Implementation of Multifamily Group Treatment for Veterans With Traumatic Brain Injury

Traumatic brain injury (TBI), affecting an estimated 22% of service members wounded in military operations in Iraq and Afghanistan, is often complicated by comorbid posttraumatic stress disorder (PTSD), depression, and other mental health problems (1,2), having a dramatic impact on the lives of the veteran’s family (3–6).

Veterans with TBI may also have deficits in memory and executive functioning that create challenges in household management and lead to friction between family members if not understood and addressed. Yet veterans and family members often lack information about the veteran’s condition and treatment needs. This gap in information and services erodes family resilience, contributing to frustration, family conflict, child distress, and consequent barriers to community reintegration (1,7–9). Although there is a growing evidence base supporting the efficacy of family and couples treatment in reducing PTSD symptoms (10–13), there is no established family-based treatment for veterans with TBI sustained during the Operation Enduring Freedom–Operation Iraqi Freedom (OEF/OIF) era.

This study evaluated the feasibility, acceptability, and initial efficacy of an intervention developed to address this gap in services. The intervention is an adaptation of multifamily group treatment (14), an evidence-based treatment for serious mental illness that uses support, education, and problem solving to reduce symptoms and improve functional outcomes. This method has been used to treat survivors of civilian TBI and family members and has had positive therapeutic effects (15).

Because underutilization of mental health services has been identified as a problem for the OEF/OIF cohort (16,17), we evaluated problems in both implementation and initial efficacy of the intervention. A companion article presents the qualitative findings of a study of postintervention focus groups about the experience and value to participants (18).

In this article and the companion article, we addressed the following implementation issues using a mixed-methods approach: systems issues or stakeholder concerns influencing buy-in at the veteran, family, and provider level; impact of group composition (level of impairment and family relationship) on group participation and retention; and adaptations of content or structure to optimize responsiveness to problems identified by veterans and family members.

We evaluated treatment efficacy by measuring change in behaviors hypothesized to respond to skills training in group problem solving. Specifically, we hypothesized that participation in the multifamily group would result in improved anger management and greater use of social support among veterans and reduced burden and increased empowerment among family members (primary outcomes). On a more exploratory basis, we hypothesized that veterans would report reduced depressive and PTSD symptoms and increased community reintegration and that greater neurocognitive impairment at baseline would be associated with less improvement in treatment outcomes. These outcomes were designated as exploratory because they were targeted less directly by intervention components or were predictors of treatment response.

Methods

Participants

Veterans with a clinical diagnosis of TBI or a positive Defense and Veteran’s Brain Injury (DVBIC) screen (19) for TBI sustained during the OEF/OIF era were identified by their polytrauma physician at the J. J. Peters Veterans Affairs (VA) Medical Center (VAMC) in the Bronx, New York, and through referral by a clinician or caseworker or examination of electronic medical records at the Durham, North Carolina, VAMC. The diagnosis of veterans identified by DVBIC screen was confirmed by the treating clinician.

Because multifamily group treatment is generally conducted as a closed group (all families begin and complete group sessions in tandem), recruitment of each group was completed within two to three months, in order to retain early recruits. Within this relatively short window, clinicians used the methods described above to identify and prescreen potential candidates and approve and refer those they considered appropriate for inclusion in the study.

Study staff contacted each of the referred veterans to arrange a meeting with the veteran and a family member or partner to provide a full description of the study. All veterans and family members who agreed to meet and gave written informed consent were screened for eligibility. The Mini-Mental State Examination (MMSE), which evaluates basic orientation, memory, and attentional processes, identified veterans whose level of cognitive impairment precluded participation. Possible scores range from 0 to 30, with higher scores indicating more intact cognitive functioning. Eligible veterans had a clinically confirmed diagnosis of TBI, at least one family member or significant other who consented to participate in the study, and a MMSE score ≥20. Veterans were excluded if they or the identified family member had an active psychosis or a severe substance use disorder. The study protocol was approved by the J. J. Peters VAMC Institutional Review Board.

Measures

Veterans and family members were evaluated before and immediately after treatment. Measures were included to test the study hypotheses and to characterize participants by using clinical and subclinical measures of distress and dysfunction frequently employed in psychosocial research to inform future work.

Veterans.

Baseline neurocognitive functioning was evaluated by the MMSE and by measures of attention and mental flexibility (Trail Making Test, parts A and B) (20) and memory (California Verbal Learning Test II) (21). The Patient Health Questionnaire (PHQ) (22) evaluated the presence of depressive, anxiety, and alcohol use disorders. The Sydney Psychosocial Reintegration Scale (23) measured change in occupational activity and interpersonal relationships, and the PTSD Checklist (PCL) (24) quantified level and type of posttraumatic stress disorder (PTSD) symptoms.

Caregivers.

The Caregiver Burden Inventory (25) evaluated the developmental, physical, social, emotional, and time burden associated with the patient’s dependence on the caregiver. The family subscale of the Family Empowerment Scale (26) assessed caregiver perceptions of coping efficacy.

Veterans and caregivers.

Three measures were administered to both veterans and family members. The 24-item AX Scale (27) measured anger management, including anger-in (suppression of angry feelings), anger-out (expression of anger toward property or people), and anger control (the frequency of attempts to control expressions of anger). The Abbreviated Duke Social Support Index (28) quantified growth in the social network (number of weekly contacts) and perceived support. The widely used Center for Epidemiological Studies Depression Scale (CES-D) (29) measured depressed mood.

Intervention

The intervention consisted of three sequential phases. In phase 1, called “joining,” clinicians met with individual families for two or three sessions to evaluate ongoing problems and define treatment goals. In phase 2, all veterans and families attended two three-hour educational workshops about TBI. In phase 3, veterans and their families attended problem-solving multifamily group meetings bimonthly for six months. The group sessions consisted of a 15-minute “socialization” period followed by a “go-round,” when each family described problems or life events experienced since the last meeting and reviewed homework assignments. On the basis of the go-round, a problem or goal was selected for a formal problem-solving exercise involving defining the problem or goal, generating all possible solutions, evaluating pros and cons of each solution, and formulating a solution and plan for implementation.

Group leaders kept a list of all problems identified by group members as candidates for problem solving and a written record of all group-generated solutions. The duration of all three phases of treatment was approximately nine months. Study clinicians attended a two-day training session about the model conducted by highly experienced multifamily group clinicians and supervisors (DN and DD).

All sessions occurred between April 2010 and March 2011 and were audiotaped and supervised by one of the trainers (DN). The trainer rated adherence to the different components of the treatment model for all sessions for the first two groups and 40% of sessions for the third group. Adherence was rated on a 3-point scale according to criteria described by McDonell and others (30), with 0 defined as nonadherence, 1, partial adherence, and 2, full adherence. The mean global adherence score across the three groups was 1.79±.43. Interrater reliability of independent ratings by trainers for six sessions of the Bronx group was adequate (kappa=.80).

Quantitative data analysis

Paired t tests were used to evaluate changes within participants over the nine months of treatment. The association between baseline levels on neurocognitive tests and the degree of change on outcome measures observed over the course of treatment (represented by posttreatment minus pretreatment scores) was evaluated by using Pearson product-moment correlation coefficients (Pearson’s r).

Results

Implementation

Buy-in by providers, veterans, and families.

A total of 59 potential candidates were identified and prescreened by their clinician, and 34 (58%) were approved and referred for study. Twenty-five (42%) of the 59 veterans identified as potential participants were excluded at the provider level because their TBI was judged to be too severe for participation (N=10), the provider was not able to identify at least one family member who was involved with the patient’s care (N=6), or there was a conflict with another treatment or study (N=4). Additionally, some clinic staff felt that adding consideration of the current intervention protocol to the already complex set of evaluation and treatment procedures for new veterans was too demanding, and this position was adopted as clinic policy. Therefore, providers also excluded five new patients.

Of the 34 veterans approved for the study, seven (21%) did not meet eligibility criteria. Two had a severe substance use disorder, and five had spouses who felt too burdened to participate or were not interested in participating (three were mothers of infants or very young children). Of the 27 eligible veterans, 13 either refused (N=6) or were lost to study (N=7). Reasons for refusal included a specific work- or school-related scheduling conflict or feeling too busy. Three of the five who had scheduling conflicts worked a 3–11 p.m. shift, which conflicted with the more popular early evening time for holding the group. Another worked a long daytime shift with a long commute.

A total of 14 veterans (52% of the 27 veterans who were eligible for the study and 41% of the 34 veterans approved for inclusion by providers) met study criteria, gave informed consent, and were enrolled in the study. [A table summarizing the reasons for exclusion or refusal is available online as a data supplement to this article.]

Veterans and family members were treated in three cohorts, one group of four and one group of five veterans and families at the VAMC in Durham and one group of four veterans and families at the VAMC in the Bronx. One couple in Durham’s first cohort withdrew after four group sessions because of the veteran’s medical condition, and two family members in the second cohort dropped out within the first few meetings, one because of a divorce and one because of relapse of a substance use disorder. In both cases the veteran remained in the study. Additionally, one couple in the Bronx withdrew during the joining sessions because of the veteran’s medical complications. Eleven (32%) veterans and nine (27%) family members and intact dyads who were referred to the study completed the treatment and were included in all quantitative analyses. Attendance at group sessions was good—across both sites, participants attended 46 of 59 total possible sessions (78%).

Tables 1 and 2 present the study sample’s demographic and military characteristics and neurocognitive data. Ten veterans reported that their TBI was associated with a loss of consciousness of <30 minutes, characterizing their TBI as mild (19). The duration of loss of consciousness of the 11th veteran was unknown. On neuropsychological testing, as a group they demonstrated intact learning and retention of new material, intact attention, and mild to moderate impairment of executive functioning, with considerable variability observed. Seventy-nine percent of veterans had comorbid PTSD, and 20% to 31% met criteria for other comorbid disorders common in this cohort (Table 2). Overall, the study sample was characteristic of OEF/OIF era veterans with mild TBI and comorbid mental health problems.

Group composition and emergent needs.

The problems identified by veterans and family members clustered in three areas: problems related to cognitive deficits, such as memory problems; relationship issues, such as poor communication; and veterans’ self-identity or community interface, such as difficulty accepting limitations or negotiating work and school accommodations. Because most participants were couples, most problems identified in the go-rounds related to disruptions in marital relationships. Although standard problem solving was a good fit for some relationship problems, such as coordinating child care, it proved less ideal for others, such as communicating negative feelings without escalation. Because relationship and communication problems were frequently brought in for group problem solving, we incorporated communication skills training methods from other evidence-based family psychoeducation models, such as behavioral family therapy (31) to facilitate problem solving by couples having relationship problems. Other modifications of the basic model adopted at the trial’s outset are discussed elsewhere (32).

Quantitative analyses

Paired t tests found significant decreases in veteran anger expression and increases in veteran social support and occupational activity (Table 3). Effect sizes ranged from .6 to 1.0. Significant decreases in caregiver burden and increases in family empowerment were observed as well.

Veterans’ baseline performance on a measure of attention and speed of processing (Trail Making Test, part A) was negatively associated with improvement on outcome measures of anger expression (r=–.72, df=10, p=.020), depression (r=–.63, df=10, p=.051), and PTSD symptoms (r=–.85, df=10, p=.004) (Table 4). Baseline performance on a measure of executive functioning (Trail Making Test, part B) was negatively associated with improvement in occupational activity (r=–.76, df=10, p=.048). These results suggested that neurocognitive processes may affect treatment response.

Discussion

This study evaluated the feasibility, acceptability, and initial efficacy of a multifamily group intervention for veterans with TBI, a model adapted from an evidence-based treatment for persons with serious mental illness. The relatively low overall ratio of veterans enrolled in treatment to veterans referred for treatment (41%) was attributable to perceptions and barriers identified by all stakeholders.

Among some providers, discussing with the veteran the potential for family involvement was viewed as incompatible with or as interfering with the primary diagnostic task of the initial visit. Given the complexities involved in diagnosis of TBI, particularly mild TBI (33,34), and the minimal knowledge of staff about the benefits of family involvement, particularly multifamily groups, this view was not surprising and suggested a need to allocate more resources to increase readiness for implementation (35). For example, education demonstrating that family members may facilitate evaluation by providing history not obtainable from the veteran because of cognitive impairment may help to engender a more receptive implementation climate.

Exclusions of veterans by providers related to the severity of the TBI, coupled with pilot data demonstrating a negative association between cognitive impairment and treatment gains on several outcome measures, suggested that implementation might be more feasible and efficacious for veterans with less severe TBI. Although our results were consistent with other research demonstrating limited efficacy of psychosocial treatment for individuals with even relatively mild cognitive impairment (36), recent work suggested the potential for adapting evidence-based psychosocial treatments for individuals with cognitive impairment (36,37).

Refusal or early dropout at the family level related to caring for an infant or young child may relate both to practical considerations, such as finding appropriate child care, and to a high degree of perceived burden. Outreach to spouses and veterans through the VA Intranet to provide education on the potential benefits of such programs, including the use of veteran or family testimonials, might provide a nonintrusive way to tip the costs-benefits ratio sufficiently to interest families in a phone consultation with clinicians, who can begin to engage the family. For these families, conducting joining sessions via teletherapy might help minimize treatment burden while building an alliance. The use of the Internet and teletherapy to extend the boundaries of traditional psychosocial interventions for veterans’ families has grown in recent years (38,39). An example is the recent development of an interactive electronic program called “Families of Heroes,” designed to educate OEF/OIF families to identify behaviors indicating a need for VA services (40).

The major barrier at the veteran level was related to scheduling or time constraints. Many veterans work afternoon shifts or attend evening classes. Offering midday groups would make the treatment accessible to these families. As Dausch and others (41) pointed out, offering effective family services requires flexibility and planning by providers and administrative leadership to accommodate working veterans and relatives.

To address the needs of participants, specifically couples, that emerged during treatment, we introduced communication training methods from individual, skills-based family psychoeducation. Although these methods were introduced within the core problem-solving framework as solutions to specific marital problems of the participants, their use represents an adaptation of the original treatment model to meet the assessed needs at one local site that had enrolled only couples.

According to Damschroder and colleagues (35), the capacity to tailor or refine the “adaptable periphery” of an intervention to meet local needs without altering “core components” is critical to successful implementation. The fact that one couple and two spouses were early dropouts at the site where the groups represented a mix of family relationships lent support to the need to provide couples with more finely-honed tools to address relationship problems within the core problem-solving model. It also suggested that implementation may be smoother with more homogeneous groups.

Our quantitative measures showed medium to large effect sizes for outcomes of all veterans and most family members and supported our hypothesized improvement in anger management and use of social support among veterans and burden and empowerment among families. However, it should be emphasized that because of the lack of a control group, we cannot attribute improvements to the intervention itself rather than to nonspecific treatment effects, such as devoting extra attention to participants. Rather the findings provided support for implementing a randomized clinical trial to evaluate changes observed in multifamily group treatment for veterans with TBI against nonspecific treatment effects. Future work should incorporate additional measures of family and family member coping as well as marital satisfaction. Our preliminary implementation data will allow us to modify our interactions with providers, veterans, and families in approaching this next, critical step.

Conclusions

An open trial of an adaptation of multifamily group treatment for OEF/OIF veterans with TBI demonstrated significant improvement in anger management and use of social supports among veterans and in burden and empowerment among families, providing a rationale for mounting a randomized clinical trial. Challenges to implementation of this intervention observed at the systems, provider, and participant levels suggested the need for building in education for both providers and family members about the benefits of family involvement. Outreach via Web-based products may help promote treatment participation. Adaptation of evidence-based treatments for a new population may require tailoring and refining the model to meet local or specialized needs of the new cohort that emerge during treatment.