Research on serious mental illness (schizophrenia, bipolar disorder, and major depression) indicates that a majority in this population have been victims of interpersonal trauma such as physical or sexual abuse (
1–
5). Mueser and colleagues (
6) found that 44.7% of individuals with serious mental illness reported childhood sexual assault and that 46.7% reported adult sexual assault. Further, rates of posttraumatic stress disorder (PTSD) are considerably higher among people with serious mental illness than in the general population, ranging from 29% to 43% depending on the sample and measure used to assess PTSD (
7). Among incarcerated populations, rates of childhood and adult interpersonal violence are also high, as shown in several studies of incarcerated individuals with serious mental illness (
8–
11; Cusack KJ, Wise A, Herring A, 2011, unpublished manuscript).
Interpersonal violence is linked with many negative outcomes. In particular, a strong link is documented between interpersonal trauma and abuse of alcohol and drugs (
8,
12–
14). Despite evidence of this association, the pathways that link interpersonal abuse and substance use are not well understood. One possibility for the association between trauma and substance use disorders is that individuals use alcohol or drugs in an attempt to cope with trauma-related distress and symptoms. Among traumatized individuals, this self-medication hypothesis states that substance use is an attempt to avoid or escape the painful symptoms of PTSD (
15). The self-medication hypothesis is supported by findings of higher rates of alcohol and drug misuse among individuals with PTSD compared with the general population (
16). A number of studies examining the onset of PTSD and substance use disorders have found that PTSD tends to precede the development of substance use disorders (
12,
17–
22).
Chilcoat and Breslau (
17) tested several hypotheses of the trauma and substance misuse relationship in a large, prospective study of patients who were members of a health maintenance organization. The investigators found that PTSD greatly increased the risk of developing substance use disorders, but trauma exposure in the absence of PTSD did not. Conversely, they found little support for the hypothesis that alcohol or drug dependence increases the risk of developing PTSD.
A longitudinal study with data collected over ten years found an increased risk of drug use disorders among young adults who were exposed to trauma and developed PTSD but no change in risk of drug use disorder among those with non-PTSD trauma (
12). A more recent prospective study of adolescents found that PTSD predicted the onset of all substance use disorders over a four-year follow-up period but substance use disorders did not predict PTSD (
21). Other studies have found that severity of PTSD symptoms is associated with greater alcohol craving and that improvement in PTSD symptom severity leads to improvement in alcohol dependence severity, without the reciprocal relationship (
20,
22,
23). Although few studies have addressed the role of PTSD in the substance abuse of persons with serious mental illness, Mueser and colleagues (
7) outlined an interactive model of trauma, PTSD, and serious mental illness symptoms that posits a central role for PTSD in the path from interpersonal trauma to substance misuse and serious mental illness symptoms.
Not only is substance abuse associated with more severe symptoms and greater rates of hospitalization among people with mental illness, but it is also strongly associated with arrest (
24,
25). One study comparing reasons for arrest found that drug and alcohol abuse was responsible for a sizeable minority of arrests among individuals with serious mental illness (
25). Swartz and Lurigio (
26) found that substance use disorders almost completely mediated the relationship between serious mental illness and criminal justice involvement for nonviolent and drug-related offenses. Schladweiler and colleagues (
27) noted several studies indicating that individuals with co-occurring substance abuse or dependence were more likely to commit crimes and have a greater number of arrests than persons with serious mental illness without a substance use disorder.
Given the high rates of comorbid substance misuse among people with mental illness and its potential role in arrests of people with serious mental illness, a better understanding of factors that may contribute to substance misuse would have important clinical implications. If PTSD is found to be a mediator of the association between sexual abuse and substance abuse, this would suggest that efforts to identify and treat PTSD among people with serious mental illness in jail diversion programs are important and may lead to better mental health and outcomes concerning substance use and criminal justice. Integrated treatment models for PTSD and substance use disorders have been developed but have not been widely disseminated in community practice settings or in jail diversion programs. Evidence of a self-medicating role of substances among people with PTSD in criminal justice settings may suggest a need to further develop and disseminate such interventions. The purpose of this study was to examine the rate of PTSD in a sample of individuals with mental illness involved in jail diversion and test the role of PTSD as a mediator in the relationship between interpersonal trauma and substance misuse.
Methods
Participants
Study participants were 386 enrollees from seven jail diversion programs funded in 2006 or 2007 through the Targeted Capacity Expansion (TCE) Jail Diversion Initiative of the Substance Abuse and Mental Health Services Administration (SAMHSA). Participants ranged in age from 18 to 66 (mean±SD=38.94±11.07). Forty-five percent (N=173) of the sample were male. Fifty-six percent (N=216) of the sample were African American, 43% (N=166) were Caucasian, and 1% (N=4) were of another race. Thirteen percent (N=50) self-identified as being of Hispanic ethnicity, and 87% did not. Thirty-seven percent (N=143) had less than a high school education, 20% (N=77) had been homeless during the past year, and 20% (N=77) held some type of employment. Primary diagnoses of the sample included bipolar disorder (28%, N=109), major depression (28%, N=108), schizophrenia spectrum disorders (22%, N=86), PTSD (6%, N=23), substance use disorders (5%, N=18), and other diagnoses (11%, N=42).
Measures
An interview protocol developed for purposes of program evaluation of SAMHSA’s TCE Jail Diversion Initiative was administered to all participants. The interview included Government Performance Reporting Act questions on demographic characteristics (gender, race, ethnicity, and date of birth), education level, employment, and alcohol and drug use. Questions on alcohol use assessed the number of days “any alcohol (beer, wine, liquor)” was consumed and the number of days of use of “alcohol to intoxication (5+ drinks in one setting)” in the past 30 days (before arrest or jail detention). For drug use, respondents were asked the number of days in the past 30 they used “illegal drugs.” The number of reported days of drinking to intoxication and using illegal drugs was used to construct a drinking-to-intoxication variable and a drug-use variable with three levels: none, some, and heavy.
Participants were asked dichotomous (yes-no) questions about the experience of a range of potentially traumatic events, including witnessing events with serious injury or death, being in a serious accident, being physically assaulted, and other events involving threat. In this study we chose to focus on sexual assault given its high prevalence and high risk for PTSD development. The item regarding forced sexual experiences (“At any time in your life, has anyone forced you to have sex when you did not want to?”) was used for the indicator variable sexual assault in this study.
The PTSD Checklist (PCL) was used to assess the severity of symptoms of PTSD (
28). The PCL is a 17-item self-report measure of PTSD symptoms based on
DSM-IV criteria, with a 5-point Likert scale response format. It has been found to be highly correlated with the Clinician-Administered PTSD Scale (CAPS), considered the gold standard for measuring PTSD (r=.929); to have good diagnostic efficiency (>.70); and to have robust psychometric properties with a variety of trauma populations (
29,
30). Scores on the PCL range from 17 to 85, and higher scores indicate greater symptom severity. Recent data also provide evidence supporting the reliability and validity of this instrument with individuals with serious mental illness (
6,
31,
32).
Procedures
Seven jail diversion programs funded in 2006 and 2007 through SAMHSA’s TCE Jail Diversion Initiative were included in the study. Funding provided over three years was used to develop the diversion programs and establish links across service systems to facilitate access to community-based treatment for diverted individuals. Participation in jail diversion meant that individuals agreed to receive services for a minimum of six months. Services received by participants varied somewhat, depending on the site, but most provided services such as medication management, case management, assertive community treatment, integrated mental health and substance abuse treatment, psychiatric rehabilitation, and gender-based trauma services. Participants at each of the sites were eligible for diversion if they were identified in jail as having a DSM-IV axis I diagnosis. Most programs targeted persons who had nonviolent misdemeanor offenses, although individuals with felonies or violent charges were allowed. Data were collected from March 2007 through January 2010. Baseline interviews were conducted with 402 individuals enrolled in the postbooking jail diversion programs. Participants were interviewed by trained independent interviewers at baseline and six months and 12 months after the baseline interview. Data from the baseline interviews are reported here. The project obtained a federal Certificate of Confidentiality, informed consent was obtained from participants, and study protocols were approved by institutional review boards at each of the seven sites.
Statistical analyses
Complete data on PTSD, lifetime history of sexual assault, gender, and age were available for 386 participants for the analysis. PTSD scores were treated as continuous in all models. As a parsimonious model for the nonlinear effects of age, we dichotomized age as ≥35 versus <35. The alcohol variable was categorized into three groups because of its highly skewed distribution with a significant excess of zeros relative to the Poisson distribution. The categories were no use of alcohol that involved intoxication (reference group), drinking to intoxication one to 19 days per month (“some intoxication”), or drinking to intoxication >19 days per month (“heavy intoxication”). For the same reasons as for alcohol use, the drug use variable was similarly categorized as no drug use (reference group), drug use one to 19 days per month (“some drug use”), or drug use >19 days per month (“heavy drug use”).
Data were analyzed with a structural equation framework, jointly modeling pathways to PTSD along with pathways to intoxication or drug use. In separate models for intoxication and drug use (
Figures 1 and
2), a generalized linear mixed model for multinomial data was used, with PTSD, age (dichotomized at 35), and any lifetime sexual assault (yes-no) as predictors, given that the assumptions of the Poisson distribution were not met. Study site was included as a random effect. Continuous PTSD scores were approximately normally distributed and were modeled with a standard mixed-effects model, with study site as a random effect and age (dichotomized at 35) and lifetime history of sexual assault (yes-no) as fixed effects.
We ran separate and combined models for men and women and found that gender was not related to either PTSD or to use of alcohol to intoxication in this study sample, either as a main effect or as a potential effect modifier. As a result, for parsimony we chose to report the combined models. To facilitate model fitting, Bayesian methods in WinBUGS were used, with relatively noninformative priors for all parameters (
33). A sensitivity analysis to prior assumptions was carried out, and results were robust to reasonable (without using highly informative specifications) modifications of the prior distributions.
Results
Descriptive statistics on the covariates in the model are presented in
Table 1. Rates of lifetime sexual assault were high for both men (21%, N=37) and women (67%, N=143), as were rates of PTSD for men (46%, N=80) and women (60%, N=127), consistent with previous studies of individuals with serious mental illness. Before fitting the models, we examined the mean (unadjusted) PCL score by sexual assault status. The mean±SD PCL score for participants with lifetime sexual assault was 57.01±16.36 compared with 45.15±17.98 for those without a history of sexual assault (t=6.74, df=384, p<.001). On the basis of the recommended cutoff score of 50 (
28), 54% (N=207) of all participants met criteria for PTSD; 68% (N=122) of participants with a history of sexual assault were above the cutoff, compared with 41% (N=84) of those without a sexual assault history (χ
2=29.33, df=1, p<.001).
In the full model for drinking to intoxication (
Figure 1), those reporting a history of sexual assault scored on average 12.13 units higher on the PCL than those who did not report such a history (95% confidence interval [CI]=8.59 to 15.65). Age ≥35 was unrelated to PTSD (estimated odds of increase=1.90 points, CI=–1.68 to 5.47). Heavy use of alcohol to intoxication was related to PTSD and age but not to a history of sexual assault. PTSD was not associated with some use of alcohol to intoxication compared with no use (odds ratio [OR]=1.14). However, a ten-point increase on the PCL was associated with 1.42 times the odds of heavy use of alcohol compared with no intoxication-related use. Participants ≥35 had 2.35 times the odds of heavy use versus no use. Age was not associated with some use of alcohol (OR=.63, CI=.38 to 1.04) compared with no use. History of sexual assault was unrelated to odds of some intoxication-level use (OR=.75) or odds of heavy use (OR=.66).
In the full model of drug use (
Figure 2), participants reporting a history of sexual assault scored on average 11.41 (CI=7.90 to 14.93) units higher on the PCL than those who did not report such a history. Age ≥35 was unrelated to PTSD (estimated increase 2.26, CI=–1.30 to 5.80 points). Heavy use of drugs was related to PTSD and age but not to a history of sexual assault. PTSD was not associated with some drug use relative to no drug use (OR=1.12). However, a ten-point increase on the PCL was associated with 1.43 times the odds of heavy use relative to no use. Age was unrelated to some drug use compared with no drug use (OR=.63, CI=.36 to 1.10). Participants 35 and older had .49 times the odds of heavy drug use relative to their younger counterparts. History of sexual assault was unrelated to odds of some drug use (OR=1.29) or heavy drug use (OR=1.12).
Discussion
In this study, experiencing a sexual assault was strongly associated with the severity of PTSD symptoms, which in turn was associated with heavy drinking and heavy drug use. These findings suggest that PTSD may be an important mediator of the relationship between lifetime sexual assault and both alcohol and drug misuse among persons with mental illness who are involved in the criminal justice system. Our results are consistent with previous studies that examined PTSD as a mediator of sexual abuse and substance misuse among women in the general population and among female victims of interpersonal violence (
34,
35).
An important implication of these findings is that greater emphasis needs to be placed on integrated treatment models for PTSD and substance abuse, as others have recommended in the literature (
36,
37). Previous research has shown that improvement in PTSD symptom severity can lead to improvements in substance use disorder severity (
22,
23). Given the high prevalence of substance use disorders among jail diversion participants, and the significant role of substance use disorders in criminal justice involvement, efforts to incorporate integrated PTSD and substance use disorder interventions into jail diversion programs are strongly needed.
Consistent with previous studies, rates of sexual assault and PTSD were high in our sample for both men and women, and the relationship between PTSD symptoms and abuse of drugs or alcohol did not vary by gender (
6,
38). Although the effects of interpersonal trauma and PTSD are infrequently considered among people with serious mental illness, most of the attention to this topic has generally focused on women (
39). This study adds to a growing body of research demonstrating gender similarities in the effect of sexual abuse and PTSD on individuals with serious mental illness.
A number of limitations of this study should be kept in mind when interpreting these findings. One important limitation is the cross-sectional design. The findings are consistent with the theory that heavier use of alcohol and drugs results from attempts to cope with PTSD symptoms and are consistent with previous research identifying PTSD as a mediator of trauma and substance abuse among individuals without serious mental illness and populations not involved in the criminal justice system (
34,
35). However, given the cross-sectional nature of the design, our findings cannot rule out the possibility that the association runs in the opposite direction—that is, the effects of drug and alcohol misuse make an individual more susceptible to PTSD symptoms. This possibility seems unlikely, especially because the study did not find a relationship between alcohol and drug use and lifetime sexual assault independent of PTSD severity. Future studies using a prospective design would provide much stronger causal evidence. Our study was also limited by the use of a self-report measure of PTSD, rather than a structured interview such as the CAPS, considered the gold standard of PTSD assessment. Although a structured interview yields better information and ability to determine diagnosis, the psychometric properties of the PCL against the CAPS have been demonstrated to be quite good in studies of people with serious mental illness (
32,
40). Finally, the item assessing sexual assault was behaviorally specific but may not have captured all experiences of sexual abuse, such as childhood molestation.
Conclusions
Despite these limitations, the results of this study provide support for the self-medication hypothesis and the interactive model of trauma and serious mental illness (
7) and suggest that substance misuse among jail diversion participants with sexual assault histories may be explained by an attempt to cope with PTSD symptoms. Greater attention to PTSD screening and treatment for both male and female jail diversion participants is recommended. Integrated treatment models have been recommended to concurrently treat PTSD and substance use disorders and should be considered for the jail diversion population with serious mental illness (
36).
Acknowledgments and disclosures
This work was funded in part by grant K01MH079343-01 from the National Institute of Mental Health to Dr. Cusack.
The authors report no competing interests.