Change in Social Support While Participating in Behavioral Activation for PTSD

Abstract

Objective:

Lack of social support predicts the development, maintenance, and exacerbation of posttraumatic stress disorder (PTSD). Moreover, social dysfunction is associated with recurrent episodes of PTSD care, and detachment/estrangement from others is a strong predictor of suicidal ideation among those with PTSD. Thus, treatments to improve social functioning among those with PTSD are needed.

Method:

Eighty veterans of recent operations in Iraq and Afghanistan participated in a randomized controlled trial comparing treatment as usual to behavioral activation (BA) for PTSD, a treatment that focuses on reducing avoidance behaviors and increasing engagement in valued goals rather than explicitly confronting trauma memories.

Results:

Mixed-model regression analyses revealed that, accounting for gender, baseline PTSD, and marital status, participants who received BA experienced greater improvements in the number of social supports from baseline to posttreatment compared with those in treatment as usual (F_1,96 = 6.29, p = .014). Gains were not maintained at 3-month follow-up, and significant results were not found for satisfaction with social supports.

Conclusions:

BA may facilitate an increase in the perceived number of social supports available to veterans with PTSD, but treatment adaptation may be necessary to maintain these gains and to increase satisfaction with social support.

Hundreds of thousands of veterans suffer from posttraumatic stress disorder (PTSD), including newer generations of veterans from operations in Iraq and Afghanistan (Fulton et al., 2015). Treatment-seeking veterans with PTSD frequently cite poor social support as a concern, and research suggests that PTSD erodes social support over time (Rosen, Adler, & Tiet, 2013; Shallcross, Arbisi, Polusny, Kramer, & Erbes, 2016). Poor social support has been shown to be associated with PTSD symptom maintenance (Schnurr, Lunney, & Sengupta, 2004) as well as interference with (Evans, Cowlishaw, Forbes, Parslow, & Lewis, 2010) and dropout from trauma-focused PTSD treatments (Gros, Price, Yuen, & Acierno, 2013). Consequently, PTSD treatments that improve social support, in addition to the PTSD symptoms they already target, may be especially effective at fostering long-term recovery.

Behavioral activation (BA) psychotherapy, which focuses on reducing avoidance behaviors and increasing engagement in valued goals and activities (e.g., improving or increasing relationships) may increase social support via an increase in the variety of activities attempted. This increase in variety then potentially increases the number and variety of other people with whom one is in contact, adds activities that have specific social components, and reduces isolative, solitary behavior. In studies of BA for depression, BA has produced significant improvements in perceived social support (e.g., Gawrysiak, Nicholas, & Hopko, 2009; Hopko et al., 2011) and social functioning (Weinstock, Munroe, & Miller, 2011) in both randomized controlled and open trials. BA may have similar effects on the social functioning of individuals with PTSD, but this hypothesis has yet to be tested. Thus, the present study explores changes in social support among American veterans from operations in Iraq and Afghanistan participating in a randomized clinical trial of BA for PTSD compared with treatment as usual (TAU). We hypothesized that individuals participating in BA for PTSD would demonstrate significantly greater improvement in self-reported number of and satisfaction with social supports compared with those in TAU from pretreatment to posttreatment and from pretreatment to 3-month follow-up.

Method

Participants and Procedure

The present study is a secondary data analysis of a randomized trial involving 80 veterans of operations in Iraq and Afghanistan seen at two PTSD specialty clinics in Veterans Health Administration facilities in the Pacific Northwest. Eligible veterans had a current PTSD diagnosis (determined by clinical interview [Clinician Administered PTSD Scale-IV; Weathers, Keane, & Davidson, 2001]) and no concurrent psychotherapy, no change in psychiatric medication in the 4 weeks prior to enrollment, and did not have current diagnoses of psychosis, bipolar disorder, substance dependence, or imminent suicidality/homicidality. Participants were randomized to either BA (N = 42) or TAU (N = 38). All procedures were approved by the VA Puget Sound Health Care System - Seattle Division and VA Portland Health Care System institutional review boards. Participants completed measures at baseline, after treatment, and 3-month follow-up. Following the first therapy appointment or 1 month after baseline (whichever came first), participants were tracked to have their posttreatment assessment 12 weeks later and the follow-up assessment 24 weeks later.

More detail regarding TAU and BA for PTSD interventions can be found in (Wagner, Jakupcak, Kowalski, Bittinger, & Golshan, 2018). In brief, participants randomized to BA for PTSD (Jakupcak et al., 2006; Wagner, Paulson, & Jakupcak, 2008) were offered up to eight individual 45- to 60-min psychotherapy sessions beginning with PTSD psychoeducation, followed by behavioral analysis of avoidance, activity scheduling, mood monitoring, and attention to present moment experience to reduce disengagement in activity, similar to mindfulness (Jacobson, Martell, & Dimidjian, 2006). Individuals randomized to TAU could participate in trauma-focused treatments (e.g., Cognitive Processing Therapy [Resick, Monson, & Chard, 2017], Prolonged Exposure Therapy [Foa, Hembree, & Rothbaum, 2007]), individual or group-based non–trauma-focused treatments, and/or pharmacotherapy offered by PTSD outpatient clinic staff. These options reflect real-world practice in PTSD outpatient clinics in the Veterans Health Administration (Freedland, Mohr, Davidson, & Schwartz, 2011). Treatment dose was determined collaboratively with the participant and provider in TAU, but participants were offered six or more individual psychotherapy sessions to approximate dose of care in BA. Despite equal access to care, individuals in BA for PTSD received significantly more individual treatment than those in TAU (F_1,73 = 44.86, p < .0001; M for BA = 6.65, SD = 2.88; M for TAU = 2.55, SD = 2.40, see Supplemental Table 1 for more information).

Measures

Demographics.

Data on age, gender, marital status, and other demographic variables were collected in a standard demographic questionnaire.

PTSD Checklist for DSM–IV, military version (PCL-M; Weathers, Litz, Herman, Huska, & Keane, 1993).

The PCL-M is a 17-item self-report measure assessing prior-month distress from military-related PTSD symptoms on a 1 (not at all) to 5 (extremely) Likert scale. Internal consistency was acceptable (baseline α = .75). Regarding the range of baseline PTSD for the two samples, the TAU baseline PCL-M range was 41–73. The BA baseline PCL-M range was 39–71.

Sarason Social Support Questionnaire–Short Form (Sarason, Sarason, Shearin, & Pierce, 1987).

This questionnaire is a six-item questionnaire, with each two-part question measuring perceptions of quantity of and satisfaction with social support. Number of supports questions (SSQN) are rated from 0 to 9, and satisfaction with support questions (SSQS) are rated from 1 (very dissatisfied) to 6 (very satisfied). Items for each type of question are averaged to produce two separate scores. Internal consistency for these scales was excellent for all time points (SSQN baseline, α = .91; after treatment, α = .94; follow-up, α = .96; SSQS: baseline, α = .92; after treatment, α = .95; follow-up, α = .96).

Data Analysis

Because of high levels of skew, the SSQN scales (positively skewed) were transformed with square root transformations and the SSQS scales (negatively skewed) were reflected and transformed with square root transformations (Shapiro-Wilk test for SSQN, baseline = .09, after treatment = .45, follow-up = .51; for SSQS, baseline = .08, after treatment = .11, follow-up = .01). Data were analyzed using mixed-effects regression modeling, accounting for missing data using restricted maximum likelihood, with a random intercept. Improved model fit indices (i.e., smaller −2 log likelihood, Aikake Information Criterion, and Bayesian Information Criterion) suggested that time should be modeled as a series of dichotomous variables (e.g., baseline vs. posttreatment, baseline vs. follow-up), rather than as a continuous variable. To account for attrition, differences among those who did and did not complete assessments were assessed on numerous characteristics. Significant differences occurred only for marital status, which was included in all analyses as a covariate. The initial model assessed SSQN as the outcome, accounting for gender, baseline PTSD, and marital status, with group by time interaction terms for each time point. A comparable model evaluated SSQS. All analyses included the intent-to-treat sample using SPSS (version 22; SPSS Inc., Chicago, IL).

Results

Baseline demographics and clinical characteristics of treatment groups are shown in Supplemental Table 1 (available in online supplemental materials). Treatment groups did not differ on demographic and clinical characteristics at baseline (Supplemental Table 2). Participants were mostly male (94%) and approximately 30 years old (SD = 6.66), with moderately high levels of baseline PTSD (M = 58.96, SD = 7.860). SSQN and SSQS were moderately significantly correlated at baseline, r = .40, p < .001. SSQN was significantly negatively correlated with PTSD, r = −.27, p = .017 at baseline. SSQS was not significantly associated with PTSD, r = −.12, p = .289 or depression, r = −.18, p = .111 at baseline. Treatment groups did not differ in baseline SSQN (p = .242) or SSQS (p = .201). As reported in (Wagner et al., 2018), there were significantly greater reductions on the PCL-M for those in BA compared with TAU (F_{2, 114} = 5.87, p < .01; 3-month follow-up PCL-M, M = 47.90, SD = 13.44 for BA, M = 51.36, SD = 15.06 for TAU).

A significant Time by Treatment interaction in the mixed model suggested that individuals in BA for PTSD demonstrated a higher rate of change in the number of support persons compared with those in TAU from baseline to posttreatment (F_1,96 = 6.29, p = .014), controlling for marital status, gender, and baseline PTSD (table of coefficients located in Supplemental Table 3 in online supplemental materials). Inspection of untransformed means revealed that those in TAU experienced declines in numbers of social supports from baseline to posttreatment (baseline, M = 2.85, SD = 1.86; posttreatment, M = 2.25, SD = 1.79) whereas those in BA experienced increases (baseline, M = 2.36, SD = 1.85; posttreatment, M = 2.78, SD = 2.33; t = −1.94, p = .06). This effect was not retained at follow-up (F_1,96 = 0.22, p = .638). With regard to satisfaction with social support, although the overall regression of SSQS was significant (p < .001), only the SSQS at posttreatment significantly predicted SSQS at follow-up (β = .68, p < .001). Treatment condition was not significantly associated with improvement in social support satisfaction.

Discussion

The present study evaluated the degree to which veterans’ self-reported number of and satisfaction with social supports changed while participating in BA for PTSD as compared with TAU. Veterans randomized to BA for PTSD reported improvements in number of, but not satisfaction with, social supports from baseline to posttreatment. Notably, perceived number of supports increased, even without explicit focus on increasing social support as a feature of the BA for PTSD protocol (Wagner et al., 2008). Thus, improved number of social supports may occur naturally as a result of increased engagement in valued activities (e.g., Hopko & Mullane, 2008). As such, BA for PTSD may be beneficial as an initial step in treatment to increase the amount of social support available to persons with PTSD. Additionally, it is important to note that differences in improvements in the number of social supports among those participating in BA compared with TAU were not maintained at the follow-up assessment; those in BA experienced subsequent reductions in support, whereas those in TAU made gains. A greater focus on social support improvement may be needed to encourage more attention to and effort in maintaining supportive relationships. The one study reporting follow-up data on maintenance of social support gains in BA for depression demonstrated that gains persisted at 12-month follow-up (Hopko et al., 2011). However, numerous sample differences make direct comparisons to the present sample of traumatized veterans problematic. Future research should explore whether an explicit focus on social support enhancement in the context of BA for PTSD could engender longer-lasting changes.

Of note, those participating in TAU experienced minor declines in their perceived number of social supports from baseline to posttreatment, although these losses were recovered by follow-up. Although the BA group experienced declines in the number of supports from posttreatment to 3-month follow-up, they reported increases over baseline levels. Increasing activation of those with PTSD may be one way to prevent decrements of social support (e.g., Shallcross et al., 2016).

Limitations and Future Directions

Although this study revealed a promising secondary outcome of participation in BA for PTSD in the form of increases in the number of social supports immediately posttreatment, it is limited by a small sample and significant attrition from assessments (33% of the sample for BA and 34% for TAU). However, mixed-effects regressions are robust to missing at random data, and analyses suggest that assessments were missing at random. Additionally, the SSQN did not provide detailed information about the types of supports (e.g., friendships) that were gained, which could be useful for treatment planning. Third, the difference in change in number of supports between the BA condition and TAU condition may be attributable to the greater amount of treatment received by the BA condition relative to TAU. Future research should investigate whether unique features of BA are responsible for the increased support.

Fourth, because the SSQS was negatively skewed, a ceiling effect may have limited changes in satisfaction within the context of the study. Nonetheless, more research is needed regarding improvement of perceived satisfaction with support among veterans, potentially with different measures of social support. Additionally, the heterogeneity of treatment in TAU prevents direct comparison with outcome studies involving any individual treatment provided. Finally, we were unable to assess the mechanisms of action responsible for the increase in social support or whether changes in social support mediated improvement in PTSD (or vice versa). Future research should assess the specific features of BA that encourage enhancement of social supports as well as specific features of support that may be enhanced (i.e., increase in actual tangible support or recall of support) as well as whether there are meditational relationships between changes in social support and changes in PTSD in the context of BA treatment.

In sum, the present study provides valuable initial information regarding secondary benefits occurring during participation in BA for PTSD. Lack of social support is strongly associated with mortality (Holt-Lunstad, Smith, & Layton, 2010) and overall quality of life in addition to PTSD symptom maintenance (e.g., Schnurr et al., 2004); thus, interventions that improve psychological symptoms and social relationships may encourage long-term recovery.

Clinical Impact Statement.

The present study suggests that behavioral activation for posttraumatic stress disorder, a novel treatment that encourages participants to reduce avoidance and increase engagement in valued activities, also helps to increase social support. This treatment may be helpful for traumatized individuals who want alternatives to existing evidence-based treatments for posttraumatic stress disorder and who have deficits in social support.