The Use of a Brief Family Intervention to Reduce Dropout Among Veterans in Individual Trauma-Focused Treatment: A Randomized Controlled Trial

Abstract

Dropout from trauma-focused treatment for posttraumatic stress disorder (PTSD) represents a daunting challenge for the field, particularly among military and veteran samples. Family involvement may help to increase the effectiveness of PTSD treatment while also improving retention. We tested a two-session brief family intervention (BFI) protocol delivered as an adjunct to individual trauma-focused treatment among a sample of 20 veteran–family member dyads (N = 40). Willingness to participate in the family-inclusive protocol was high, with over 85% of veterans and family members who were screened agreeing to take part. All enrolled veterans were beginning a course of either cognitive processing therapy (CPT) or prolonged exposure (PE), delivered in outpatient Veterans Affairs clinics. Family members were randomized to either receive or not receive the BFI from study clinicians. In the BFI condition, 20.0% of veterans dropped out of CPT/PE before the 16-week study end; the remainder were either still attending on-protocol sessions or had completed the full protocol. In the control condition, 40.0% of veterans dropped out of CPT/PE before the end of the study. Observed significant, large magnitude decreases in PTSD symptoms over time did not differ by condition, ESsg range = −1.12 to −2.04. Accommodation did not significantly decrease over time in either condition, ESsg range = 0.18 to −0.98. The BFI represents a promising option for veterans, family members, and clinicians who are seeking a brief, feasible, narrowly focused method for incorporating families into veterans’ individual trauma-focused therapy and potentially reducing the rate of dropout.

Trauma-focused psychotherapies, such as cognitive processing therapy (CPT; Resick et al., 2017) and prolonged exposure (PE; Foa et al., 2019) have been shown to result in substantial reductions in posttraumatic stress disorder (PTSD) symptoms (U.S. Department of Veteran Affairs [VA] and Department of Defense, 2017). However, there are persistent difficulties with treatment retention and residual symptoms at the end of a course of treatment (Hembree et al., 2003; Imel et al., 2013; S. E. Larsen et al., 2019; Schnurr & Lunney, 2019). These problems are particularly acute in veteran and military samples, which have shown poorer outcomes in trauma-focused treatment compared with civilians (Hoge & Warner, 2014; Steenkamp et al., 2015). In addition, approximately 30%–60% of veterans drop out of trauma-focused treatments prematurely (Hale et al., 2019; Kehle-Forbes et al., 2016; Meis et al., 2019).

There have been calls to overhaul trauma treatment (Steenkamp et al., 2020), and it is clear that improvements are needed. However, it would be premature to abandon the strongest treatments entirely; an alternative would be to enhance existing treatments through the addition of family members. The findings from meta-analyses have indicated that a lack of social support is one of the strongest contributors to the development and maintenance of PTSD (Ozer et al., 2008), suggesting that correctly leveraging these associations could be quite impactful. Significant others have been shown to strongly influence mental health treatment initiation for veterans with PTSD (Meis et al., 2010; Sayer et al., 2009; Spoont et al., 2014); family members may provide initial motivation for treatment, and their involvement may increase treatment retention. Meis et al. (2019) found that when veterans’ family members encouraged them to face difficult situations—a crucial feature of trauma-focused treatments— the veterans were twice as likely to finish PE or CPT. Furthermore, the results of surveys of veterans receiving VA health care have consistently demonstrated that large numbers of veterans (i.e., 80% or more) desire higher levels of family involvement in their mental health care (Batten et al., 2009; Meis et al., 2013).

In addition to their potential salutary effects on veterans’ treatment experience, families can also potentially engage in behaviors that undermine treatment. Evidence from interviews with close loved ones of veterans undergoing CPT and PE in the VA health care system indicates that most family members know very little about the treatment (Thompson-Hollands et al., 2019). If family members are not educated about the rationale and evidence supporting these treatments, both of which involve explicit conversations about the traumatic experience, they may be worried that such a focus is harmful. Studies have shown that such concerns are not uncommon even among providers who are less familiar with exposure-based treatments (Becker et al., 2004; Frueh et al., 2006), so it is reasonable to assume that family members may similarly share these fears and could potentially advocate for veterans to drop out of trauma-focused treatment.

Of equal concern is that family members might also engage in symptom accommodation, or behaviors that prevent or reduce disorder-related distress through avoidance. In PTSD, accommodation may involve family behaviors, such as immediately turning off the television if a program touches on the military or combat, or not going out to eat because the veteran feels unable to tolerate crowded restaurants (Fredman et al., 2014). Accommodation ultimately reinforces the pathology of the disorder by allowing patients to avoid their emotional experience, and levels of family accommodation have been shown to be associated with patients’ levels of PTSD symptoms (Fredman et al., 2014). Furthermore, accommodation undermines a key mechanism of PTSD treatment: The experience and acceptance of emotions surrounding the traumatic event (Foa et al., 2007; Resick et al., 2014).

There are several existing family-inclusive interventions in use within the VA. Some of these interventions are brief but not disorder-specific, such as Veteran-Centered Brief Family Psychoeducation (VA, 2016); others are tailored to veterans with PTSD and their loved ones but focus on improving relationship satisfaction, communication, and empowerment rather than on reducing PTSD symptoms directly (e.g., Fischer et al., 2013). Such protocols contribute to the important goal of addressing veterans’ larger social contexts and may provide specific skills, such as enhanced communication, but they are not PTSD treatments.

There are also a number of family-inclusive treatments for PTSD (Johnson, 2002; Monson et al., 2012; Sautter et al., 2015), but all of the existing protocols require multiple conjoint sessions over several months, posing a significant barrier to treatment access. The most well-known and commonly used family-inclusive treatment is cognitive behavioral conjoint therapy for PTSD (CBCT; Monson et al., 2012), a 15-session, fully dyadic intervention based on the cognitive behavioral interpersonal theory of PTSD (Monson et al., 2010); an 8-session CBCT protocol (Morland et al., 2019) and an intensive multicouple CBCT protocol (Fredman et al., 2020) have also been tested. The CBCT intervention includes many treatment elements, including addressing maladaptive trauma- and relationship-related thoughts and encouraging couples to undertake in vivo exposure exercises. The treatment has shown promise across a small number of studies (e.g., Monson et al., 2012) and is currently available to VA providers as part of a dissemination “rollout,” but the significant time commitment of the protocol may be a barrier to treatment for many families. Furthermore, not all providers feel confident in delivering a fully conjoint intervention, reducing the number of available clinicians. If some of the strengths of family-inclusive treatment could be gained with a less-intensive protocol, a much wider pool of veterans and their loved ones would potentially stand to benefit.

The VA has invested substantial resources into the rollout of CPT and PE across facilities (Rosen et al., 2016) to better address the needs of veterans and provide them with some of the most strongly supported PTSD treatments. Pairing family involvement with such widely disseminated interventions has the potential to make implementation more successful by capitalizing on existing structures and incentives. We therefore developed a brief, PTSD-specific family intervention, meant to be delivered as an adjunct to CPT and PE, to address what we saw as a gap in treatment options for providers and families.

The Brief Family Intervention for PTSD (BFI; Thompson-Hollands et al., in press) is a two-session protocol that focuses on psychoeducation and skill-building for family members of veterans with PTSD who are beginning a course of CPT or PE. Only the adult family member attends the BFI sessions; veterans are not present. Most of the BFI content is focused on psychoeducation about PTSD and trauma-focused treatment, and the protocol is delivered early in the veteran’s course of individual treatment. To reduce concerns about a possible breach of confidentiality, sessions are delivered by a provider who has expertise in PTSD treatment but is not the veteran’s individual trauma-focused therapy provider. The clinician describes the symptoms of PTSD and how the disorder is thought to be maintained through avoidance and distorted trauma-related cognitions. The clinician then describes trauma-focused treatment (i.e., CPT or PE, depending upon which applies to the veteran), focusing on the treatment rationale as well as the safety and efficacy of the treatment. The provider encourages family members to brainstorm ways they might actively support the veteran’s treatment efforts, such as through expressing gratitude or encouragement or by providing practical assistance with treatment-related tasks. Finally, the provider describes accommodation and helps the family member to identify how they may be inadvertently facilitating avoidance, with the goal of reducing accommodation in a way that is feasible and caring. Throughout the sessions, the family member is provided with ample opportunities to ask questions or express concerns. The intent of the BFI is to both enhance the efficacy of CPT or PE by reducing accommodation and, therefore, increasing opportunities for exposure and to increase retention in treatment by increasing veterans’ perceived social support and providing family members with accurate treatment-related information. A more detailed description of the BFI protocol and a case example can be found in (Thompson-Hollands et al., in press).

The current study was a pilot trial of the BFI. Our goals were to assess the initial feasibility and acceptability of the program through recruitment and retention data. We hypothesized that having a family member participate in the BFI would make veterans more likely to complete their individual trauma-focused treatment, given the results reported by Meis et al. (2019). We also wished to determine the extent to which having a family member participate in the BFI resulted in PTSD symptom reduction and reductions in accommodation relative to trauma-focused treatment alone, although we were largely focused on measuring effect sizes associated with participation in the BFI and, therefore, these analyses were exploratory.

Method

Participants

We sought to recruit a sample of dyads consisting of one veteran and one adult family member. The study flow is depicted in Figure 1. The inclusion criteria for veterans were: (a) current PTSD according to the criteria in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013), (b) engaged in the early stages of individual CPT or PE treatment (i.e., the veteran had completed no more than three sessions at baseline), (c) psychiatric medication stability for at least 4 weeks prior to baseline, and (d) identification of an eligible family member and willingness to have that person participate in the study. Veterans of any gender were eligible for inclusion. Family members were required to be at least 18 years of age, a family member of the veteran (i.e., spouse or romantic partner, parent, adult child), and currently living with or in frequent contact with the veteran. The exclusion criteria for veterans were current moderate or severe substance use disorder (mild substance use disorder was not an exclusion); current psychosis diagnosis, unstable bipolar disorder, or organic mental disorder; severe suicidal risk; and current participation in couples or family counseling with the identified family member. Exclusion criteria for family members were the same as those for the veterans, with the additional requirement that family members could not currently meet the DSM-5 criteria for PTSD. Restricting the participation of “dual PTSD dyads” is consistent with other family-inclusive PTSD treatment trials (e.g., Fredman et al., 2016, 2020).

Figure 1.

CONSORT Diagram of Participant Flow

Note. PTSD = posttraumatic stress disorder; BFI = brief family intervention.

Twenty dyads (N = 40 participants) were randomized to receive the BFI or not receive the BFI, with a 1:1 allocation. The sample size was small due to the pilot nature of the study. The randomization sequence was determined via a blind draw by an individual who was not a part of the research team, and the allocation sequence was maintained in a protected file that was not accessed until the eligibility of the dyad was confirmed. The large majority of veterans in the sample identified as male (95.0%); one individual (5.0%) identified as a transgender man. The veterans report their race/ethnicity as Caucasian (70.0%), Black (15.0%), and Hispanic (15.0%), and the mean participant age was 44.05 years (SD = 15.27). The average time since the veteran experienced their index traumatic event was 236.80 months (SD = 190.69). Most veterans were initiating CPT (80.0%), with the remainder initiating PE. Veterans reported their index traumatic events as combat exposure (35.0%), physical assault (25.0%), childhood sexual assault (15.0%), traumatic loss (15.0%), childhood physical assault (5.0%), and adult sexual assault (5.0%).

All family members (100.0%) identified as female and all were the cohabitating wives or romantic partners of the participating veterans. Regarding race and ethnicity, 80.0% of the family members identified as Caucasian, 15.0% as Black, and 15.0% Hispanic, and the mean family member age was 42.00 years (SD = 14.74).

Procedure

All procedures were approved by the VA Boston Healthcare System (VABHS) Institutional Review Board, and the study was registered at Clinicaltrials.gov (NCT02956434). Veterans who were beginning individual CPT or PE were recruited from mental health clinics at VABHS, including the PTSD Clinical Team and the Center for Returning Veterans. Study personnel approached mental health clinicians directly to inform them about the study opportunity, and clinicians then discussed the research with veterans in their caseload who were planning to begin CPT or PE. Study recruitment and follow-up occurred between June 2018 and May 2020.

Clinical or research staff first approached veterans about study participation; if a veteran appeared to be eligible based on a phone screen and was interested in participating, they were asked to provide contact information for the family member they wished to involve. The research team then contacted family members and the study was explained to them. Baseline visits were scheduled once the participants had successfully completed the phone screen. Baseline visits were held separately for veterans and family members, and written consent was obtained. If both dyad members were found to be eligible at baseline and were still interested in participating, they were considered to be enrolled, with follow-up assessments conducted at 6, 12, and 16 weeks.

Family members were randomized to receive or not receive the BFI. Family members were given the choice to receive the BFI sessions in person or by phone, with sessions conducted shortly after randomization. Doctoral-level psychologists who received direct training and consultation from the BFI’s developer (Johanna Thompson-Hollands) conducted all sessions (more information on training is provided in later sections).

Veterans received individually delivered CPT or PE from their usual-care clinicians at VABHS. This trauma-focused treatment was not a part of the study procedures, and clinicians were not study staff members. However, providers in the recruiting clinics at VABHS are highly experienced in the delivery of evidence-based PTSD treatments. Providers who delivered CPT and PE to the study participants were largely staff psychologists and psychology postdoctoral fellows and interns, although four veterans were treated by psychology practicum students. As noted, providers tracked whether sessions were on- or off-protocol; we found that 75.11% versus 24.89% were on- and off-protocol, respectively, with most off-protocol sessions occurring prior to the start of CPT or PE. Examples of off-protocol sessions were sessions devoted to anger management skills or other preparations for trauma-focused treatment. Any family involvement initiated by the veteran’s CPT or PE provider was also tracked. Across all veterans enrolled in the study, only one had a joint session with a family member. This session occurred after the veteran had already discontinued his PE treatment and transitioned into a non–trauma-focused treatment approach.

Training of Assessors and BFI Clinicians

Baseline assessments were conducted by the first author, who had extensive experience in both semistructured assessments and PTSD treatment trials. Follow-up Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) interviews were administered by members of the National Center for PTSD Assessment Core. These individuals hold masters- or doctoral-level degrees in psychology and had received comprehensive training in the assessment of PTSD and other diagnoses as well as training on the CAPS-5 measure specifically. Assessment Core personnel conduct assessments across numerous clinical trials simultaneously and are blind to treatment condition.

All BFI clinicians were doctoral-level psychologists currently engaged in regular clinical work at VABHS, including caseloads that primarily or exclusively consisted of veterans diagnosed with PTSD. The first author, and BFI developer, trained BFI clinicians. All BFI clinicians received a copy of the BFI manual and engaged in one-on-one didactic information sessions. There was ample time provided to address any questions the providers had and to conduct role-plays and demonstrations as needed. The first author also provided ongoing supervision of any active BFI cases, including reviewing all tapes of the BFI sessions and providing feedback to the clinicians. None of the BFI sessions were judged to be nonadherent.

Impact of COVID-19

The COVID-19 pandemic occurred during the period of data collection for this study. VABHS began canceling in-person visits and transitioning to telehealth sessions in mid-March 2020; this continued through the completion of the study period. The large majority of participants in the present sample had completed all study procedures when these changes began, but three dyads were still in the follow-up period for the study. Two of the three veterans in these dyads had dropped out of CPT or PE several weeks prior to the COVID-19–related changes; accordingly, they are counted as dropouts in the data presented herein. The third veteran’s treatment was disrupted after Session 3 of CPT because his clinician, who was a practicum student, was not permitted to conduct telehealth sessions due to the need for onsite supervision coverage. This veteran participant was transitioned to a more senior clinician who completed treatment with the veteran.

Measures

PTSD Symptoms and Diagnostic Status

The CAPS-5 (Weathers et al., 2013) was administered to veterans to assess PTSD symptom severity and diagnostic status. The CAPS-5 has demonstrated excellent validity and reliability (Weathers et al., 2018). Trained, blinded, masters- or doctoral-level evaluators administered the assessment. To check for reliability, 26.0% of the CAPS-5 assessments were double-coded (κ = 1.00). In the present sample, Cronbach’s alpha for the CAPS-5 at baseline was .77.

Exclusion Criteria

Relevant sections of the Structured Clinical Interview for DSM-5 (SCID-5; First et al., 2015), a widely used measure of psychiatric disorders, were administered at baseline to all participants to rule out exclusion criteria.

Retention and Dropout

Veterans’ treatment dropout, retention, and homework compliance were reported by their individual CPT or PE clinician. Treatment progress and dropout were reported using the Session Tracking Form, which included session dates and noted whether each session was on-protocol. The Homework Compliance and Quality form (Leung & Heimberg, 1996) was used to assess homework compliance. Clinicians were asked to rate how much of the assigned homework the veteran completed (i.e., “none completed,” “attempted some,” “completed most, “completed all”) and the extent to which the veteran appeared to understand the homework assignment and produce good-quality work (i.e., “poor quality and minimal understanding,” “very good quality and very good understanding”). Both forms were collected from the clinicians at the end of the veteran’s study participation or when the veteran finished or dropped out of treatment, whichever occurred first.

Family Member Accommodation

Family members completed the Significant Others’ Responses to Trauma Scale (SORTS; Fredman et al., 2014) to measure their levels of accommodation. The SORTS is the only validated measure of accommodation that is specific to PTSD. Items assess accommodation behaviors, such as avoiding the person with PTSD because of their irritable mood, avoiding physical contact because of the person with PTSD’s discomfort, and avoiding discussing the traumatic event in front of the person with PTSD. The SORTS has demonstrated strong internal consistency and has been associated with both PTSD severity and relationship distress (Fredman et al., 2014). In the present sample, Cronbach’s alpha for the SORTS at baseline was .96.

Treatment Expectancy and Satisfaction

Family members who received the BFI also completed the Treatment Expectancy Questionnaire (TEQ; Borkovec & Nau, 1972) after Session 1 and the Client Satisfaction Questionnaire (CSQ; D. L. Larsen et al., 1979) after Session 2. Items on the TEQ are rated on a 0–10 scale, with higher scores indicating higher expectations for treatment success. Items on the CSQ are rated on a 1–4 scale, with higher scores indicating higher levels of treatment satisfaction (maximum = 20).

Data Analysis

All data were analyzed in an intent-to-treat format. The degree of missing data was moderate: Overall, 78.1% of scheduled assessments were completed. We examined the impact of the BFI on PTSD symptoms and accommodation in two ways. First, using hierarchical linear modeling (HLM) in R (R Core Team, 2017), we constructed two-level models consisting of time (Level 1) nested within persons (Level 2). The dependent variable was the CAPS-5 total score for the veteran model and SORTS total score for the family-member model. Models included a linear time variable to capture the slope of change during the study (baseline to Week 16, centered at baseline), an intercept, and a random intercept. A random slope was tested but did not improve the model fit and, thus, was not retained. Finally, a treatment condition variable (BFI vs. control) was included, as well as a Treatment Condition x Time moderating variable. In addition, we used Mplus (Version 8) to calculate CAPS-5 and SORTS score means and standard deviations using full information maximum likelihood (FIML) estimation (Muthén & Muthén, 1998–2017). We used these estimates to calculate standardized mean gain scores (ESsg) and standard errors of the standardized mean gain scores (SEsg; Lipsey & Wilson, 2001). These effect sizes quantify the magnitude of change in repeated-measures designs and can be interpreted in a similar fashion to Cohen’s d; for example, an ESsg of .80 or higher is indicative of a large-magnitude change over time.

Determining sample sizes for multilevel models is a complex issue, and the “rules of thumb” are disputed (Maas & Hox, 2005; Mok, 1995). Furthermore, “effective” sample sizes, as opposed to nominal sample sizes, in HLM will vary according to the model’s predictors (Bickel, 2007). Given the pilot nature of the present examination, we recruited sample sizes that would allow for the detection of large effects in the HLM analysis (McCoach & Garcia, 2018) and calculated effect sizes to more precisely characterize the impact of the intervention.

Results

Recruitment

As shown in Figure 1, most veterans screened for the study were interested in having a family member participate in the BFI. Of the 31 veterans who were phone-screened, only two reported that they were not willing to have a family member involved, and two other veterans stated that their desired family member was too busy to take part. The remaining veterans (87.1%) were willing to have a family member participate in BFI sessions and felt that the time commitment would be reasonable for their family member to complete. Of the 25 family members who were initially approached regarding participation, only two declined, leaving 92.0% who were open to participating in the BFI.

BFI Retention and Satisfaction

All family members in the BFI condition completed all sessions of the protocol; five family members chose to complete both of their sessions in person, whereas four completed the sessions by phone and one completed one session in person and one by phone. Family members were quite engaged: There was only one weather-related cancellation, which was promptly rescheduled. Postintervention ratings were high. After Session 1, family members largely reported via the TEQ that the intervention seemed logical to them (M rating= 8.13, SD = 0.99) and that they were somewhat confident that the BFI would improve their ability to help the veteran (M rating = 7.50, SD = 0.93). Following the receipt of both sessions, family members reported an average rating of 17.38 (SD = 2.62) on the CSQ.

CPT or PE Retention Among Veterans

In total, 20.0% of the veterans in the BFI condition dropped out of CPT or PE prior to the end of the 16-week study period. The remainder of the veterans in the BFI condition were either still attending on-protocol sessions or had completed the full protocol. In the control condition, 40.0% of veterans dropped out of CPT or PE prior to the end of the study period. The difference in dropout between the two conditions was not statistically significant, χ²(1, N = 20) = 0.952, p = .329. The severity of PTSD symptoms at baseline was not predictive of dropout from CPT or PE, B = 0.01, SE=0.05, p = .814.

Symptom Change in Veterans and Family Members

PTSD Symptoms

The intraclass correlation coefficient (ICC) for the null model was .367, indicating that 36.7% of the variance in CAPS-5 scores was between rather than within veteran participants. The results of HLM analyses indicated that PTSD symptoms declined significantly over time across conditions, B = −1.46, SE = 0.22, t(39.2) = −6.58, p < .001. There was no effect of treatment condition on the intercept of symptoms, B = −6.48, SE = 4.69, t(26.6) = −1.38, p = .179, or the slope of change over time, B = 0.42, SE = 0.31, t(39.2) = 1.38, p = .177.

Table 1 displays FIML-estimated mean CAPS-5 scores, by time and condition, and Table 2 displays the within-group standardized mean gain score effect sizes by condition. Figure 2 shows the pattern of change in CAPS-5 scores across both groups. Veterans in both groups experienced substantial decreases in PTSD symptoms overall, with large effect sizes at each assessment point.

Table 1.

Full Information Maximum Likelihood Means and Standard Deviations for Veterans (CAPS-5) and Family Members (SORTS)

Group and intervention	N	Baseline		Week 6		Week 12		Week 16
		M	SD	M	SD	M	SD	M	SD
Veterans
BFI	10	37.80	8.81	22.23	9.91	22.89	12.26	18.60	10.98
Non-BFI	10	42.40	8.81	29.60	9.04	25.78	10.86	21.21	9.91
Family members
BFI	10	33.30	21.79	32.18	14.27	30.84	25.04	22.82	11.16
Non-BFI	10	45.40	27.24	51.22	25.61	41.87	28.03	29.88	24.87

Note. CAPS-5 = Clinician-Administered PTSD Scale for DSM-5; SORTS = Significant Others’ Responses to Trauma Scale; BFI = brief family intervention.

Table 2.

Within-Group Effect Sizes From Baseline to Follow-Up in Veterans (CAPS-5) and Family Members (SORTS)

Group and intervention	Baseline to Week 6		Baseline to Week 12		Baseline to Week 16
	M 1	SE	M	SE	M	SE
Veterans
BFI	−1.24	1.21	−1.19	1.07	−1.37	1.35
Non-BFI	−1.12	1.09	−1.47	1.11	−2.04	1.62
Family members
BFI	−0.25	0.62	−0.08	0.16	−0.98	1.26
Non-BFI	0.18	0.29	−0.17	0.27	−0.30	0.23

Note. CAPS-5 = Clinician-Administered PTSD Scale for DSM-5; SORTS = Significant Others’ Responses to Trauma Scale; BFI = brief family intervention; ESsg = standardized mean gain score; SEsg = standard error of the standardized mean gain score.

¹M = standardized mean gain score; SE = standard error of the standardized mean gain score

Figure 2.

Full Information Maximum Likelihood Estimate Mean Clinician-Administered Posttraumatic Stress Disorder Scale for DSM-5 Total Scores Over Time

Note. CAPS-5 = Clinician-Administered Posttraumatic Stress Disorder (PTSD) Scale for DSM-5; BFI = brief family intervention.

Accommodation

The ICC for the null model was .764, indicating that 76.4% of the variance in SORTS scores was between rather than within family members. The results of HLM analyses indicated that accommodation did not change significantly over time across conditions, B = −0.63, SE = 0.42, t(33.9) = −1.49, p = .146. There was no effect of treatment condition on the intercept of accommodation, B = −13.67, SE = 11.33, t(21.1) = −1.21, p = .241, or the slope of change over time, B = 0.14, SE = 0.60, t(33.7) = 0.23, p = .820.

Table 1 displays the FIML-estimated mean SORTS scores, by time and condition, and Table 2 displays the within-group standardized mean gain score effect sizes. Figure 3 shows the pattern of change in SORTS scores across both groups. The observed standard deviations and standard errors of the effect sizes indicate these scores are characterized by substantial heterogeneity both at each time point with regard to change over time. The largest between-condition difference was observed at Week 6, when a slight increase in accommodation was observed in the control condition and a slight decrease was observed in the BFI condition.

Figure 3.

Full Information Maximum Likelihood Estimate Mean Significant Others’ Responses to Trauma (SORTS) Total Scores Over Time

Note. BFI = brief family intervention.

Discussion

The present results indicate that this novel, BFI is feasible and acceptable to family members and veterans. Recruitment was robust, no family members dropped out, and satisfaction ratings were high. Elsewhere, we have described qualitative feedback from both veterans and family members indicating that they found this approach to be valuable (Thompson-Hollands et al., in press). The quantitative outcomes presented herein reveal a similar story and clearly suggest that this format of family involvement is appealing and desirable to both veterans and their family members. These feasibility and acceptability outcomes illustrate the coherence of this approach with patient-centered care.

In addition to offering veterans the care they desire and providing a novel option for family involvement that is less onerous than existing approaches, our aim was to determine whether the BFI offered measurable clinical benefits for the pressing problems of treatment dropout and poor response. Although the results for dropout were not statistically significant, which is likely due to the small sample size, we found that having a family member participate in the BFI resulted in substantially reduced treatment dropout relative to dropout observed among veterans who did not have a family member receive the BFI. Recently published program evaluation data from the VABHS PTSD clinic, the main recruitment clinic for the present study, has shown a dropout rate of 63.8% among veterans who initiate CPT or PE (Niles et al., 2018). We note that even the control group in the present study had better retention outcomes than those reported by Niles and colleagues; we suspect that simply having a supportive family member that the veteran was willing to invite into the present study may have had a beneficial effect on veterans’ treatment experience, even if the family member was not ultimately randomized to receive the BFI. Future research might seek to quantify this difference by following a group of veterans who either have no family members available or who explicitly decline to involve them.

The results of the HLM analyses demonstrated no condition effects of the BFI on either PTSD symptoms or accommodation. We opted to construct the present study as a rigorous comparison of gold-standard individual trauma-focused therapy plus or minus the addition of a brief adjunctive family intervention. As might be expected given that all veterans in the present sample were receiving CPT or PE, on average, the veteran participants experienced large reductions in their PTSD symptoms at each assessment point regardless of whether their family member received the BFI. Future examination in larger samples will help determine whether the BFI confers any advantage with regard to the rate of PTSD symptom change and if the potential benefits are mediated by reduced family accommodation or simply by increased retention in CPT or PE.

Regarding the lack of effect of condition on accommodation symptoms in the HLM analyses, we expect that our small sample size and substantial heterogeneity, as indicated by the standard deviations and standard errors in our analyses, contributed to the null finding. Furthermore, family members in the control group slightly increased their accommodation early in treatment, perhaps in an attempt to soothe veterans who were displaying normative distress early in their trauma-focused work, whereas those in the BFI condition slightly decreased their accommodation during the same period. Such differences are a promising signal regarding one potential mechanism of the BFI’s effects.

Our study had a number of limitations, including the small sample size. Our next step will be to evaluate the BFI in a much larger sample that will allow for a more robustly powered analysis of main effects as well as an investigation of potential moderating variables. Second, our follow-up window was somewhat brief at only 16 weeks postbaseline. Although our analysis was strengthened by having multiple follow-up assessments during the treatment itself rather than relying on a simple pre- and posttreatment assessment, a longer follow-up would have been preferable. Third, our sample was largely homogenous regarding veteran and family member gender; therefore, the results cannot speak to the impact of family involvement among more diverse dyads. There was diversity within our sample with respect to race and index trauma type. We are currently engaged in qualitative interviews with women veterans with PTSD to learn about their preferences for family involvement. Finally, we offered remote BFI sessions exclusively by phone rather than providing a video-based version, which may have diminished the effectiveness compared to in-person or video sessions.

Overall, the present results are encouraging with regard to the promise of the BFI. There was a great deal of openness to family involvement among the veterans and family members we approached, and our retention and satisfaction for the BFI itself were excellent. These are all important markers as the VA strives to expand veteran-centered treatment options. Furthermore, it appears that having a family member participate in the BFI may reduce dropout from trauma-focused treatment. This is consistent with previous work (Meis et al., 2019) and positions the BFI as an exciting option for integrating families into well-established, trauma-focused treatments and keeping veterans engaged while demanding a far-reduced time cost for both providers and families compared to existing treatments.

As providers seek to deliver the best possible PTSD treatments, harnessing veterans’ closest relationships continues to be an important avenue to pursue. Veterans, family members, and providers should have a range of family-inclusive treatment options available to them, as no single treatment will be ideal for all family constellations or situations. There will be times when a very intensive protocol will be preferable and times when a simpler approach is sufficient. The BFI will likely be broadly feasible to a range of families and providers. The present results suggest that the BFI will be acceptable to family members and can potentially lead to retaining a veteran in treatment when they might otherwise drop out. For veterans looking to make choices with regard to the intensity of family involvement in their PTSD treatment, the BFI is an excellent addition to the existing family-inclusive options.

Open Practice Statement

The study reported in this article was not formally preregistered. Neither the data nor the materials have been made available on a permanent third-party archive; requests for the data or materials can be sent via email to the lead author at johanna.thompson-hollands@va.gov.