PTSD Screening Among Incarcerated Men

Trauma exposure is overrepresented in incarcerated male populations (Wolff, Huening, Shi, & Frueh, 2014). Surveys of trauma exposure among incarcerated men have found rates of trauma exposure ranging from 62.4% to 100% (Gibson et al., 1999; Saxon et al., 2001; Wolff et al., 2014; Wolff & Shi, 2009; Wolff, Shi, Blitz, & Siegel, 2007), compared with rates of 43% to 92% for community-based male populations (Breslau, 2009; Teplin, McClelland, Abram, & Weiner, 2005). Moreover, incarcerated men, compared with men residing in the community, are more likely to report experiencing physical and sexual assaultive violence. For example, the lifetime rate of assaultive violence for a community sample of men is estimated at 43.3% (Breslau, Chilcoat, Kessler, Peterson, & Lucia, 1999), less than half the rate of 96.5% estimated for incarcerated men (Wolff et al., 2014). The difference in lifetime rates of sexual trauma between male community and incarcerated populations is even more striking. The lifetime prevalence of sexual assault reported by incarcerated men is estimated at 15% to 16% (Saxon et al., 2001; Wolff et al., 2014), compared with rates of 1% to 3% for adult male populations (Breslau et al., 1999; Lukaschek et al., 2013).

Trauma exposure is linked to psychiatric morbidity, particularly posttraumatic stress disorder (PTSD). Based on community samples, 5% of men have experienced PTSD, representing approximately 7% to 14% of those who have experienced a severe traumatic event (Breslau, 2009). By contrast, for incarcerated men, roughly 33% have severe PTSD symptoms, increasing to 60% having moderate to severe PTSD symptoms (Wolff et al., 2014). Depending on type of trauma exposure, 30% to 60% of incarcerated men exposed to a physically violent traumatic event manifest moderate to severe symptoms of PTSD, increasing to 43% to 75% for those exposed to sexual assault. Rates are higher yet for incarcerated men with other co-occurring mental disorders (Wolff et al., 2014).

The high rates of trauma and PTSD among incarcerated men are less surprising than the lack of screening and treatment for these conditions in incarcerated settings. To date, most of the expanding but limited attention on trauma and PTSD in corrections has focused on women (Federal Partners Committee on Women and Trauma, 2013), even though men comprise 93% of the incarcerated population (Carson & Sabol, 2012). Only a minority of male inmates receive treatment for behavioral health disorders while incarcerated (Ditton, 1999; Peters & Matthews, 2003). Focusing on the male population is critical to the rehabilitation effort of modern corrections, especially in light of the high correlation among trauma, behavioral health problems, and criminality (Wolff & Shi, 2009).

Background

Developing cost-effective screening strategies is essential for population-wide diffusion. Screening must impose minimal fiscal and staff burden to be adopted by departments of corrections that are facing pressures to lower costs. For this reason, we explored the feasibility, reliability, and validity of computer-administered screening for PTSD symptoms among incarcerated men. This modality of screening has been tested using cognitive and non-cognitive survey and screening instruments in health clinics, mental health clinics, and educational settings with psychiatric, low-income, minority, and low computer literacy populations (Campbell et al., 1999; Richman, Kiesler, Weisband, & Drasgow, 1999). There has been no study to our knowledge testing this modality among incarcerated populations, although they share many of the characteristics of populations used in previous studies (e.g., HIV-positive individuals, injecting drug users, people with serious mental illnesses, substance users).

Advantages most frequently identified for computer-administered interviewing (CAI) focus on efficiency and validity. In terms of efficiency, because CAI is administered by computer, minimal staff and training are needed to administer or score the instrument, although there are immediate start-up costs for software and perhaps computers if none are available. Agencies with information management systems may, with minimal investment, integrate CAI with other databases, such as inmate clinical records, to improve case management and treatment planning. Efficiency is also enhanced by the conservation of paper and data handling (e.g., transcription and data entry).

In terms of validity, self-report data collected by CAI may be more complete and accurate, increasing the quality of the data. Administratively, data quality is improved in part because non-response bias is nearly eliminated by computer formatting and sequencing; in part because random and non-random interviewer-bias is eliminated with standardization of question formatting; and in part because with self-pacing, users have more time to reflect as needed prior to answering questions (Chang & Krosnick, 2009; Nicholls, Baker, & Martin, 1997). The accuracy of self-report data is also expected to improve if CAI is perceived as being a more private mode for reporting behaviors that are socially undesirable or stigmatizing (e.g., sexual dysfunction, illegal behavior, risky sexual behavior, sexual or physical abuse, abortion). With CAI, respondents are less motivated to answer questions in more socially desirable or undesirable ways because there is no social pressure or context to motivate distortion. Intentional impression management is minimized to the extent that respondents feel more anonymous (i.e., no one is directly observing and perhaps judging their responses) when answering questions administered by computer. Indeed, there is an extensive literature showing a negative association between anonymity and social desirability distortion, and a positive association between anonymity and self-disclosure (Richman et al., 1999).

Validity, however, may be decreased with computer administration if respondents feel threatened by having identifiable information about them stored in a computer (known as the “big brother syndrome”; Rosenfeld, Booth-Kewley, Edwards, & Thomas, 1996) or misunderstand or misinterpret questions that could be clarified by an interviewer. Clarification of questions by interviewers, however, introduces the possibility of interviewer-bias as interviewers may differ in their explanation of survey questions or provide inappropriate cues in their explanations that could distort responses (Fowler, 1990).

The evidence on the relative quality of data reported by CAI and traditional formats (TF; for example, paper-and-pencil, face-to-face interviews, oral administration) is mixed and depends chiefly on the type of data being collected. A meta-analysis of 28 studies comparing CAI and face-to-face interviewing (FFI) found that there was less social desirability distortion with CAI than with FFI when the survey probed highly sensitive personal information (Richman et al., 1999). In general, respondents tend to more completely and accurately report stigmatizing behaviors (i.e., behaviors that are shaming or embarrassing) with CAI compared with FFI, while the opposite holds for reporting psychological distress (DeLeeuw, Hox, & Kef, 2003; Ghamen, Hutton, Zenilman, Zimba, & Erbelding, 2005; Newman et al., 2002; Nicholls et al., 1997; Richman et al., 1999). Newman et al. (2002) argue that the “impersonal” nature of computers enhances the reporting of information that is stigmatizing but hinders the reporting of information that requires human engagement, such as feelings of depression, where empathy may be required to fully elicit information on feelings. The presence of an interviewer in a face-to-face condition may distort responses: While orally administering survey questions, social expectations may be created directly through subtle or nuanced voice intonation, hesitations, gestures, and facial expressions; indirectly through appearance and body language; as well as by the sense of being directly observed while responding to orally administered questions, reducing privacy (Richman et al., 1999; Sudman & Bradburn, 1974). Overall, research evidence suggests that CAI is a valid and reliable method for collecting sensitive survey and behavioral data, but not data on psychological distress, especially for respondents who are comfortable interacting with computers (Metzger et al., 2000).

Several instruments are available to screen for PTSD symptoms (National Center for PTSD, 2014), with the PTSD Checklist (PCL) being one of the most commonly used screens in studies with samples exposed to motor vehicle accidents, sexual assault, sexual harassment, cancer, natural disasters, and combat (Blanchard, Jones-Alexander, Buckley, & Forneris, 1996; Hoge et al., 2004; Krause, Kaltman, Goodman, & Dutton, 2007; Palmieri & Fitzgerald, 2005; Palmieri, Weathers, Difede, & King, 2007; Shelby, Golden-Kreutz, & Andersen, 2005). It also has been used in studies of trauma and PTSD among people with serious mental illnesses (Mueser et al., 2001; Wolford et al., 2008) and incarcerated veterans (Saxon et al., 2001). The PCL—a 17-item self-report measure of PTSD symptoms based on Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association [APA], 1994) criteria—uses a 5-point Likert-type scale format generating a score ranging from 17 to 85, with higher scores indicating greater symptom severity, to assess presence of current PTSD symptoms (Wilkins, Lang, & Norman, 2011). The psychometric properties of the PCL are robust for internal consistency, test–retest reliability, and convergent validity (Blanchard et al., 1996; Keen, Kutter, Niles, & Krinsley, 2008). Comparative studies of PCL administered by computer and other modalities (e.g., face-to-face interview, paper-and-pencil self-administration) have demonstrated psychometric equivalence (Campbell et al., 1999; Wolford et al., 2008).

No studies have assessed the feasibility, reliability, and validity of using CAI to screen for psychiatric disorders among people who are incarcerated. Standard screening practice for psychiatric conditions in prisons is the FFI conducted by correctional mental health care staff. For this reason, we compare the relative psychometric performance of computer versus oral administration of screening for PTSD among incarcerated men. While the literature suggests that CAI is positively perceived by respondents and yields data quality that is equal to or better than interviewer-involved administration when eliciting information about stigmatizing behavior, it is unclear whether these findings will apply to screening for PTSD among incarcerated men for several reasons. First, incarcerated people do not have access to computers and, for those incarcerated since the 1980s, they have had minimal or no experience using a computer with a mouse device. As such, incarcerated men, particularly older men, may be uncomfortable with computer administration, reducing their willingness to participate in computerized screening. Feelings of discomfort with the technology may also trigger suspicion if computer technology is seen as threatening safety. Not knowing where the information goes and how it will be used after being entered into the computer can cause some people to be distrustful of the alleged privacy and anonymity expected with computer administration (Rosenfeld et al., 1996). Incarcerated people, by virtue of their custody status and environment, are reflexively suspicious, which may make them less candid when responding to the computer-administered survey questions.

Second, CAI may not perform optimally because the PCL instrument is tapping into behaviors that are stigmatizing (e.g., sexual trauma, childhood abuse) and feelings that are distressful. The literature suggests that CAI outperforms interviewer-involved administration when questions probe stigmatizing behaviors but interviewer-involved administration outperforms CAI when questions probe psychological distress. Which effect dominates in the administration of the PCL will depend on how incarcerated men respond to questions that probe both stigmatizing behaviors and psychological distress. To the extent that incarcerated men are more inclined to hide their feelings to protect their vulnerability, they may be more comfortable revealing their feelings and experiences to an impersonal computer rather than in the presence of an interviewer who may arouse evaluation apprehension.

Current Focus

The purpose of this study is to test the feasibility, reliability, and validity of using CAI versus orally administered interview (OAI) to screen for PTSD among incarcerated men. Feasibility is tested by the ability to recruit incarcerated men to complete a computer-administered survey. Test–retest reliability is determined using a 2 × 2 factorial design with random assignment to one of four administration conditions: (a) CAI and CAI, (b) CAI and OAI, (c) OAI and CAI, and (d) OAI and OAI. Validity is assessed by comparing PCL scores on symptom severity with the Clinician-Administered PTSD Scale (CAPS), which is widely acknowledged as the gold-standard measure for PTSD diagnosis (Blake et al., 1990; Weathers & Litz, 1994).

Method

This study screened for PTSD among male residents housed at a high security prison operated by the Pennsylvania Department of Corrections from March to June 2012. The primary focus was to compare CAI screening with OAI screening for PTSD symptoms. The protocols for recruitment and interviewing were approved by the appropriate institutional review boards. All participants signed informed consent forms after the conditions of participation (including confidentiality, duty to inform, privacy, risks, benefits, and right to withdraw or refuse to answer questions) were reviewed with them by research staff. Participants received a calendar for completing the first interview and a reentry manual customized to Pennsylvania for completing the second interview.

Participants

Residents eligible for the screening were 18 years or older and had at least 10 months remaining on their mandatory minimum sentence to be completed at the host facility (to ensure sufficient time to complete the parent study prior to release). Excluded were residents with active psychosis or organic brain impairment (limiting the ability to give informed consent), or currently on or had been on suicide watch in the past 3 months. According to prison administrative records, of the estimated 4,000 residents, 1,887 were eligible for the study. Half of these men were randomly invited to be screened, and 592 (63%) gave written consent and participated in the screening interviews. Those who declined mentioned several reasons for not participating, including not being ready to address trauma issues, expecting to be released or transferred, or scheduling conflicts with other required programs. Of the 592 screened participants, 61 were ineligible for the reliability and validity analysis because they did not meet the inclusion criteria (retest was completed outside the 14-day evaluation period, n = 57; missing retest, n = 2; required reading assistance on the computer, n = 2). Of the 531 eligible cases, the first 100 participants for each modality (CAI–OAI, OAI–CAI, CAI–CAI, OAI–OAI) were selected for analysis to ensure balanced group sizes. The OAI–OAI group only had 96 eligible cases, limiting the analysis to 100 per modality.

Results

On average, the study sample of 396 incarcerated men was 43 years old, African American, at least high school graduates or equivalent, and non-Veterans. Most were serving time for a violent offense and had served on average 15 years in prison since turning 18 (see Table 1). The sub-samples randomly assigned to the different combinations of survey modality did not differ statistically in their demographic characteristics with four exceptions: The OAI–CAI sample was several years older, more likely to be college educated, and spent more time incarcerated since age 18, while the CAI–CAI sample was more likely to have some college, compared with the OAI–OAI sample.

Conclusion

In conclusion, our study indicates that incarcerated men are willing and able to reveal information about their PTSD symptoms whether engaged by a computer or a clinician, and the information reported is equally complete and valid. Computer-administered screening for PTSD using the PCL-C has the potential to reliably and accurately identify incarcerated men who would benefit from evidence-based treatment for PTSD. Using a wide net to identify true positives has merit because of the high prevalence of trauma exposure among incarcerated men and the high correlation among PTSD, substance use, and criminality. If the goal is to prepare incarcerated men for the community, using screening methods for PTSD that are efficient, reliable, and valid merit adoption in an effort to protect the public and improve the welfare of those who are incarcerated.

Click here to read the study in its entirety on PubMed.