Replicating Outcomes of Survivors Healing From Abuse: Recovery through Exposure (SHARE): A Brief Exposure-Based Group Treatment for Incarcerated Survivors of Sexual Violence

Abstract

Objective:

Survivors Healing from Abuse: Recovery through Exposure (SHARE) is a group treatment for incarcerated women that targets sexual abuse sequelae. Previous research on SHARE found significant reductions in posttraumatic stress disorder (PTSD), depression, and generalized anxiety disorder (GAD) symptoms among completers. However, evidence for SHARE has been limited to open pilot studies conducted in one prison. Therefore, this study sought to replicate the results of previous SHARE studies in a separate facility.

Method:

Thirty-two women incarcerated in a minimum-security prison participated in an open trial of SHARE. Six separate groups were conducted, each consisting of eight 1.5-hr sessions.

Results:

Pre- to posttreatment symptom reductions in PTSD, depression, and GAD were statistically significant with large effect sizes. Moreover, 21–37% of treatment completers evidenced reliable change in their symptom reduction during the course of treatment. Most women who did not evidence reliable change were already below the clinical cutoff on the corresponding symptom measure at pretreatment and remained below by posttreatment.

Conclusions:

These results are consistent with prior studies of SHARE and provide additional support for the positive outcomes of this brief, targeted trauma-focused treatment for incarcerated women.

People who become incarcerated report greatly elevated prevalence of trauma exposure compared with community peers (Poister Tusher & Cook, 2010). Sexual victimization is especially common among women who become incarcerated, with studies using validated assessment tools and large samples finding lifetime prevalence of 56–82% (Karlsson & Zielinski, 2018). Early exposure to sexual violence has even been conceptualized as a pathway to prison for women, possibly through the development of mental health problems such as posttraumatic stress disorder (PTSD), depression, and substance abuse (Lynch et al., 2017; Salisbury & Van Voorhis, 2009). Indeed, a recent study of jail detainees (Lynch et al., 2017) found that mental health diagnoses mediated the relationship between traumatic exposure as a child and/or adult and number of convictions as an adult.

Thus, there is a clear need for trauma-focused treatments that are feasible to deliver in correctional facilities. Survivors Healing from Abuse: Recovery through Exposure (SHARE), is a brief, structured group treatment that was originally developed for incarcerated women who have experienced sexual violence and have a range of commonly associated outcomes such as PTSD, depression, anxiety, anger, and trust issues (see Karlsson, Bridges, Bell, & Petretic, 2014; Karlsson, Zielinski, & Bridges, 2015; Zielinski, Karlsson, & Bridges, 2016; and Method section for more detail). SHARE has been ongoing at a minimum-security women’s prison in the mid-South since January of 2012 (approximately 5 groups/year, for a total of 38 groups as of July 2019). To date, SHARE has been evaluated primarily for its feasibility and acceptability (e.g., Stage 1 of The National Institute on Drug Abuse stage model; Karlsson et al., 2014, 2015; Onken, Blaine, & Battjes, 1997). An initial pilot examined outcomes from 14 treatment completers from three separate groups. Participants had statistically significant declines in depression and generalized anxiety symptoms over the 8-week treatment, with large effect sizes (Karlsson et al., 2014). Moreover, analysis of clinically significant change per the reliable change index (RCI; Jacobson & Truax, 1991) found that 86% of the sample reported significantly less depression and 50% reported significantly less anxiety by posttreatment. Eight participants (57%) recovered on the PTSD measure (e.g., above the clinical cutoff at pretreatment but below at posttreatment). A larger open trial analyzed outcomes from 36 treatment completers who participated in nine separate SHARE groups (Karlsson et al., 2015). Consistent with the initial pilot, participants reported significantly fewer depression and GAD symptoms by posttreatment, with a large effect size. In the second outcome study, the number of PTSD symptoms were also compared from pre- to posttreatment and the result showed statistically significant reduction in symptoms with a large effect size. Analysis of clinically significant change using the RCI revealed that 39% of participants reported significantly less depression and generalized anxiety. Nine women (25%) were recovered on the PTSD measure at posttreatment. Subsequent follow-up analysis revealed that symptom reductions were maintained or further reduced during a 2–6 month follow up period (Karlsson, Zielinski, Calvert, & Bridges, 2016). However, SHARE has not yet been evaluated in other correctional facilities and research seeking to replicate the effects of SHARE outside of the initial correctional facility is warranted.

Therefore, this study sought to replicate the results of previous research on SHARE by examining treatment outcomes in a correctional facility other than the one where SHARE was developed and has been evaluated thus far. Consistent with prior evaluations (Karlsson et al., 2014, 2015), we hypothesized that participants at this second site would report significant declines in symptoms of PTSD, depression, and anxiety from pre- to posttreatment, with declines evidencing large effect sizes. We also hypothesized similar rates of clinical improvement as those obtained in prior research (i.e., that >40% of the sample would show clinically significant improvement in symptoms from pre- to posttreatment).

Method

SHARE Intervention

SHARE’s core components are based on imaginal exposure and cognitive restructuring, techniques commonly included in evidence-based trauma-focused therapy (e.g., Foa, Hembree, & Rothbaum, 2007; Resick & Schnicke, 1993). The therapy is conducted over eight 1.5-hr sessions. The first two sessions focus on building rapport, providing education about sexual violence and its sequela, and preparing participants for imaginal exposure to memories of sexual victimization. Sessions 3–7 then transition to a primary focus on imaginal exposure (Foa et al., 2007). Each session, one or two women share their personal story. Group leaders help participants attend to “hot spots” in the memory and guide women in attending to and labeling trauma-related emotions, sensations, and cognitions. They also challenge trauma-related cognitions using gentle probes. After each participant is finished sharing, she has the option of receiving feedback from the leaders and other group members. If she invites feedback, group members are invited to share supportive reflections and comments. None of the group members are required to give or receive feedback. In Sessions 5–7, psychoeducation on the impact of trauma on safety, trust, power/control, esteem, and intimacy (McCann, Sakheim, & Abrahamson, 1988) is provided and discussed if there is time remaining after exposures are complete. All exposures are completed by the end of Session 7. The final session of SHARE then focuses on relapse prevention and continuing recovery. Compared with other established trauma treatments, SHARE is unique in that there is no required homework between sessions. SHARE is in-tended to be idiographic and, therefore, address a range of trauma-related outcomes, including, but not specific to, PTSD.

Participants and Facility

Thirty-two women voluntarily attended one of six separate rounds of SHARE while incarcerated in a minimum-security work-release center in Kentucky over the 1.5-year period during which data were collected for this study. The center has a maximum capacity of 200 women, all incarcerated for nonviolent felonies (e.g., selling or using illicit drugs, committing financial crimes). Each group included 3–8 women (mode = 6).

Procedure

All procedures were approved by the institutional review board at Murray State University and by the Kentucky Department of Corrections. Each SHARE group was led by three therapists. One therapist was the first author, a licensed clinical psychologist. The other therapists were female graduate students (two per group, with a total of five students across all groups) completing a Master’s degree in clinical psychology. Before the start of the first group of the semester, the therapists would meet for approximately 90 min to prepare for the first session as well as to discuss general expectations (e.g., structure of group and supervision, as well as ethical obligations such as the Prison Rape Elimination Act, 2003). Otherwise, supervision was conducted while driving to and from the prison, which was approximately a 60-min drive from the clinical team’s university. On the way to the prison, the supervision focused on preparing for the session (e.g., clarifying every-one’s role, reviewing therapy techniques, and addressing concerns in the group), and on the way back, the focus was on debriefing the session.

Group participants were recruited via oral announcements made by the group leaders in the center’s dormitories as well as other occupied spaces (e.g., outdoor area, cafeteria). The announcement included a description of the treatment dates, structure and purpose, behaviorally specific examples of sexual violence, and a range of possible areas of difficulties related to the sexual violence. Interested individuals were invited to submit a brief form containing basic information that could impact their ability to participate in the group such as parole date. This form was submitted to one specifically identified facility administrator.

Inclusion/exclusion criteria.

Inclusion criteria were: (a) being a woman incarcerated at the facility (age 18 or older) during the study enrollment period; (b) endorsing lifetime experience of at least one instance of sexual violence (examples provided during recruitment included being touched when you did not want, being threatened if you did not have sex, and being gang raped); (c) self-endorsing having at least one difficulty potentially related to the trauma (examples provided during recruitment included sad-ness, anxiety, anger, or trust issues); and (d) expressing interest in this study. The only exclusion criteria were (a) current participation in the substance abuse program at the facility; and (b) having a parole date that would preclude participation in the 8-week treatment protocol. Women currently enrolled in the facility’s substance use program were excluded from SHARE temporarily, as the facility had a policy that women in the substance use program could not be in any other programs concurrently. However, they were able to participate in SHARE before or after they participated in the substance abuse program. Because the majority of the women at the study facility are incarcerated because of substance use related charges (e.g., selling or using illicit drugs, financial crimes associated with receiving access to substances), most of them are recommended to participate in the substance abuse program while incarcerated.

Measures

Participants completed the following self-report measures during the first and last treatment sessions, in addition to a general demographics form (completed during first session) and assessment of exposure to 10 distinct categories of potentially traumatic events (i.e., the trauma checklist from the Posttraumatic Diagnostic Scale, PDS; Foa, Cashman, Jaycox, & Perry, 1997; completed during the last session; see online supplemental material).

Posttraumatic stress disorder (PTSD).

The 20-item PTSD Checklist for Diagnostic and Statistical Manual for Mental Disorders-Fifth Edition (DSM–5; PCL-5; Weathers et al., 2013) was used to assess PTSD symptoms over the past week at pre- and posttreatment. Response options are used to indicate the severity of each PTSD symptom and range from 0 (not at all) to 4 (extremely). Previous research has suggested a clinical cutoff score of 33, resulting in 93% sensitivity and 72% specificity for DSM–5 diagnosis of PTSD (Wortmann et al., 2016). Internal consistency reliabilities in the study sample were .94 at both assessment points.

Depression.

The 9-item Patient Health Questionnaire (PHQ-9; Kroenke, Spitzer, & Williams, 2001) was used to assess for depressive symptoms over the past week at pre- and posttreatment. Response options range from 0 (not at all) to 3 (nearly every day); higher sum scores indicate greater symptom severity. Previous research suggests a clinical cutoff score of 10 (88% sensitivity and specificity for diagnosing major depressive disorder; Kroenke et al., 2001). Internal consistency reliabilities in the study sample were .90 and .83 for the pretreatment and posttreatment assessments, respectively.

Generalized anxiety disorder (GAD).

The two-item Generalized Anxiety Disorder Scale (GAD-2; Kroenke, Spitzer, Williams, Monahan, & Löwe, 2007) was used to assess participants’ symptoms of general anxiety over the past 2 weeks at pre- and posttreatment. Response options range from 0 (not at all) to 3 (nearly every day); higher sum scores indicate greater symptom severity. Prior research recommended a clinical cutoff score of 3 (86% sensitivity and 83% specificity for diagnosing GAD; Kroenke et al., 2007). Correlations between the two items comprising the GAD-2 at pre- and posttreatment were r = .61 (pre) and r = .90 (post) in this sample.

Analytic Approach

Statistically significant change was determined using paired samples t tests comparing pre- and posttreatment levels of PTSD, depressive, and GAD symptoms. These analyses were conducted with the intent-to-treat sample as well as with treatment completers who provided pre- and posttreatment data. Missing values for the intent-to-treat analyses were imputed using last observation carried forward method (LOCF; Salim, Mackinnon, Christensen, & Griffiths, 2008).

Clinically significant change was determined by calculating effect sizes and improvement rates. Effect sizes for the paired samples t tests were determined using eta squared calculations; interpretations of these sizes were consistent with Cohen’s (1988) guidelines—that is, small effect size = .01, medium = .06, and large = .14. Improvement rates were determined via the reliable change index (RCI; Jacobson & Truax, 1991; see online supplemental material for more detailed description).

Results

Of the 32 group participants, 24 women completed treatment (i.e., attended at least one of the first two sessions and completed an imaginal exposure) and 19 both completed treatment and completed research measures at both assessment points. There were no pretreatment differences in demographics or symptom levels between completers with research measures at both time points and noncompleters or completers with research data at only one timepoint (i.e., online supplemental material Table 1s). In the intent-to-treat sample, participants were 81.3% White (1 identified as Black, 1 Pacific Islander, and 2 as Bi- or Multiracial), 65.6% nonmarried, with an average age of 36.72 (SD = 7.25) and average of 2.57 (SD = 1.33) children. The 20 women who completed the PDS at posttreatment reported experiencing an average of 5.05 (SD = 2.37; range 1–9) traumatic event categories. All treatment completers attended at least six sessions (M = 7.5, SD = 0.69; median and mode = 8 sessions).

Bivariate correlational analyses were conducted for the intent-to-treat sample to test whether any of the demographic variables (i.e., age, number of children, previous therapy experiences, and treatment dose) were significantly associated with pre- and posttreatment symptom levels as well as the change in symptom levels during treatment (i.e., posttreatment symptom score subtracted from pretreatment symptom score). Age was significantly negatively correlated with pre- and/or posttreatment symptom levels of PTSD, depression, and GAD (rs = −.56 to −.35, ps < .05), but it was not significantly correlated with change in symptom levels (rs = −.02 to .04, ps > .86). The other variables were not significantly associated with symptom levels or symptom change.

Posttraumatic Stress Disorder (PTSD)

Consistent with our hypothesis, paired samples t-tests revealed statistically significant declines in PTSD symptoms from pre- to posttreatment (ps < .001) with large effect sizes, η²s ≥ .40, in both the intent-to-treat sample and the treatment completers with pre- and posttreatment data. See Table 1 for more details about the descriptive and inferential statistics.

Table 1.

Comparing Pre- and Posttreatment Data for Intent-to-Treat Sample and Treatment Completers With Pre and Posttreatment Data

Variables	Pretreatment M (SD)	Posttreatment M (SD)	Test statistica	p-value	Effect sizeb
Intent-to-treat (N = 32)
PTSD (PCL-5)	32.83 (17.12)	23.17 (18.58)	t(29) = 4.21	<.001	.40
Depression (PHQ-9)	9.07 (6.11)	6.73 (4.66)	t(29) = 2.77	.010	.21
Anxiety (GAD-2)	3.43 (1.79)	2.83 (2.02)	t(29) = 2.63	.013	.19
Pre-and posttreatment data (n = 19)
PTSD (PCL-5)	28.11 (16.54)	12.84 (10.72)	t(18) = 5.19	<.001	.60
Depression (PHQ-9)	8.68 (6.82)	5.00 (3.62)	t(18) = 2.98	.008	.33
Anxiety (GAD-2)	3.00 (1.80)	2.05 (1.84)	t(18) = 2.81	.012	.30

Note. N = total sample size; n = number of participants in subsample; PTSD = posttraumatic stress disorder; GAD = generalized anxiety disorder; PCL-5 = PTSD Checklist for Diagnostic and Statistical Manual for Mental Disorders-Fifth Edition; PHQ-9 = 9-item Patient Health Questionnaire.

^aTest statistics compares pre- and posttreatment data for the various subdivision of the total sample.

^bEta squared calculations; interpretations of these sizes were consistent with Cohen’s (1988) guidelines (i.e., small effect size = .01, medium = .06, and large = .14).

Seven women evidenced significant improvement with RCIs above 1.96, which is 21.9% of the intent-to-treat sample or 36.8% of the treatment completers with pre- and posttreatment data. Out of the 13 women who were above the clinical cutoff for probable PTSD at pretreatment (≥33; Wortmann et al., 2016), 38.5% (n = 5) evidence significant improvement. The majority (n = 11; 57.9%) of the treatment completers with pre- and posttreatment data who did not have significant RCIs were below the clinical cut-off for probable PTSD at both assessment points. One participant (5.3%) remained above the clinical cutoff at both assessment points.

Depression

As expected, depression symptoms also declined significantly from pre- to posttreatment ps ≤ .010, with large effect sizes, η²s ≥ .21. This pattern was evident in both sample analyses (see Table 1 for details).

Seven of the women evidenced significant improvement with RCIs above 1.96, which is 21.9% of the intent-to-treat sample or36.8% of the treatment completers with pre- and posttreatment data. Out of the 10 women who were above the clinical cutoff for probable depressive disorder (≥10; Kroenke et al., 2001) at pretreatment, 60.0% (n = 6) evidence significant improvement. The remaining participants with pre- and posttreatment data (n = 12;63.2%) who did not evidence any significant improvement were below the clinical cut-off for probable depressive disorder (≥10; Kroenke et al., 2001) at both assessment points.

GAD

Consistent with our hypothesis, generalized anxiety symptoms significantly declined from pretreatment to posttreatment (ps < .02), with large effect sizes, η²s ≥ .19. As for the depression and PTSD outcomes, this pattern was evident in both analyses (see Table 1 for details).

Four women evidenced significant improvement with RCIs above 1.96, which is 12.5% of the intent-to-treat sample or 21.1% of the treatment completers with pre- and posttreatment data. Out of the 18 women who were above the clinical cutoff for probable GAD (≥3; Kroenke et al., 2007) at pretreatment, 22.2% (n = 4) evidence significant improvement. As for the other two outcomes measures, the majority of participants who did not evidence significant improvement were below the clinical cutoff for probable GAD at both assessment points (n = 10; 52.6% of those with pre- and posttreatment data). Four participants (21.1%) remained above the clinical cutoff at both assessment points and one participant (5.3%) who was above the clinical cutoff at pretreatment was below at posttreatment.

Discussion

SHARE is a group treatment program for helping women recover from the effects of sexual violence while incarcerated. In this open trial replication study, we examined SHARE’s effects in a separate facility in a different state than where it was developed. Consistent with past research, participants reported significant and large reductions in symptoms of PTSD, depression, and generalized anxiety after completing SHARE. We found support for these results in both the intent-to-treat sample and with treatment completers who provided pre- and posttreatment data.

Rates of clinically significant change, as determined by the RCI, were somewhat lower than expected. In prior studies, at least 40% of the total sample of SHARE participants had reliable symptom reductions (Karlsson et al., 2014, 2015). In this study, these rates ranged from 12.5–60.0%, with the lowest percentages for the intent-to-treat sample. It could be that the treatment was less effective in the new facility. On the other hand, findings could be because of sample differences. Women in this study were less symptomatic at pretreatment than women in prior studies of SHARE, which made it more difficult for them to experience a comparable magnitude of score decreases that would be required for changes to be considered reliable. If improvement rates are calculated from only those women above the clinical cutoff at pretreatment, results align with those in past research for the PTSD and the depression outcomes (i.e., 39–60% improvement). Together, the results of this study not only strengthen the evidence for the feasibility and acceptability of SHARE but provide further evidence that incarcerated women—who typically have histories of chronic trauma exposure and mental illnesses—can benefit from brief evidence-informed trauma treatment.

Results of this study should be considered preliminary evidence of replication of SHARE treatment outcomes, as they are based on a relatively small sample of incarcerated women with limited racial and ethnic diversity, who were all incarcerated in a minimum-security facility. Evaluating SHARE in a more diverse array of facilities and participant samples would further bolster evidence of its feasibility. Additionally, the program was still delivered, supervised, and evaluated by a team that included its developers. Studies by independent research teams are needed to control for researcher allegiance, which previous research has shown accounts for a portion of the observed effects in treatment studies (Munder Brütsch, Leonhart, Gerger, & Barth, 2013). Related to this, although one of the developers was directly involved in both projects, and that would suggest attention to protocol integrity, there was no formal assessment of treatment adherence and competence. Our results also relied fully on self-report responses to relatively brief measures answered privately—measurement limitations that should be considered when interpreting our findings. There was evidence that at least one woman had difficulty relating to the trauma assessment items, had difficulty sustaining attention, and/or responded randomly; she did not endorse any of the questions assessing exposure to sexual violence (three separate items) and yet shared about her sexual trauma while completing SHARE. Participants’ trauma history was assessed at the end of treatment and, therefore, this information was missing for those who did not complete the postassessment. This limited the ability to compare treatment completers and noncompleters in terms of their exposure to traumatic events. Participants were also not directed to answer the PCL-5 while considering a single index trauma, such as the sexual assault that they shared about in group, a deviation from the measure’s typical administration. The impact of this methodological deviation is unknown. Moreover, the GAD-2 contains only two questions about participants’ level of anxiety and worry. This measure also evidenced the lowest percentage of reliable improvement, possibly because of the limited number of items and low range of symptom severity scores possible.

Still lacking in the evaluations of SHARE are comparisons to other bona fide treatments or the inclusion of other control groups, which would provide stronger evidence for treatment efficacy and effectiveness. Future studies would benefit from randomly assigning participants into the SHARE group and one or more meaningful comparison groups (e.g., Seeking Safety, which is also being implemented at correctional facilities for women; Najavits, 2002). Such a study would allow for comparison between SHARE’s trauma-focused treatment approach, designed to target mechanisms known to maintain trauma-related pathology such as avoidance behaviors, and a trauma-informed treatment approach, designed to educate participants about the interplay between trauma and substance use without any active treatment components (e.g., primarily a nonspecific treatment that is informed by research on trauma and substance use).

Summary and Implications

Results of this study should be considered preliminary evidence of replication of SHARE treatment outcomes, in an uncontrolled trial. Although additional studies are needed, this small-scale study suggests that SHARE can be efficacious with a different sample, in a different location and facility from where the treatment was developed. Given the clear and compelling evidence that incarcerated women need treatment for lingering sequela of sexual victimization and other trauma (Karlsson & Zielinski, 2018), further research on groups such as SHARE is critical to developing evidence-based practices that are specific to the needs of this underserved population.

Clinical Impact Statement.

This study shows that the outcomes from Survivors Healing from Abuse: Recovery through Exposure (SHARE), a therapy group designed for incarcerated women who have been sexually victimized, were replicated in a facility outside where the treatment was originally developed. This open trial adds to the literature indicating SHARE’s promise. Future randomized controlled trials should be conducted to determine SHARE’s efficacy.