Lifetime traumatic events and high-risk behaviors as predictors of PTSD symptoms in people with severe mental illnesses
Research is limited regarding the role of high-risk behaviors, trauma, and posttraumatic stress disorder (PTSD) symptoms in people with severe mental illnesses (SMI).The current survey of 276 community mental health clients diagnosed with either a schizophrenia spectrum disorder or a major mood disorder examined the mediating role of lifetime high-risk behaviors with regard to lifetime traumatic events and PTSD symptom severity measured by the PTSD Symptom Scale-Interview version. Structural equation modeling revealed that lifetime high-risk behaviors had significant direct effects on PTSD symptom severity, and high-risk behaviors partially mediated the relationship between lifetime trauma and PTSD symptom severity. Lifetime trauma had the greatest total effect on PTSD symptoms, followed by primary Axis I diagnosis and lifetime high-risk behaviors. Implications for practice with people with SMI include more thorough assessments of trauma, PTSD, and high-risk behaviors and use of coping skills approaches to reduce high-risk behaviors. Limitations of the study include the cross-sectional nature of the data.
KEY WORDS: high-risk behaviors; major mood disorders; posttraumatic stress disorder; schizophrenia; severe mental illness
**********
People with severe mental illnesses (SMIs), including schizophrenia spectrum disorders (SSDs) and major mood disorders (MMDs), experience higher rates of trauma, such as physical and sexual abuse and exposure to violence, than do people in the general public (Mueser et al., 1998; Resnick, Bond, & Mueser, 2003), and they show rates of posttraumatic stress disorder (PTSD) four to five times greater (between 33% and 43%) than do people in the general population (about 8%) (Mueser et al., 1998; O'Hare, Sherrer, & Shen, 2006). People with SMI are also more likely to engage in high-risk behaviors--such as suicide attempts, substance abuse, drug overdoses, unprotected sex, and self-mutilation--and to demonstrate behaviors associated with more severe psychiatric symptoms, have more frequent hospitalizations, be at increased risk of trauma, and have poorer treatment outcomes (Gearon, Kaltman, Brown, & Bellack, 2003; Mueser et al., 1998; O'Hare et al., 2006). Few research investigations have focused on the relationship of high-risk behaviors to trauma and PTSD. Given that the consequences of high-risk behaviors can be retraumatizing, the current study tests whether high-risk behaviors mediate the relationship between lifetime trauma and PTSD symptom severity.
AIMS OF THE CURRENT STUDY
The relationships among traumatic events, high-risk behaviors, and PTSD remain poorly understood in the SMI population (Butler, Mueser, Sprock, & Braff, 1996; Mueser, Rosenberg, Goodman, & Trumbetta, 2002; O'Hare et al., 2006). Although PTSD symptoms in this population have been shown to be positively related to other measures of emotional distress (O'Hare et al. 2006; Resnick et al., 2003), there is only limited research on the linkages between the frequency of lifetime trauma and PTSD. In addition, the relationship between lifetime trauma and high-risk behaviors is even less understood. Although a recent study of clients, most of whom participated in the current investigation, showed that subjective distress from highrisk behaviors--that is, overdosing on medication, self-mutilation, and suicide attempts--mediated the connection between trauma-related subjective distress and PTSD symptoms (O'Hare et al., 2006), it is not known if high-risk events mediate the relationship between lifetime traumatic events and PTSD symptoms. Discovery of such a link could help improve our understanding of the mediating role of high-risk behaviors with regard to traumatic events and PTSD symptom severity.
Gender is a key factor to consider when examining trauma, high-risk events, and PTSD. Women with serious psychiatric disorders appear to be at greater risk for certain stressors, including sexual abuse and interpersonal violence as adults, and men are more likely to witness serious acts of violence (Goodman et al., 2001; Mueser et al., 1998; Resnick et al., 2003). Although studies have shown no differences in rates of PTSD diagnosis between men and women in the SMI population (Mueser et al., 1998; O'Hare et al., 2006), women do report greater perceived threat from stressful and traumatic events (Olff, Langeland, Draijer, & Gersons, 2007) and greater subjective distress from an array of stressful and traumatic events, including sexual and physical abuse and having experienced an unexpected death of a loved one (O'Hare et al., 2006). Therefore, gender is included as a key moderator in the analysis.
The nature of the relationship between symptoms of SMI and PTSD is also in need of further examination (Butler et al., 1996; Mueser et al., 2001, 2002). Because clients with SMI appear to be more vulnerable to traumatic events throughout their lives, it makes sense that they would also be at greater risk for meeting PTSD criteria. However, the large disparity (ratio of 2:1) in rates of PTSD among those with an MMD compared with those with an SSD, respectively, also demands further study.Mueser et al. (1998) found that 58% of clients with mood disorders and 28% with schizophrenia met PTSD criteria. O'Hare et al. (2006) found comparable results in a separate community mental health sample (50.9% of those with MMDs and 21.8% of those with schizophrenia also met criteria for PTSD). It is expected that diagnosis will add significant explanatory power to the current analysis.
The current study of community mental health clients suffering from SMI examined the relationships among primary Axis I diagnosis, frequency of lifetime traumatic or stressful events, and the mediating role of high-risk behaviors as predictors of PTSD symptom severity. Bivariate analyses were used to test the relationships among gender and diagnosis with lifetime events, high-risk behaviors, and PTSD symptom severity, but structural equation modeling (SEM) was used to examine the effects of these factors on PTSD symptom severity as well as the mediating role of high-risk behaviors.
METHOD
Sample and Procedure
All adult clients in the community support program (18 or older) of a mental health center in southeastern New England were eligible to participate in this face-to-face survey. Of 423 active clients at the time the survey commenced, 273 completed the survey, for a 65% response rate. In addition, of 165 outpatient clients sampled, 81 completed the survey, for a 49.1% response rate. Response rate for the combined sample of 588 adult services clients was 60.2% (n = 354). Two hundred (56.5%) of this initial sample were female. The overwhelming proportion were identified as white (92.4%), with each of the Native American, Asian, Hispanic, and African American groups constituting 2% or less of the sample. (These percentages reflect county racial demographics.) About one-half (49.0%) reported their annual family income as between $0 and $7,999, about one-quarter (26.9%) earned between $8,000 and $15,000, and the rest (24.1%) earned over $15,000. The large majority (69.5%) reported their primary insurance as Medicaid or Medicare, about one-fifth (18.4%) reported private insurance, and the rest reported either "none" (5.9%) or "other." PrimaryAxis I diagnoses applied by staff psychiatrists were as follows: SSDs (41.2%), MMDs (46.2%), adjustment disorders (7.6%), and other disorders (4.8%). Of those with SSDs, the largest individual diagnostic subgroups were schizoaffective disorder (44.6%) and paranoid type (37.2%). All other subtypes constituted no more than 6% of the SSD group. Of those with MMDs, most were diagnosed with major depression (53.7%), followed by bipolar I (29.9%) and bipolar II (14.0%). Axis II diagnoses were summarized as personality disorders (12.7%). Most clients (83.9%) had been hospitalized at least once in their lifetime, and almost all clients (93.8%) were taking prescribed psychiatric medication.
Because the purpose of the current study included a focus on two key diagnostic groups, further analysis began with a subsample of 310 clients with a primary diagnosis of either SSD (54.1%, n = 139) or MMD (45.9%, n = 118). Of these 310 cases, 34 (11%) were dropped because of missing data on variables essential to this study: PTSD scale and high-risk event data. Thus, subsequent analysis for this study was based on 276 cases (56.2%, n = 155 women; 42.8%, n = 118 men; three cases did not report gender). Diagnostically, these clients were divided as follows: SSD (44.6%, n = 123) and MMD (55.4%, n = 153). Women (64.5%, n = 100) were significantly more likely than men (43.2%, n = 118) to be diagnosed with MMD, and men (56.8%, n = 67) were significantly more likely than women (35.5%, n = 55) to be diagnosed with an SSD [[chi square](1, N = 273) = 12.29, p < .01].
This survey, conducted from fall 2002 through winter 2003, was designed to be integrated into routine client care by staff social workers and case managers who had a current working relationship with the study participants and had considerable experience in collecting data as part of routine assessment. Staff members were given additional training in the use of the instruments used in the current study. The questionnaire was designed to take about one client contact hour to complete. However, flexibility was emphasized to accommodate clients who required more time. The research supporting the reliability and validity of self-report in people with SMI is substantial (for example, Drake et al., 1998; O'Hare, Sherrer, LaButti, &: Emrick, 2004; O'Hare et al., 2006). Nevertheless, practitioners were advised not to initiate discussion about the survey if the client was in crisis or was otherwise experiencing inordinate distress. The documents and procedures for this study were approved by the agency's Research Review Committee (accredited by the Joint Commission on the Accreditation of Hospitals) and by the Boston College Institutional Review Board.
Instruments
The Risky Behavior and Stressful Events Scale (RBSES) (O'Hare et al., 2006) is an inventory of 15 items drawn from research related to traumatic events and high-risk behaviors (Mueser et al., 1999, 2002; Norris & Riad, 1997). Items are grouped as follows: stressful or traumatic events (having been physically or sexually abused; saw another harmed or killed; saw significant other harmed or killed, experienced life-threatening accident, witnessed life-threatening accident; experienced natural disaster, life-threatening illness, sudden death of friend or loved one, and homelessness) and high-risk behaviors (suicide attempts, self-mutilation, having run away, having overdosed on medications, and having had unprotected sex with an unfamiliar partner). Interviewers asked clients to estimate how many times each event occurred in their lifetime. The RBSES has been validated with virtually the same sample surveyed for the current study (see O'Hare et al., 2006). Exploratory factor analysis (EFA) with the traumatic and high-risk event items measured by client self-reported intensity of distress (that is, "none" to "extreme") resulted in excellent factorial validity, with high factor loadings (.50 to .80) and good separation between factors. Internal consistency reliability for the subscales was .72 and .80 for traumatic and high-risk events, respectively. Concurrent validity was also supported in that subjective distress from trauma and high-risk behaviors were shown to significantly predict PTSD symptom severity. Self-report of traumatic events is well validated in the SMI population (for example, Mueser et al., 1998).
The PTSD Symptom Scale-Interview version (PSS-I), originally developed with female sexual assault victims (Foa, Riggs, Dancu, & Rothbaum, 1993; Foa, Riggs, & Gurshuny, 1995), comprises 17 items that correspond to Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev.) (American Psychiatric Association, 1987) criteria for PTSD, and has demonstrated good internal consistency, excellent interrater and test-re-retest reliability, and good concurrent validity with the Clinician-Administered PTSD Scale (Foa, Cashman, Jaycox, & Perry, 1997; Foa & Tolin, 2000). Recently, the PSS-I was tested using exploratory and confirmatory factor analysis (CFA) with the current sample of clients with SMI (O'Hare, Shen, & Sherrer, 2007). Results showed excellent internal consistency (alphas in the high .80s) for all three subscales and strong evidence of concurrent validity with measures of client-reported subjective distress from both traumatic events and high-risk behaviors. Two strong factors emerged in the CFA supporting reexperiencing and avoidance subscales, the two subscales used in the current study. Although the three-factor model of PTSD (reexperiencing, avoidance-numbing, and hyperarousal) remains the clinical standard, previous research has revealed a range of factor solutions (for example, Palmieri & Hobfoll, 2005).The construct validity of PTSD remains a subject of lively debate in the empirical literature (for example, Weathers & Keane, 2007).
A previously reported EFA (O'Hare et al., 2007) resulted in the following two-factor solution (factor loadings are in parentheses): factor 1--item 8, "unable to remember" (.66), item 9, "less interest in activities" (.77), item 10, "feeling cut-off from people" (.80), item 11, "emotionally numb" (.74), and item 12, "feeling that plans, hopes will not come true" (.77); factor 2--item 1, "upsetting thoughts or images" (.81), item 2, "bad dreams, nightmares" (.84), and item 3, "reliving the trauma" (.85). Factor 1 (avoidance) explained 37.9% of the variance, and factor 2 (reexperiencing) accounted for 31.4% of the variance, for a total of 69.3%. Cronbach's alpha for factor 1 revealed a much lower interitem correlation for item 8. Thus, it was dropped, resulting in a four-item scale (items 9 to 12) and a Cronbach's alpha of .88. Analysis of factor 2 resulted in an alpha of .84. Thus, that initial analysis suggested that this two-factor version of the PSS-I reliably measured two dimensions of PTSD in this sample: avoidance and reexperiencing, factors used in the current study. CFA strongly validated this factor structure (see O'Hare et al., 2007).
The 17 items on the PSS-I are scored 0 to 3, for a possible total score ranging from 0 to 51. Each of the 17 items is scored according to the following scale: 0 = not at all, 1 = once per week/a little, 2 = two or four times per week/somewhat, and 3 = five or more times per week/very much. The interviewer first asks the Client (on the basis of the client's responses to the RBSES) how often, during the past seven days, the client experienced particular symptoms in reaction to what the client considers to be the "most traumatic event" that has ever occurred to him or her. Subscale scores for reexperiencing, avoidance, and arousal are calculated by simple summation.
Gender (male = 1, female = 2) and diagnosis (SSD = 1, MMD = 2) are also included in the SEM analysis.
RESULTS
Bivariate Analysis of Key Study Variables
By their very nature, traumatic and high-risk behaviors, even in low frequencies, are considered clinically significant. Because frequency distributions showed a strong positive skew for all 15 items, they were recoded into three ordinal levels--"never," "one time," and "multiple times"--to make bivariate analysis more interpretable. Frequencies for all lifetime traumatic and high-risk event items appear in Table 1 for the total sample, by gender and diagnostic category. Chi-square analyses (including gamma) showed that women reported having been abused physically more frequently than men [[chi square](2, N = 276) = 27.7, p < .01; [gamma] = .51, p < .01] and sexually abused significantly more frequently than men [[chi square](2, N = 276) = 20.5, p < .01; [gamma]= .27, p < .01].Women reported greater frequencies of having run away [[chi square](2, N = 276) = 6.1; p < .05; [gamma] = .15, p > .05], although the magnitude of the difference was weak (as indicated by the nonsignificant gamma correlation).Women were also more likely to have overdosed on medication, with multiple overdoses occurring more than two and one-half times more often than for men [[chi square](2, N = 276= 17.2, p < .01; [gamma] = .44, p < .01]. Clients with MMDs were significantly more likely than those with SSDs to have been physically abused [[chi square](2, N = 276) = 20.7, p < .01;[gamma] = .43, p < .01], have been sexually abused [[chi square](2, N = 276) = 12.7, p < .01; [gamma] = .35, p < .01), have seen a significant other harmed or killed [[chi square](2, N = 276) = 7.7, p < .05; [gamma] = .33, p < .05], have witnessed a serious accident [[chi square](2, N = 276) = 11.5, p < .01; [gamma] = .36, p <. 01], have experienced a life-threatening illness [[chi square](2, N = 276) = 7.1, p < .05; [gamma] = .27, p < .05], have attempted suicide [[chi square](2, N = 276) = 9.2, p < .05; [gamma] = .22, p < .01], have engaged in self-mutilation [[chi square](2, N = 276) = 7.9, p < .05; [gamma] = .40, p < .01], have run away and been unaccounted for [[chi square](2, N = 276) = 7.6, p < .05; [gamma] = .27, p < .05], and have overdosed on medication [[chi square](2, N = 276) = 11.3;p < .01; [gamma] = .35, p < .01].
Subscales for the 10 trauma items and the five high-risk items were summed into two cumulative indexes (total lifetime trauma and total high-risk events). The summing of these items mitigated the non-normal distributions for the total scores somewhat. Nevertheless, positive skew for these two subscales exceeded an absolute value of 2. Thus, Mann-Whitney U was selected to test for differences in total self-reported traumatic events and high-risk behaviors by gender and diagnosis. Results revealed that women (mean rank = 146.65) reported more lifetime traumatic events than men (mean rank = 124.32) (U = 7,649, p < .05), and women (mean rank = 147.57) reported significantly more high-risk behaviors than men (mean rank = 123.11) (U = 7,506, p < .05). In addition, clients with MMDs (mean rank = 156.73) reported significantly more lifetime traumatic events than clients with SSD (mean rank = 115.83) (U = 6,621, p < .01), and clients with MMDs (mean rank = 153.87) reported significantly more high-risk behaviors than did clients with SSDs (mean rank = 119.38) (U = 7,058, p < .01).
The distribution for total PTSD symptoms showed a moderate positive skew (.99), well within the threshold needed for conducting parametric tests. Independent t tests were conducted to examine differences by gender and diagnosis. The slightly greater reported symptoms for women were not statistically significant [men: n = 118, M = 9.70, SD = 12.98; women: n = 155, M= 12.80, SD = 13.07; t(271) = 1.90, p >. 05]. However, clients with MMDs reported significantly higher PTSD symptoms than clients with SSDs [MMDs: n = 153, M = 15.50, SD = 14.30; SSD: n = 123, M = 6.79, SD = 9.61; t(266.1) = -6.03;p <.01].
SEM
SEM was designed in two stages: The first stage was used to test whether high-risk events mediated the relationship between lifetime traumatic events and PTSD symptom severity; in the second stage, gender and primary mental health diagnosis (that is, SSDs and MMDs) were added to test their influence on the model. Given the excessive skew on lifetime traumatic events and high-risk behaviors, these variables were transformed by taking the cubed root of each factor, an approach that resulted in more normal distributions.
To test for the mediating role of high-risk behaviors, first, a simple SEM model including lifetime traumatic events and PTSD symptoms was used to measure the direct effect (standardized beta) of trauma on PTSD symptoms, showing a significant result (B = .39, p < .01). Next, the variable "total lifetime high-risk events" was introduced into the model as a mediator. Mediators are intervening variables that change the relationship between two other variables. If a variable mediates, it reduces the amount of variance explained between the two main variables--in this case, trauma and PTSD. A fully mediated relationship (relatively rare in the social sciences) means that the significant correlation between the two main variables would be reduced to a nonsignificant correlation because of the mediating effect of the third variable (high-risk behaviors). If partially mediated (a condition that is more common), the correlation between the two main variables would be reduced substantially by the mediator (Baron & Kenny, 1986; MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002).To test the possible mediating effect of high-risk behaviors on trauma and PTSD symptoms, paths were created from lifetime trauma to high-risk behaviors and from high-risk behaviors to PTSD symptoms, as depicted in the conceptual model in Figure 1. If high-risk behaviors partially mediate the relationship between trauma and PTSD symptoms, the original beta (that is, B = .39) should be reduced substantially.When high-risk events was introduced as a mediator, the standardized beta between trauma and PTSD symptoms was reduced substantially (B = .25, p <.01), demonstrating a relatively robust partial mediation. The amount of variance in PTSD symptoms explained by traumatic events was reduced from 15% to 6% as a result of the mediating effects of high-risk behaviors. Thus, the hypothesis that high-risk behaviors significantly mediate the relationship between lifetime traumatic events and PTSD symptom severity was supported.
[FIGURE 1 OMITTED]
To this mediating model were then added gender and primary diagnosis, shown in the foregoing bivariate analyses to be related significantly to both trauma and high-risk behaviors. This mixed measurement and path model (N = 276) was recursive and identified with four observed variables (gender, diagnosis, lifetime traumatic events, and lifetime high-risk behaviors), three latent variables (total PTSD symptoms, reexperiencing, and avoidance), and six residual terms. Gender was included as an exogenous variable, and diagnosis as an endogenous variable. Maximum likelihood estimation procedure was used. The SEM model in Figure 2 includes standardized regression weights. Overall, the model demonstrated a high degree of congruence with the data. All seven fit indices were well within acceptable limits [[chi square](39, N = 276) = 38.0, p >.05]; normed fit index = 97, relative fit index = .95, incremental fit index = .99, Tucker-Lewis index = .99, comparative fit index = .99, and root mean square error of approximation = .01. Diagnosis showed a moderate and significant direct effect on PTSD symptom severity (B = .22, p <.01), as did high-risk behaviors (B = .20, p < .01). The direct effect of lifetime traumatic events on lifetime high-risk behaviors was also significant and large (B = .54, p < .01). Gender had no significant effect on lifetime traumatic events or high-risk behaviors but was significantly related to primary diagnosis (B = .22, p < .01). (A covariance matrix is available from the first author upon request.)
DISCUSSION
The additive and interacting impact of diagnosis, traumatic events, and high-risk behaviors on PTSD symptom severity in this model suggest a complex relationship among these factors. Although gender had a small effect overall, diagnosis showed a moderate significant effect, revealing that those with MMDs had greater PTSD symptoms and greater reports of lifetime trauma. The greater association between mood disorders and PTSD symptoms might be accounted for by overlapping symptom profiles (for example, sleep problems, loss of interest, diminished hope for the future, difficulty concentrating) and the effects of information processing difficulties on the ability of clients with schizophrenia to recall traumatic events accurately (McDevitt-Murphy, Weathers, Adkins, & Daniels, 2005; Mueser et al., 1998; O'Hare et al., 2006). Other evidence using dimensional measures suggests substantial correlation between PTSD and depression as demonstrated with the Personality Assessment Inventory (McDevitt-Murphy et al., 2005) and the CAPS (Foa et al., 1993). In addition, evidence is emerging that people who demonstrate a heightened negative appraisal of traumatic events--a key indicator of depression--are more likely to develop and maintain PTSD symptoms (Ehlers & Clark, 2000; Fairbrother & Rachman, 2006). This phenomenon might hold for people with SMIs as well, particularly for those with MMDs.
[FIGURE 2 OMITTED]
Trauma showed a large and significant direct effect on high-risk behaviors, supporting previous evidence that people with SMI who are traumatized are probably more likely to engage in high-risk behaviors (Mueser et al., 2002; O'Hare et al., 2006). The finding that frequency of lifetime high-risk behaviors mediates the relationship between traumatic events and PTSD symptoms is also important. Theoretical explanations for this mediating effect might include the following: Traumatic events may predispose some clients to engage in high-risk behaviors (for example, sexual abuse can increase the risk of self-mutilation or suicide attempts); the stressful consequences of high-risk behaviors may exacerbate PTSD symptoms; and high-risk behaviors may also increase the likelihood of further trauma (for example, sex with strangers leading to sexual assault, suicide attempts leading to involuntary hospitalization). Rather than a linear view of trauma and PTSD, it is reasonable to consider interactive and reciprocal relationships among trauma, high-risk behaviors, and PTSD symptoms. Although mental health professionals have become more aware of the relationship between mental illnesses and co-occurring conditions such as substance use disorders, further research should emphasize the interacting role of high-risk behaviors in a model that includes early childhood and adult trauma, symptoms of mental illness, substance use disorders, client appraisal of traumatic events, the role of social supports in mitigating the effects of trauma, and the role that client high-risk behaviors play in precipitating traumatic events. A model developed by Mueser et al. (2002) asserts that PTSD influences the course of serious mental illnesses, both directly (a worsening of psychiatric symptoms) and indirectly through retraumatization, substance abuse, and interpersonal problems. Such models need to be further developed with longitudinal designs.
Given the high prevalence rates of trauma and PTSD among SMI community clients, screening for trauma histories and related symptoms, including PTSD, should be conducted as part of routine assessment protocols, using well-validated measures. Training social workers to provide specialized PTSD treatment is another important consideration, especially in the public mental health sector, where trauma exposure in SMI clients is often overlooked (Cusack, Grubaugh, Knapp, & Frueh, 2006). In recent years, promising cognitive--behavioral approaches for individual and group treatment of PTSD in SMI clients have been developed (Mueser et al., 2007; Mueser, Rosenberg, Jankowski, Hamblen, & Descamps, 2004). Data from a randomized controlled trial (Mueser et al., 2008) have suggested that this tailored cognitive-behavioral treatment--a combination of education about PTSD, breathing retraining, cognitive restructuring, and coping skills--can be effective in reducing PTSD symptoms in SMI clients who are not judged to be good candidates for exposure-based treatments that can produce distressing emotional reactions and exacerbate symptoms.
Despite current ambiguities in understanding these complex relationships, practitioners should conduct a detailed assessment that identifies the temporal relationships between multiple lifetime traumatic events and the onset of high-risk behaviors, including substance abuse, suicide attempts, self-mutilation, and other behaviors that may precipitate further trauma. Closer examination of trauma history and high-risk behaviors is likely to result in more accurate assessments and more effective treatments for clients with SMIs (Rosenberg & Mueser, 2008; Sherrer & O'Hare, 2008).
There are a number of limitations to this study. First, cross-sectional studies provide a limited basis for asserting causal relationships among variables, including mediation. Longitudinal studies would shed more light on the ongoing relationship between trauma and high-risk behaviors in this challenging population. Second, although a 60% response rate is very good for a voluntary clinical sample, significant differences with the nonresponders cannot be ruled out. Third, although the application of Axis I diagnoses of major mental illnesses is generally considered reliable, there is no way to test the reliability of psychiatric diagnoses as applied in the current study. Last, generalizations from these data to other groups of people with SMI are limited.
Original manuscript received December 24, 2007
Final revision received April 2, 2009
Accepted May 12, 2009
REFERENCES
American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author.
Baron, R. M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.
Butler, R..W, Mueser, K.T., Sprock, J., & Braff, D. L. (1996). Positive symptoms of psychosis in posttraumatic stress disorder. Biological Psychiatry, 39, 839-844.
Cusack, K.J., Grubaugh, A. L., Knapp, R. G., & Frueh, B. C. (2006). Unrecognized trauma and PTSD among public mental health consumers with chronic and severe mental illness. Community Mental Health Journal, 42, 487-500.
Drake, R. E., McHugo, G.J., Clark, R. E., Teague, G. B., Xie, H., Miles, K., & Ackerson, T. H. (1998). Assertive community treatment for patients with co-occurring severe mental illness and substance use disorder: A clinical trial. American Journal of Orthopsychiatry, 68, 201-215.
Ehlers, A., & Clark, D. M. (2000). A cognitive model of posttraumatic stress disorder. Behavior Research and Therapy, 38, 319-345.
Fairbrother, N., & Rachman, S. (2006). PTSD in victims of sexual assault: Test of a major component of Ehlers-Clark theory. Journal of Behavior Therapy and Experimental Psychiatry, 37, 74-93.
Foa, E. B., Cashman, L., Jaycox, L., & Perry, K. (1997). The validation of a self-report measure of posttraumatic stress disorder: The Posttraumatic Stress Disorder Scale. Psychological Assessment, 9, 445-451.
Foa, E. B., Riggs, D. S., Dancu, C.V., & Rothbaum, B. O. (1993). Reliability and validity of a brief instrument for assessing post-traumatic stress disorder. Journal of Traumatic Stress, 6, 459-473.
Foa, E. B., Riggs, D. S., & Gershuny, B. S. (1995). Arousal, numbing and intrusion: Factor structure of PTSD following assault. American Journal of Psychiatry, 152, 116-120.
Foa, E. B., & Tolin, D. E (2000). Comparison of the PTSD Symptom Scale-Interview version and the Clinician-Administered PTSD Scale. Journal of Traumatic Stress, 13, 181-191.
Gearon, J. S., Kaltman, S. I., Brown, C., & Bellack, A. S. (2003).Traumatic life events and PTSD among women with substance use disorders and schizophrenia. Psychiatric Services, 54, 523-528.
Goodman, L.A., Salyers, M. P., Mueser, K.T., Rosenberg, S. D., Swartz, M., Essock, S. M., Osher, F. C., et al. (2001). Recent victimization in women and men with severe mental illness: Prevalence and correlates. Journal of Traumatic Stress, 14, 615-632.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83-104.
McDevitt-Murphy, M. E., Weathers, F. W., Adkins, J.W., & Daniels, J. B. (2005). Use of the personality assessment inventory in assessment of posttraumatic stress disorder in women. Journal of Psychopathology and Behavioral Assessment, 27, 57-65.
Mueser, K.T., Bolton, E., Carry, P. C., Bradley, M. J., Ahlgren, K. F., DiStaso, D. R., et al. (2007). The trauma recovery group: A cognitive-behavioral program for post-traumatic stress disorder in persons with severe mental illness. Community Mental Health Journal, 43, 281-304.
Mueser, K.T., Goodman, L.A., Trumbetta, S. L., Rosenberg, S. D., Osher, F. C., Vidaver, R., et al. (1998). Trauma and post-traumatic stress disorder in severe mental illness. Journal of Consulting and Clinical Psychology, 66, 493-499.
Mueser, K.T., Rosenberg, S. D., Drake, R. E., Miles, K. M., Wolford, G., Vidaver, R., & Carrieri, K. (1999). Conduct disorder, antisocial personality disorder, and substance abuse in schizophrenia and major affective disorders. Journal of Studies on Alcohol, 60, 278-284.
Mueser, K.T., Rosenberg, S. D., Goodman, L.A., & Trumbetta, S. L. (2002).Trauma, PTSD, and the course of severe mental illness: An interactive model. Schizophrenia Research, 53, 123-143.
Mueser, K.T., Rosenberg, S. D., Jankowski, M. K., Hamblen, J. L., & Descamps, M. (2004). A cognitive-behavioral treatment program for posttraumatic stress disorder in persons with severe mental illness. American Journal of Psychiatric Rehabilitation, 7, 107-146.
Mueser, K.T., Rosenberg, S. D., Xie, H., Jankowski, M. K., Bolton, E. E., Lu, W., et al. (2008).A randomized controlled trial of cognitive-behavioral treatment for posttraumatic stress disorder in severe mental illness. Journal of Consulting and Clinical Psychology, 76, 259-271.
Mueser, K.T., Salyers, M. P., Rosenberg, S. D., Ford, J. D., Fox, L., & Carty, P. (2001). Psychometric evaluation of trauma and post-traumatic stress disorder assessments in persons with severe mental illness. Psychological Assessment, 13, 110-117.
Norris, F. H., & Riad, J. K. (1997). Standardized self-report measures of civilian trauma and post traumatic stress disorder. In J. P. Wilson & T. M. Keane (Eds.), Assessing psychological trauma and PTSD (pp. 7-42). New York: Guilford Press.
O'Hare, T., Shen, C., & Sherrer, M.V. (2007).Validating the Posttraumatic Stress Disorder Symptom Scale with persons who have severe mental illnesses. Research on Social Work Practice, 17, 720-728.
O'Hare, T., Sherrer, M.V., LaButti, A., & Emrick, K. (2004). Validating the Alcohol Use Disorders Identification Test with persons who have a serious mental illness. Research on Social Work Practice, 14, 36-42.
O'Hare, T., Sherrer, M.V., & Shen, C. (2006). Subjective distress from stressful events and high-risk behaviors as predictors of PTSD symptoms severity in clients with severe mental illness. Journal of Traumatic Stress, 19, 375-386.
Olff, M., Langeland, W., Draijer, N., & Gersons, B.P.R. (2007). Gender differences in posttraumatic stress disorder. Psychological Bulletin, 133, 183-204.
Palmieri, P., & Hobfoll, S. (2005, November). Factor structure of post-traumatic stress symptoms in a community sample of physically and sexually abused women. Paper presented at the meeting of the International Society of Traumatic Stress Studies, Toronto, Ontario.
Resnick, S. G., Bond, G. R., & Mueser, K.T. (2003). Trauma and posttraumatic stress disorder in people with schizophrenia. Journal of Abnormal Psychology, 112, 415-423.
Rosenberg, S. D., & Mueser, K.T. (2008). Trauma and post-traumatic stress syndromes. In K.T. Mueser & D.V. Jeste (Eds.), Clinical handbook of schizophrenia (pp. 447-458). New York: Guilford Press.
Sherrer, M.V., & O'Hare, T. (2008). Clinical case management. In K.T. Mueser & D.V. Jeste (Eds.), Clinical handbook of schizophrenia (pp. 309-318). New York: Guilford Press.
Weathers, F.W., & Keane, T. M. (2007). The criterion A problem revisited: Controversies and challenges in defining and measuring psychological trauma. Journal of Traumatic Stress, 20, 107-121.
Thomas O'Hare, PhD, is associate professor, Graduate School of Social Work, Boston College, Chestnut Hill, MA. Margaret V. Sherrer, MSW, is assistant professor, Department of Psychology and Human Services, Lyndon State College, Lyndonville, VT. Send correspondence to Thomas O'Hare, 338 Dolloff Pond Road, Sutton, VT 05867