Restorative Retelling for Violent Death: An Investigation of Treatment Effectiveness, Influencing Factors, and Durability

Abstract

Many adults who have lost a loved one to violent death suffer from depression, posttraumatic stress disorder (PTSD), and complicated grief. Limited research has examined structured group interventions for violent death survivors or characteristics (e.g., types of loss, quality and type of relationship with the deceased) that may impact response to intervention. This records review of 91 survivors examined the effectiveness of Restorative Retelling (RR), a brief structured group intervention for violent loss survivors. Participants completed depression, PTSD, and complicated grief measures at pre- and post-treatment and at 1-year follow-up for a subset of participants. Findings revealed statistically significant changes in depression and PTSD symptoms (Cohen's d values ranged from .33–.46) at post-treatment, with significant changes observed across all domains at 1-year follow-up. Treatment response appeared to be influenced by high distress, gender, and relationship with the deceased. Results imply a large-scale randomized control trial to determine treatment efficacy.

It is estimated that over 50% of Americans have lost a loved one to violent death (homicide, suicide, accidents, or disasters; Kilpatrick et al., 2013). Common psychological sequelae of violent death include posttraumatic stress disorder (PTSD), depression, and complicated grief, disorders that are often co-occurring as well as chronic (e.g., Sung et al., 2011, Zinzow, Rheingold, Hawkins, Saunders, & Kilpatrick, 2009). Evidence-based individual treatments now exist for PTSD, depression, and complicated grief separately. However, group treatments offer a number of potential advantages that individual treatment does not, including support from others who have experienced similar events (Yalom & Lesczc, 2005). The purpose of this study is to examine the effectiveness of restorative retelling (RR), a brief structured group intervention for violent death survivors, among a sample of adults seeking counseling services for bereavement issues at a medical center–affiliated grief clinic.

A few group interventions have been developed for and tested with adult and adolescent survivors of violent death. These groups provide a structured, multifaceted treatment approach that includes at a minimum psychoeducation, coping skills training, and the provision of emotional support from leaders and other members. Indeed, preliminary evidence supports the use of structured group interventions for mental health problems associated with the violent death of a loved one. For example, in a randomized controlled trial of a 12-session, prevention-focused group intervention for parents who had suffered the violent death of a child, the intervention reduced general distress in highly distressed bereaved mothers, but not fathers (Murphy et al., 1998).

While the above-mentioned treatments target particular groups of violent death survivors based on relationship to the deceased or age (e.g., parents, adolescents), RR is a treatment geared toward the more general population of adults (such as siblings, spouses, or adult children of the deceased) who have lost a loved one to a violent death (Rynearson, 2001; Rynearson & Correa, 2008). The treatment model underlying the intervention conceptualizes non-recovery from violent death of a loved one as involving a combination of separation distress and trauma distress. The combination of separation and trauma distress results in a narrative dilemma that, through retelling the death story, is reconstructed into a coherent narrative that restores a sense of autonomy and meaning to survivors. Further, Rynearson (2001) posits that a group format offers survivors the universality of a shared experience, vicarious learning, and increased social support. To achieve this, the 10-session intervention includes distress-management skills (relaxation training), commemoration of the life of the deceased through sharing of positive memories, and traumatic stress symptom reduction via exposure-based drawings of the story of death imagery. Restorative retelling has been shown to be effective in open trials at reducing symptoms of PTSD, depression, and complicated grief among treatment-seeking adults receiving services at a community clinic (Rynearson, Correa, Favell, Saindon, & Prigerson, 2006; Saindon et al., 2014). These studies suggested that the treatment is broadly effective, particularly for people with high baseline distress (Saindon et al.), but the studies did not explore other factors that might moderate RR's effectiveness.

Research on various psychological treatments for bereaved individuals suggests that the effectiveness of grief-related therapies depends on several client-specific characteristics including baseline distress (Murphy et al., 1998; Saindon et al., 2014), time since loss (Allumbaugh & Hoyt, 1999; Currier, Neimeyer, & Berman, 2008), gender (Murphy et al., 1998; Schut, Stroebe, & van den Bout, 1997; Sikkema, Hansen, Kochman, Tate, & DiFranceisco, 2004), and relationship (in terms of both quality and kinship) with the deceased loved one (e.g., Allumbaugh & Hoyt, 1999; Joyce, Ogrodniczuk, Piper, & Sheptycki, 2010; Neimeyer & Currier, 2009). However, many of the findings are mixed in terms of the extent to which these factors influence treatment outcome. Besides mixed findings regarding influencing factors, violent death treatment outcome research is limited, with several initial open trials but few randomized controlled trials. Moreover, little is known at this point about the duration of treatment effects for existing treatment protocols, especially RR.

The current study has three aims: (a) to assess whether previous findings of RR's effectiveness can be replicated in this new sample, (b) to test whether characteristics of the bereaved individuals (baseline symptoms, time since loss, and gender) and their relationships to the deceased (type of relationship and quality of relationship) influence treatment outcome, and (c) to present preliminary findings on the duration of treatment effects at a 1-year follow-up. We hypothesized that RR would reduce symptoms of depression, PTSD, and complicated grief in the overall sample, but that it would be differentially effective in those with high baseline symptoms versus those with low baseline symptoms. Further, we expected that the following variables would be associated with better treatment outcomes: more recent loss, a closer relationship to the deceased, loss of a loved one other than a child, and female gender.

METHODS

Participants

Records of 91 violent death survivors were reviewed. These records included all survivors who had lost a loved one to violent death, were seeking counseling services for bereavement issues, and were provided with the RR group treatment at a medical center–affiliated grief counseling center between 2001 and 2011. Survivor inclusion criteria for the group treatment included (a) adults who experienced a violent death (i.e., homicide, suicide, or accident) and (b) English speaking. Exclusion criteria included significant personality disorders, active drug and=or alcohol abuse, or other active psychosis that would contraindicate group therapy. Of those survivors who were screened and agreed to participate in RR, two had significant missing data at baseline and were excluded from the analyses.

Descriptively, the majority of survivors engaged in RR were female (n = 63; 70.8%), with a mean age of 45.34 years (SD = 12.71). In terms of race/ethnicity, survivors were predominantly European-American (n = 71; 79.8%), with smaller percentages of African Americans (n 3; 3.4%),Hispanics (n = 2; 2.2%), or other racial/ethnic backgrounds (n 5; 5.6%). The majority of the sample was college educated (n = 55; 61.8%).

Sample Characteristics

A total of 89 records were included in the pre-post analyses. Of these survivors, 73 completed treatment and post-assessment. A total of 11 survivors completed the 1-year follow-up assessment. Table 1 presents all 89 participants’ demographic information. No significant differences were noted in demographic and loss characteristics between survivors who completed 1-year follow-up assessments and those who were eligible to complete the follow-up.

TABLE 1.

Participants’ Demographic Characteristics.

Characteristic	Mean or frequency (SD or %)	No. answering question
Means of death		89
Homicide	62 (69.7%)
Suicide	17 (19.1%)
Accident	8 (9.0%)
Multiple/different types	1 (1.1%)
Other/unknown	1 (1.1%)
The deceased person was the survivor's:		89
Parent	10 (11.2%)
Romantic partner	14 (15.7%)
Child	38 (42.7%)
Sibling	14 (15.7%)
Friend	4 (4.5%)
Other relative or multiple losses	9 (10.1%)
Positive relationship with the deceased	8.27 (1.46)	76
Months since loss	21.92 (34.63)	87
Saw the death occur	9 (10.1%)	75

Note. Percentages are based on the full sample (n = 89). Positive relationship is measured on a Likert scale ranging from 1 = least positive to 10 = most positive. Although participants were, on average, 22 months post-loss, median time since loss was 12.37 months.

Treatment Package

RR is a 10-session group therapy developed to treat distress and grief responses of violent death survivors. A free version of the training manual is available via the developers’ Web site (http://www.vdbs.org/html/training-manuals.html). Group size ranges from six to 10 members, and group members are taught various resiliency-building stress reduction techniques such as relaxation strategies and then encouraged in a structured manner to engage in commemorative imagery (sharing the life stories of their loved ones) and death imagery (retelling the violent death story).

The groups were facilitated by a licensed social worker who is also one of the RR developers (FC). This open trial record review study (i.e., no control, and individuals know what treatment they are receiving) explored data collected from a medical center–affiliated grief counseling center.

Measures

The demographic and loss characteristic measure was completed at pre-treatment, and all dependent measures were administered by the counseling center staff at pre-treatment, at post-treatment (last day of group), and at the 1-year follow-up (for a subsample of survivors). The assessment battery required approximately 20–30 minutes to complete and included the following measures.

Demographics And Loss Characteristics

At pre-treatment, demographics and characteristics related to the death were collected, which included survivor gender, age, race=ethnicity, marital status, education, and relationship to the deceased. Information about the deceased was collected including gender, age at time of death, time since loss, type of loss, and whether the survivor witnessed the death or found the deceased.

Relationship Quality

The pre-assessment also included six items assessing level of positivity of the survivor's relationship with the deceased (closeness, peacefulness, ease, supportiveness, compatibility, and attachment). The first five were rated on a Likert scale of 1–10 where 1 is least and 10 is most; attachment was rated on a Likert scale where 1 is least and 5 is most, and it was multiplied by 2 so that it was on roughly the same scale as the first five items. The six scores were then averaged to form a total relationship quality score. Cronbach's alpha for the scale was .85 in this sample. As Table 1 shows, participants generally reported having positive relationships with the deceased.

Complicated Grief Assessment Self-Report (Cga-Sr)

This self-report is based on the Inventory of Complicated Grief (Prigerson et al., 1995), and permits diagnosis of complicated grief as well as a total intensity score (9 = no symptoms, 45 = extreme symptoms). Assessed are Criterion A (separation distress), Criterion B (other symptoms such as difficulty with the following: accepting the death, trusting others, experiencing feelings other than numbness, and moving on, as well as feeling bitter, as though life were meaningless, as though the future holds no purpose, and on edge), Criterion C (impairment in functioning), and Criterion D (duration of symptoms greater than 6 months). In the current sample, the internal consistency of the nine items assessing symptom intensity was α = .84.

Beck Depression Inventory (Bdi)

The 21-item BDI (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) is among the most widely used instruments to measure depressive symptoms. Item scores range from 0 to 3; total scores range from 0 to 63. Higher scores indicate greater symptom severity. The BDI had high internal consistency in the current sample (Cronbach's α = .91).

Impact Of Events Scale–Revised (Ies-R)

The IES-R (Weiss & Marmar, 1997) is a 22-item self-report measure that assesses distress related to the trauma symptom clusters of intrusions, avoidance, and hyperarousal. Items are rated on a 5-point scale from 0 (not at all) to 4 (extremely). In the current sample, the IES-R showed high internal consistency (Cronbach's α = .89).

Death Imagery Scale (Dis)

The DIS (Rynearson & Correa, 2008) is a five-item self-report measure that assesses the presence of five distinct forms of post-loss, death-related thoughts and images, including reenactment (‘‘I experienced a fantasied replay of the dying’’), rescue (‘‘I experienced a fantasy of rescuing the person from dying’’), revenge (‘‘I experienced a fantasy of retaliation for this dying’’), reunion (‘‘I experienced a fantasy of reunion with the deceased family member and=or friend’’), and remorse (‘‘I experienced a fantasy that I should have somehow prevented the dying from happening’’). Items are scored on a 4-point scale ranging from 0 to 3 where higher scores are associated with more frequent death imagery Although no studies have yet established the psychometric properties of the DIS, a recent publication that includes participants from the current study found that the DIS has acceptable reliability (Baddeley et al., in press). In this sample of participants, the DIS demonstrated acceptable reliability (Cronbach's α = .71).

Procedures

According to records and program staff, survivors were initially contacted by outreach or were referred by community service providers or mental health workers. They sought treatment for bereavement and trauma-related distress following the violent death of a close friend or family member. All survivors were assessed in a semistructured, individual interview to provide requisite crisis support while clinically judging the presence of comorbidity and consideration of psychiatric consultation before enrollment in the RR group intervention. After agreeing to RR, each survivor completed the above-mentioned battery of questionnaires before beginning treatment. We should note that different measures of complicated grief were used in the clinic throughout the time period specified in this record review, and the complicated grief assessment completed by most survivors was the CGA-SR. Those survivors who did not complete the CGA-SR before and after treatment were not included in the subset of analyses looking at changes in complicated grief symptoms. Survivors participated in a weekly 2-hour RR group treatment for 10 sessions. To maximize treatment fidelity, all groups were facilitated by a social worker (FC) who followed the group manual developed in collaboration with Dr. Rynearson. Groups averaged six to eight members each. The same measures were repeated at the end of the intervention. Within the past year, program staff started collecting 1-year follow-up measures from survivors. Of the eligible 19 survivors they attempted to contact, 11 completed follow-up measures. The other eight did not complete measures after several reminder calls and e-mails. No significant differences in terms of demographic and loss characteristics were noted between survivors who completed 1-year follow-up assessments and those who did not complete the follow-up. For the purposes of this open trial, records for RR group members during 2001–2011 were de-identified, reviewed, and entered into a SPSS database.

Data Analyses

A series of mixed-model repeated measures ANOVAs were conducted to examine pre-post differences on all primary outcome variables as well as to examine pre-post differences in outcomes while accounting for categorical influencing factors (i.e., baseline symptoms [high versus low distress], gender, relationship status). An advantage of using a mixed-model approach is that these models can accommodate missing data resulting from missed appointments or dropout by estimating parameters based on information available for any given individual with missing data (Gueorguieva & Krystal, 2004). A series of mixed-model regression analyses were also used to examine the effects of influencing factors that were continuous (i.e., relationship quality and time since loss). Because multiple relatives of the deceased were allowed to participate in the RR group (resulting in six dyads and one triad of related individuals in this sample), each mixed-model analysis included a random effect that clustered subjects by deceased loved one to account for some participants reporting the same loss. In four analyses, variance due to the random effect could not be estimated (likely due to the small number of related participants), and random effects were thus removed from the model. Repeated measures ANOVAs were conducted on outcome measures examining change in symptoms over three time periods (pre, post, 1-year follow-up) for those survivors who were offered the chance to participate in and completed the 1-year follow-up assessment.

RESULTS

Treatment Effectiveness

Survivors’ depression and PTSD symptoms both overall and across all three clusters (intrusion, avoidance, and hyperarousal) decreased significantly over the course of treatment from pre- to post-assessment, with effect sizes in the small to medium ranges. Complicated grief symptoms did not change significantly over the course of treatment. Survivors also evidenced an overall decrease in death-related thoughts and images from pre- to post-treatment, with statistically significant decreases in reenactment, F(1, 75.19) 14.50, p<.001, and remorse, F(1, 68.71) = 12.12, p =.001, imagery (see Table 2 for means and standard deviations).

TABLE 2.

Changes in Symptoms of Depression, PTSD, and Complicated Grief from Pre- to Post-Treatment.

	Pre M (SD)	Post M (SD)	F	p	Cohen's d
BDI	20.97 (10.06)	16.74 (9.28)	F(1, 74.59) = 28.03	<.001	.44
IES total	40.05 (14.45)	33.43 (14.34)	F(1, 73.08) = 37.38	<.001	.46
Intrusions	17.47 (5.87)	14.86 (6.07)	F(1, 73.29) = 25.86	<.001	.44
Avoidance	11.84 (5.87)	9.88 (6.05)	F(1, 76.47) = 13.06	.001	.33
Hyperarousal	10.74 (5.22)	8.69 (4.56)	F(1, 72.21) = 29.68	<.001	.42
CGA symptoms	27.68 (7.28)	27.67 (7.07)	F(1, 44.58) = 2.72	.106	.00
DIS total	6.99 (3.77)	5.86 (3.51)	F(1, 71.10) = 13.88	<.001	.31

Note. N = 88 for all mixed-model analyses except for analyses with CGA (n = 58) and DIS (n = 87) as the main outcomes; ns for pre- and post-means vary slightly due to missing data.

Effects of Participant and Relationship Characteristics on Treatment Outcome

Baseline Symptoms

A series of 2 × 2 mixed-model repeated measures ANOVAs were conducted for each outcome examining the impact of baseline severity (above vs. below median) by time (pre- vs. post-treatment). There were significant interaction effects of baseline symptom severity for symptom domains including depression symptoms, F(1, 80.04) = 10.69, p = .002; overall PTSD symptoms, F(1, 77.33) = 5.53, p = .021; avoidance symptoms, F(1, 80.53) = 20.59, p < .001; intrusions, F(1, 77.14) 4.36, p = .040; and hyperarousal, F(1, 73.34) = 9.60, p = .003. There was no statistically significant interaction effect of baseline symptom severity for complicated grief, F(1, 46.47) = 1.65, p = .205. Follow-up pairwise comparisons revealed that those above the median in symptom severity for depression, PTSD symptoms, and all PTSD symptom domains experienced statistically significant improvement from pre- to post-treatment, while those below the median only experienced statistically significant change in overall PTSD symptoms, intrusions, and hyperarousal symptoms (see Table 3). In terms of death-related thoughts and images, a statistically significant interaction was found in terms of overall death-related imagery, F(1, 74.40) = 9.57, p = .003. Follow-up analyses revealed that those above the median in symptom severity experienced statistically significant decreases in all forms of death imagery from pre- to post-treatment, including reenactment, rescue, revenge, reunion, and remorse, while those below the median also experienced decreases in reenactment and rescue imagery (all ps < .05).

TABLE 3.

Changes in Symptoms of Depression, PTSD, and Complicated Grief From Pre- to Post-Treatment in Participants With Baseline Symptoms Below vs. Above the Median.

	Below median		Above median
	Pre M (SD)	Post M (SD)	Pre M (SD)	Post M (SD)
BDI	13.09 (4.77)	11.00 (6.46)	28.90 (7.33)	21.61 (8.58)*
IES total	28.81 (8.73)	24.23 (10.46)*	51.05 (9.22)	41.23 (11.95)*
Intrusions	13.79 (3.72)	11.92 (4.83)*	22.88 (3.44)	18.84 (4.82)*
Avoidance	8.17 (3.48)	7.88 (5.50)	18.14 (3.43)	13.00 (5.55)*
Hyperarousal	7.47 (2.99)	6.16 (3.16)*	15.52 (3.12)	11.82 (3.93)*
CGA symptoms	23.28 (4.75)	22.44 (5.31)	33.42 (3.67)	31.42 (5.96)*
DIS total	4.43 (2.15)	3.95 (2.58)	10.50 (1.85)	8.13 (3.05)*

Note. These results are derived from pairwise comparisons of the dependent variable as part of each separate mixed-model repeated measures ANOVA.

^*p < .05.

Time Since Loss

Baseline time since loss was entered as a variable in a series of mixed-model regression analyses to investigate the impact of time since loss on post-treatment symptom severity after controlling for pre-treatment symptoms. There were no significant main effects of time since loss for depression, overall PTSD symptoms, intrusions, avoidance, hyperarousal, or complicated grief symptoms.

Gender

When gender was entered as a fixed factor in each equation, a statistically significant gender (male vs. female) by time (pre- vs. post-treatment) interaction emerged for complicated grief symptoms, F(1, 43.02) = 9.00, p = .004. Follow-up pairwise comparisons revealed a reduction in complicated grief symptoms for women (p = .003) but not for men. No other gender by time interaction effects were significant for any of the outcome variables.

Quality Of Relationship To Deceased

A series of mixed-model regressions controlling for pre-treatment symptoms showed that relationship quality predicted post-treatment symptoms on multiple symptom scales. Specifically, more positive relationships with the deceased were associated with greater post-treatment symptom severity in terms of complicated grief (t = 2.22, p = .033, 95% confidence interval [CI] = 0.09, 1.89) and overall PTSD symptoms (t = 2.30, p = .025, 95% CI = 0.21, 2.99). Looking at PTSD symptom subscales, relationship quality did not significantly predict intrusive thoughts or avoidance symptoms, but more positive relationships with the deceased were associated with greater post-treatment hyperarousal symptoms (t = 2.44, p = .018, 95% CI = 0.11, 1.12). Relationship quality was not predictive of post-treatment depression symptom severity.

Type Of Death

A series of 2 × 2 mixed-model repeated measures ANOVAs with time as the within-subjects factor and type of death (homicide vs. non-homicide) as the between-subjects factor showed no significant interaction effects of time by type of death for depression symptoms, overall PTSD symptoms, intrusions, avoidance, hyperarousal, or complicated grief symptoms (ps > .10). However, a statistically significant main effect for type of death emerged such that, across time, homicide survivors endorsed more severe overall PTSD symptoms, F(1, 79.50) = 5.00, p = .028; avoidance, F(1, 82.51) than = 5.31 p .024; and hyperarousal symptoms suicide and accident survivors, F(1, 77.45) = 6.09, p = .016.

Type Of Relationship To Deceased

A series of 2 × 2 mixed-model repeated measures ANOVAs with time as the within-subjects factor and type of relationship with the deceased (child vs. non-child) as the between-subjects factor showed no interaction effects of time by type of relationship for depression symptoms, complicated grief symptoms, overall PTSD symptoms, intrusions, or arousal (ps > .05). However, there was an interaction effect for avoidance, F(1, 75.29) = 4.37, p = .040, such that those who lost a child showed greater decreases in avoidance symptoms than those who lost another type of loved one.

Treatment Outcome at 1-year Follow-Up

To assess whether treatment gains were maintained after the end of treatment, repeated measures ANOVAs were run for each of the outcome variables to assess changes from pre- to post-treatment to the 1-year follow-up for the small subset of survivors where 1-year follow-up was assessed. Because of the small sample size for these analyses, we did not cluster survivors by association with the deceased loved one. There were significant, sustained improvements at the 1-year follow-up in terms of depressive symptoms, F(2, 20) = 6.80, p = .006; PTSD, F(2, 20) 12.23, p < .001; and complicated grief, F(2, 20) = 10.87, p = .001, with large effect sizes (Cohen's d values ranged from .97 to 1.21). No statistically significant main effect for time emerged for overall death imagery.

DISCUSSION

Given the unique characteristics of violent death, these survivors are at increased risk of PTSD, depression, and complicated grief (e.g., McDevitt-Murphy, Neimeyer, Burke, Williams, & Lawson, 2012). Therefore, specific interventions tailored for this population warrant development and investigation within the grief and trauma fields to address these mental health difficulties. Several models have been developed to address this need but currently with limited empirical support. This open trial attempted to provide evidence for restorative retelling, a group treatment package for violent death survivors, while taking into account various potential factors that may impact intervention results.

The results were consistent with prior open trials of RR (Saindon et al., 2014), replicating support that RR may be an effective treatment for violent death–related difficulties. Of note, the sample characteristics from this study were different from Saindon et al.'s sample, suggesting generalizability of results to different samples. For example, this sample included more variability in type of traumatic loss. Saindon et al.'s study consisted mostly of homicide survivors, whereas this sample included a wider variety of loss (homicide, suicide, motor vehicle crash). In addition, this sample had a longer time period since loss than Saindon et al.'s sample (18 months vs. 22 months). Despite these differences, results were consistent in that symptoms of posttraumatic stress and depression decreased from pre- to post-treatment.

In addition to examining whether RR would be an effective treatment intervention, this study also tested whether characteristics of the bereaved individuals (baseline symptoms, time since loss, and gender) and their relationships to the deceased (type of relationship and quality of relationship) influence treatment outcome. As hypothesized, results revealed that survivors who reported more distress appeared to have greater symptom reduction with RR. These findings are consistent with the broader literature on bereavement research indicating that more distressed individuals are more responsive to interventions (e.g., Currier et al., 2008). However, treatment was not influenced by time since loss, which is consistent with results from an open trial study examining a behavioral activation and exposure-based brief protocol for complicated grief in older adults (Acierno et al., 2012). Contrary to our hypotheses, survivors who reported having a closer relationship to the deceased had more severe symptoms post-treatment. Contrary to other findings (Boelen, de Keijser, van den Hout, & van den Bout, 2011), those participants in our sample who lost a child made greater improvements on the PTSD avoidance scale. There were no other differences in symptom reduction between those who lost a child and those who did not. Also contrary to other findings, we found no differences in gender outcomes for PTSD or depressive symptoms, although, consistent with other findings, women experienced a greater reduction in complicated grief symptoms than men. It appears RR is not gender specific in response to PTSD and depressive symptoms, which may suggest its generalizability. Moreover, although this open trial sample involved only a small subset of the total sample, results are promising as treatment effects were present at the 1-year follow–up. These findings are encouraging but should be interpreted cautiously given the small sample size and poor follow-up rate.

Limitations should be noted when interpreting the findings of this study. First, this is an open trial of a community sample; therefore, no randomization or control was used to determine causation of results. Improvements may have been due to time or regression to the mean. Second, history and maturation represent threats to the internal validity of the findings. Third, treatment was provided to all treatment seekers despite reports of significant levels of psychopathology on measures, potentially diluting treatment results. Fourth, even though we modeled effects due to association with the deceased, we did not model group-level effects or other potential sources of non-independence of observations. And lastly, follow-up data were only recently collected for a small subsample of participants.

Despite these limitations, our results are promising and provide evidence that RR may be a useful and well-tolerated treatment intervention for distress related to violent death. Open trials are a necessary first step within treatment development and evaluation before large-scale randomized controlled trials (RCTs) are conducted. This study, along with other recent open trials of RR (Rynearson et al., 2006; Saindon et al., 2014), gives credence to the need for a large-scale RCT that could help clinicians and researchers more confidently speak to the overall efficacy of RR compared to the non-directive grief groups that are typically provided in the community. Because adult survivors in this sample, as with other samples in the limited violent death treatment outcome literature (e.g., Murphy et al., 1998), were predominantly Caucasian, efforts should be made to recruit more racially diverse, representative samples of violent loss survivors as part of any future RCT in order to increase the generalizability of any conclusions about the efficacy of RR. Lastly, this study provides further understanding of factors that may relate to treatment outcomes such as time since loss, level of distress, type of death, and relationship with the deceased that should be considered in future RCTs of RR.

Biography

Alyssa A. Rheingold is an associate professor at the National Crime Victims Research and Treatment Center in the Department of Psychiatry and Behavioral Sciences at the Medical University of South Carolina. Her research interests include evidence-based treatment of trauma-related mental health problems and traumatic grief.

Jenna L. Baddeley is a clinical psychologist at the Ralph H Johnson VA Medical Center and a clinical assistant professor at the Medical University of South Carolina. Her clinical and research interests include bereavement, depression, and couples interventions.

Joah L. Williams is a postdoctoral research fellow at the National Crime Victims Research and Treatment Center in the Department of Psychiatry and Behavioral Sciences at the Medical University of South Carolina. His clinical and research interests include topics related to traumatic grief and treatment development and evaluation.

Clara Brown is an undergraduate intern at the National Crime Victims Research and Treatment Center at the Medical University of South Carolina.

Megan M. Wallace is clinic case manager at the National Crime Victims Research and Treatment Center at the Medical University of South Carolina. She is also a candidate for a master's in social work from the University of South Carolina.

Fanny Correa is the clinical director of Separation and Loss Services, Virginia Mason Medical Center, and adjunct professor at Seattle University, Master's in Criminal Justice Program.

Edward K. Rynearson is a clinical psychiatrist from Seattle, Washington, where he founded the section of psychiatry at the Virginia Mason Medical Center. In addition to full-time clinical practice, he has served on the clinical faculty of the University of Washington as a clinical professor of psychiatry.