as murder, aggravated assault, rape, or child abuse, with non-violent offenses (e.g., property, drug, public order, or unknown) serving as the reference category (1 = Non-violent, 2 = Violent). Correctional history is also dichotomous and notes whether an inmate had ever spent time in another correctional facility prior to their current admission (1 = No, 2 = Yes).

Prison Experience

The prison experience of inmates is based on five measures: time in prison, facility type, prison visits, prison infractions, and prison program participation. Time in prison is a continuous measure, coded in years, and was computed by subtracting the inmate’s date of admission from the year the survey was administered. Facility type is a dichotomous measure indicating whether inmates were housed in either a state or federal correctional facility (1 = State, 2 = Federal). Prison visits is a binary measure based on whether, over the past month, inmates had received a visit from individuals other than their lawyer (1 = No, 2 = Yes). Given that research indicates incarcerated veterans may engage in less institutional misconduct than nonveterans, prison infractions are captured using a variety score index of 15 separate rule violations the inmate could have been written up for or found guilty of since their admission. These offenses included drug or alcohol violations, weapon possession, physical or verbal assaults against staff or other inmates, general disobedience, and other major or minor infractions. Higher scores indicate a greater variety of rule-breaking violations committed by the inmate (α = .708). Finally, our measure of prison program participation is dichotomous and indicates whether inmates, since their admission, had ever participated in an educational, vocational, or job-related training program (1 = No, 2 = Yes).

Analytical Strategy

A series of negative binomial, ordinary least squares (OLS), and logistic regression analyses were used to estimate the effects of veteran inmate status and other relevant control variables on the presence of mental disorders and suicidal behaviors, as well as on various indicators of prison adjustment, including the exhibition of psychological symptoms and the use of mental health treatment while incarcerated. We conducted our analyses in an organized manner by examining disorders, suicide, and adjustment in successive fashion, allowing us to control for these variables in subsequent outcomes. In addition, while we accounted for state or federal facility type, there were no other aggregate-level factors available that were pertinent to our individual-level hypotheses. As a result, although we considered the use of multilevel modeling to assess these relationships, to aid interpretability we chose to estimate our models using Huber-White corrected standard errors which accounts for the possible dependence of observations of inmates clustered within 261 facilities (Rogers, 1993).Footnote 2 Similar to a multilevel model, where prison type would be controlled at level-2, robust clustering takes into consideration the likelihood that inmates housed in the same facility might be more similar to one another than inmates across different facilities. Moreover, collinearity diagnostics were run to ensure that multicollinearity among the covariates was not an issue. The average value using the variance inflation factor (VIF) was 1.23, with no value exceeding 2 which is within the acceptable range in the social sciences (Fox, 2008). Again, it is important to reiterate that our primary objectives in this study are to (1) determine if veteran inmates display a greater lifetime prevalence of mental disorders and suicidal behaviors and to (2) determine if these prior mental issues affect recent psychological adjustment and the obtainment of mental health treatment in prison.

Results

The sample characteristics for veteran and nonveteran inmates are presented in Table 1. Consistent with the prior literature, veteran inmates were older (45.6 vs. 34.4 years old), more likely to be white (53.6% vs. 30.9%), more highly educated (12.3 vs. 10.6 “years” of education), more likely to have been employed prior to incarceration, and to be married. Although veteran inmates had fewer arrests and were less likely to have spent time in a correctional facility prior to their most recent admission, they were more likely to be serving time for a violent offense (51.6% vs. 43.6%) and to have spent a longer time in prison on their current sentence (5.5 vs. 4.1 years).

Table 1 Sample characteristics for veteran and nonveteran inmates (N = 14,278)
Full size table

Mental Disorder Diagnoses and Suicidal Behavior

Negative binomial regression analyses were used to examine the relationship between veteran inmate status and prior mental disorder diagnoses controlling for the inmate’s demographic characteristics, criminal history, and prison experiences. To conserve space, the following models are presented in the supplementary file. As anticipated, veteran status was significantly associated with higher scores on the mental disorder index compared to nonveterans (B = .375, p < .001). Additionally, a within-group analysis found that veterans who had experienced combat were significantly more likely to report mental disorders (B = .353, p < .01). Next, we conducted a series of multinomial logistic regression analyses, also presented in the supplementary file, to estimate the effect of veteran status on past suicidal behaviors controlling for the mental disorder index. We found that veterans were significantly more likely to report having considered suicide (B = .362, RRR = 1.43, p < .001), but they were not more likely to report having attempted suicide.Footnote 3 Interestingly, a within-group analysis revealed that combat experience was significant and negatively associated with attempting suicide among veterans after controlling for reported mental disorders (B = −.775, RRR = .460, p < .01). However, this finding is not entirely unexpected given that (1) in studies of larger military samples, deployment to a war zone is sometimes associated with a reduced suicide risk, (2) suicidal behavior and its timing are highly contextualized among veterans, and (3) these behaviors are often a result of comorbid mental disorders including PTSD, aggressive psychoses, anxiety, and depression (Kang et al., 2015; Reger et al., 2015).

Psychological Maladjustment

Table 2 presents results from our three-step OLS regression analyses estimating the effect of veteran inmate status on psychological maladjustment during the past year. Similar to the previous analyses, the first step controls for the basic demographic and history variables. Next, taking into consideration the fundamental role of individual mental health history in the presentation of psychological symptoms, step two includes additional controls for prior mental disorder diagnoses and suicidal behavior. Lastly, step three presents a within-group analysis for veterans only. In Model 1, veteran status is significant and positively associated with greater psychological maladjustment (B = .340, p < .01). However, in Model 2, after controlling for the significant effects of prior mental disorder diagnoses and suicidal behavior, veteran inmate status was rendered non-significant. Finally, in Model 3, our within-group analyses found that combat and discharge status did not exert any significant effect among veterans. Across all models, younger inmates, those who had spent less time in prison, and those who committed a greater variety of prison infractions remained significantly more likely to report symptoms of psychological maladjustment.

Table 2 Linear regression for psychological maladjustment
Full size table

Mental Health Treatment

Table 3 presents the results of our binary logistic regression analyses estimating the likelihood of veteran inmates receiving each of the three mental health treatment outcomes: medication, counseling, or hospitalization. We once more utilized a two-step design where mental disorder diagnoses, suicidal behavior, and recent symptoms of psychological maladjustment are added in the second step after the inmate’s demographics and individual history. No within-group analyses for veterans are presented for these outcomes in order to maintain parsimony as combat and discharge status did not display significant effects in any model. Beginning with Model 1, veteran inmates were initially 33% more likely than nonveterans to have received medication since their current incarceration (OR = 1.33, p < .001). However, upon controlling for the aforementioned covariates in Model 2, veteran status was rendered non-significant. As expected, individuals with more mental disorders (OR = 2.78, p < .001), those who engaged in suicidal behavior (OR = 1.69, p < .001), and those who exhibited psychological maladjustment were significantly more likely to receive medication (OR = 1.05, p < .001).

Table 3 Logistic regression for prison treatment outcomes
Full size table

In Model 3, veterans were again 46% more likely to have received counseling, therapy, or other treatment services than nonveterans since their admission (OR = 1.46, p < .001). Upon controlling for the mental health variables in Model 4, the effect of veteran status was rendered non-significant. Again, the mental disorder index (OR = 2.10, p < .001), suicidal behaviors (OR = 1.64, p < .001), and maladjustment exerted significant positive effects on the likelihood of receiving counseling (OR = 1.04, p < .001). Lastly, Model 5 shows that veteran inmates were twice as likely to be hospitalized in a mental health unit or treatment program compared to nonveterans (OR = 2.00, p < .001). After controlling for the mental health variables in Model 6, this effect is reduced, but remains significant (OR = 1.52, p < .05). While the mental disorder index was statistically significant (OR = 1.69, p < .001), psychological maladjustment was not, and suicidal behaviors exhibited a stronger effect than in the previous models (OR = 2.68, p < .001). This may suggest that mental health hospitalization is a result of actions where the inmate has presented an observable and active threat to himself requiring immediate or mandatory action by prison staff.

In sum, across all three outcomes, these results indicate that veteran inmates were more likely than their nonveteran counterparts to seek out or receive various forms of mental health treatment during incarceration. However, after controlling for the effects of prior mental disorder diagnoses, suicidal behavior, and psychological maladjustment, the relationship between veteran inmate status and two of the outcomes was fully mediated while the third was diminished. This implies the presence of an indirect effect of veteran status occurring between reported mental health issues and the likelihood of receiving treatment within the prison environment.

Discussion

The goal of the current study was to build on the limited empirical literature regarding the prison experiences of veteran inmates. Specifically, we sought to determine whether and how veteran inmates differed from their nonveteran counterparts with respect to mental disorders, suicidal behaviors, psychological adjustment, and mental health treatment outcomes while incarcerated. Consistent with past research and our first hypothesis, we found that veteran inmates had higher rates of mental disorder diagnoses, were more likely to consider suicide, and, before controlling for the aforementioned variables, reported poorer psychological adjustment in prison. However, contrary to our second hypothesis, we found that veteran inmates were actually more likely to seek out or receive mental health treatment during incarceration. Moreover, this effect was fully mediated for two of the outcomes after accounting for mental disorders, suicidal behavior, and issues of maladjustment within the last year. Based upon our analyses, these findings appear to be driven primarily by mental health history as opposed to veteran inmate status alone. Indeed, given that veterans do not significantly differ from nonveterans across two of the three treatment outcomes and are actually more likely to receive hospitalization indicates that they are presumably receiving care in accordance with their needs. Consequently, these results challenge the assumptions made in previous studies that veterans—by nature of the military’s cultural indoctrination toward values of emotional suppression and stoicism—are less likely to seek out resources related to mental health treatment, irrespective of whether they received an official diagnosis (Hajjar, 2014; Lunasco et al., 2010; Vogt, 2011).

Our findings, therefore, suggest that the barriers to mental health treatment might be qualitatively different for incarcerated veterans compared to those in the community context. For instance, upon incarceration, inmates are subjected to a variety of intake procedures—many of which involve assessments of mental health or suicide risk (Hardyman, Austin, & Peyton, 2004). Accordingly, if an inmate is diagnosed during admission, it is possible that the extent to which they receive treatment for a disorder is contingent upon the mandate of a particular facility (Adams & Ferrandino, 2008; Prins, 2014). Although veterans may underreport their mental health issues as a consequence of being less likely to seek out treatment prior to arrest, it is also probable that reviews of their official military records could potentially negate any such omission during the screening process; however, the survey used in this study did not collect such information. As previously mentioned, prisons are highly structured environments characterized by a lack of privacy and constant monitoring (Sun et al., 2007). In essence, veteran inmates might not face the same barriers to treatment within prison simply because they are under complete supervision and are thus forced to receive treatment based upon their intake assessment or prior medical history.

Alternatively, the nature of prisons might encourage inmates to be more proactive in seeking out various treatments than they otherwise would in the community context. As noted earlier, May and colleagues (May et al., 2017, p. 15) found that some veterans may prefer custodial over noncustodial sanctions because they “are better able to tolerate the predictable, regimentalized nature of prison over the uncertain nature of community sanctions”—the likes of which can make them feel “more anxious, irritable, and [impatient]” (see also Sayers et al., 2009). If veteran inmates view the prison environment as organized and predictable, then the barriers to treatment that they face in the community, including perceptions of stigma and weakness, might be mitigated—at least within the institutional setting. It is important to emphasize, however, that we are unable to distinguish whether or not the treatment received was by the volition of the inmates or whether they were based on actuarial risk or need assessments.

Notably, beginning in 2009, the national Veterans Justice Outreach (VJO) program was established to provide licensed specialists to assess justice-involved veterans’ needs, “refer and link them to care, and assist with access to care barriers” as they progress through each initial stage or “intercept point” of the criminal justice system (Blue-Howells, Clark, Berk-Clark, & McGuire, 2013, p. 49). Acting as a liaison, these specialists also collaborate with “law enforcement, defense counsel, prosecutors, jails, and courts”—potentially augmenting the services received upon incarceration (Finlay et al., 2016, p. 206). While the VJO program is often used in conjunction with veterans treatment courts to assist with mental health or substance abuse issues, homelessness, and pre-trial diversion—only about half of these settings offer empirically-based treatments capable of targeting the most serious recidivism risks such as possessing antisocial tendencies or having criminal associates (Blonigen et al., 2017). For convicted veterans exiting prison, the Health Care for Re-entry Veterans (HCRV) program was created in 2007 to serve as a nexus between correctional institutions and local Veterans Affairs facilities to provide pre-release assessments, treatment referrals, transitional support, and short-term case management (Finlay et al., 2017). While these outreach efforts appear promising in meeting the needs of veterans, they would not have existed during the time of the current study’s survey in 2004, and so future research on inmates may benefit from collecting these measures of agency contact.

Understanding the needs of veteran inmates and the potential barriers they face with respect to mental health treatment in correctional facilities has important policy implications. First, our results show that contrary to the previous research on the general population of veterans, veteran inmates are not necessarily less likely to seek out or receive treatment for a mental health condition. However, congruent with past research, we also found that veteran inmates had higher rates of mental illnesses, were more likely to consider suicide, and initially displayed poorer psychological adjustment. This suggests that although veterans are not necessarily a disadvantaged group, they still represent a distinct, at-risk population. Indeed, recent research has shown that the risk and protective factors correlated with institutional maladjustment are likely group-specific (McCuish & Corrado, 2017) and that some inmate groups—such as veterans—might face unique challenges in coping with institutional life, relative to others (Albertson, Banks, & Murray, 2017).

Second, if veteran inmates are qualitatively different (i.e., a distinct and culturally diverse population) from others in terms of their level of risk, then prison administrators should consider tailoring various programs to their specific needs, with a specific focus on “military service awareness training” for all correctional staff (Albertson et al., 2017, p. 29). Recent research by Blonigen et al. (2016), for instance, suggests that there are veteran-specific criminogenic needs related to factors such as combat exposure, PTSD, and traumatic brain injury (TBI)—all of which may warrant closer examination alongside the already established criminogenic risk factors. Along these lines, Fontana and Rosenheck’s (2005) analysis of the National Vietnam Readjustment Study indicated that the effects of combat exposure on antisocial behavior were completely mediated by the effects of PTSD. Furthermore, some research indicates that PTSD and TBI are comorbid, and that distinguishing between the two conditions regarding their impact on offending (and by extension, prison adjustment) is particularly difficult (Hill III, Mobo Jr, & Cullen, 2009; Hoge et al., 2008; Pinals, 2010). Given these findings, it may be instructive for future instruments to consider the potential interactive effect of PTSD and TBI in predicting criminal justice outcomes for veterans and, more specifically, indicators of adjustment among incarcerated veterans (Sreenivasan et al., 2013).

Third, for some correctional facilities in the United States, addressing the criminogenic needs of veterans has translated into developing specific initiatives within the jail or prison context, including the creation of separate “veteran wings.” These wings, analogous to the efforts of specialized veterans courts, were created for the specific purpose of helping veterans successfully reintegrate into society, and are characterized by a focus on programs directed at mental health, addiction, and substance abuse treatment, as well as employment skills training, and have thus far been used in several states across the country (Edelman, Berger, & Crawford, 2016; Hill, 2018; National Institute of Corrections, 2018). Although veteran-only cellblocks have only been recently implemented, correctional officials have posited that recreating a prison atmosphere similar to the trappings of military life will reduce the pains of imprisonment for incarcerated veterans—characteristics over which these former soldiers are likely to bond (Connor, 2017). Importantly, however, the effectiveness of these veteran-specific initiatives needs to be rigorously evaluated, as “policy and practice reform [in this area] has been hampered by a lack of empirical data,” which increases the risk of “responding to [incarcerated] veterans based on … unsubstantiated assumptions which [could] impact negatively on ex-service personnel” (Albertson et al., 2017, p. 30).

Limitations and Future Research

Although we produced several important findings, there are, as in all studies, avenues for improvement in future research. To start, we acknowledge that our measure of veteran status did not account for the historical period during which the inmates in our sample served. It is possible that prison experiences might differ between inmates who served in different cohorts and among those who served in multiple wars. For instance, those who enlisted during the Vietnam War might be qualitatively different on important measures of age, education, and aggressiveness than those who served during the wars in Afghanistan and Iraq. Additionally, the characteristics of soldiers could have been affected by more rigorous screening techniques employed by the United States military, including the recruitment of an all-volunteer force rather than draftees since 1973 (Culp, Youstin, Englander, & Lynch, 2013). As service members continue to return from deployments across the world, it is advisable that future studies should explicitly account for these cohort differences, so as to disentangle their potential effects when assessing differences in the prison experience between veteran and nonveteran inmates. In the same way, our data are cross-sectional and do not permit the analysis of veteran inmate trajectories throughout their incarceration or across additional treatment outcomes, leaving us with an incomplete picture of their prison experiences. As a result, we are unable to establish causality with respect to mental diagnoses, suicidal behavior, and psychological maladjustment as our measures predominately capture lifetime prevalence or an occurrence within the last year. The incorporation of longitudinal designs would therefore serve to illuminate these screening trends and potentially establish the temporal ordering of key variables.

Researchers should also explore the contribution of offense type for incarcerated veterans when examining prison outcomes. Indeed, violent offense type remained a significant predictor across models for considering and attempting suicide, as well as all treatment outcomes. While analyzing the specific crime type was beyond the scope of this study, suicide risk has been shown to be markedly increased in offenders who commit violent crimes—with more serious violent crimes exerting a stronger overall effect (Webb et al., 2012). The risk of suicide among veterans also appears to correspond to cohort era with Vietnam and Gulf War veterans exhibiting what has been termed a “healthy soldier effect,” wherein veterans display a lower risk of mortality than the general population due to strict standards of physical fitness and access to universal health care both during and after their service (Kang et al., 2015, p. 98). Conversely, recent OEF/OIF veterans appear to be presenting significantly higher rates of suicide than expected—veterans who are less likely to exist in this dataset. Although not significant in our study, receiving a discharge status “other than honorable” has been linked to increased suicide risk in the short-term upon separation from the military (Reger et al., 2015). Moreover, veterans that are discharged in this manner can be disqualified from receiving government benefits related to medical care, education assistance, and other financial support that may contribute to subsequent justice system involvement.Footnote 4

Finally, too few female veterans were available in our study which necessitated omitting them from our analyses; nevertheless, the growing presence of female soldiers on the battlefield has sparked new inquiries into their unique stressors with exploratory research suggesting they may be twice as likely as males to be diagnosed with PTSD (Street, Vogt, & Dutra, 2009). Indeed, among all justice-involved veterans, recent research has found that although female veterans have significantly fewer substance use disorders than male veterans, they are significantly more likely to have been diagnosed with a mental disorder and they also have lower odds of entering residential treatment (Finlay et al., 2015). Importantly, any actuarial assessment tailored to the experience of veteran inmates should account for the specific influence of gender, as it is probable that the criminogenic needs of male and female veteran inmates differ from one another (Van Voorhis, Wright, Salisbury, & Bauman, 2010). Along these lines, common reintegration efforts to reduce recidivism rates, such as cognitive behavioral therapy, are also likely to require “significant adaptation to veterans’ needs” and have yet to be empirically evaluated (Timko et al., 2014, p. 635). Timko et al. (2016), for example, have shown that veterans treatment courts continue to grow at a rapid rate, are beginning to relax their eligibility criteria to include a wider breadth of veteran offenders, and that nearly two-thirds now offer a peer mentoring component. The establishment of these special courts, cellblocks, and diversionary programs represent the first steps in addressing the mental health problems of veterans in a rehabilitative fashion through the criminal justice system (Seamone, 2011). However, the success of these approaches will hinge on a concerted effort between multiple government and community stakeholders as the demand for these support systems increases in the decades ahead (Christy, Clark, Frei, & Rynearson-Moody, 2012).