(PS2-A13) Working Alliance as a Potential Mediator of Change in Positive Affect with Behavioral Activation and Exposure-based Therapy

Background: Anxiety and depression are often characterized by high negative affect and low positive affect (PA). Existing behavioral interventions primarily, and successfully, target the negative affect system. While they have also been found to have benefits for PA, these effects are smaller and post-treatment self-report of PA often remains below normative levels. Delineating treatment mechanisms contributing to individual differences in PA changes could inform treatment optimization. Therapeutic working alliance (WA) has been consistently linked to better negative affect treatment outcomes, irrespective of treatment type. Thus, the current study sought to explore whether alliance is a specific mediator of change and, therefore, a possible mechanism of PA treatment response.

Method: The current study includes a total of 94 (mean age = 36.4; 82 female) treatment seeking adults meeting criteria for Generalized Anxiety Disorder (GAD) who were randomized to receive 10 sessions of either Behavioral Activation (BA; n = 48) therapy or Exposure (EXP; n = 46) therapy. Participants completed the NIH Toolbox Positive Affect measure at pre- and post-therapy, as well as the Working Alliance Inventory – Client version at week 6 of treatment. Linear mixed-effects (LMEs) models were used to examine the impact of treatment arm on PA (PA ~ time x treatment) and WA (WA ~ time x treatment). A moderated mediation model was tested using the ‘bruceR’ package in R. Post-treatment PA was the outcome variable, pre-treatment PA was the predictor, WA was the mediator; the moderating effect of treatment-arm (BA, EXP) on the association between pre-treatment PA and WA, and on the association between WA and post-treatment PA, was tested. Age, gender, and years of education were included in the model as covariates.

Results: Positive affect scores increased from pre- to post-therapy across both treatments (p = < .001), though post-treatment PA scores remained below normative levels for both BA (mean T-score = 42.38; SD = 5.96) and EXP (mean T-score = 41.76; SD = 7.59). There was a significant direct effect of pre-treatment PA on post-treatment PA (β = 0.428, p < 0.001). There was no significant indirect effect of pre-treatment through WA on PA, pre-treatment PA was not a significant predictor of WA, and treatment-arm did not moderate either relationship. The effect of WA on post-treatment positive affect was significant for both EXP (β = 0.576, p < .001) and BA (β = 0.297, p = .010).

Conclusions: Results indicate that regardless of pre-treatment PA, WA contributes to increases in post-treatment PA for both BA and EXP-based behavioral therapies. Results suggest that WA may serve as one potential mechanism of change that contributes to PA outcomes, but is likely not the only contributor. Identifying strategies for fostering optimal working alliance (e.g., therapist interaction style, cultural humility/empathy, client collaboration, desire to invest in the therapeutic process) within manualized behavioral interventions may be an important aspect of enhancing positive affect outcomes.