Size of Social Network and Strength of Working Alliance Predict Perceived Valence of Social Rewards during Psychotherapy for Late-life Depression

Abstract

<h3>Introduction</h3> Processing of monetary and social rewards is inhibited in late-life depression¹. Social isolation predicts depression severity² and poor response to psychotherapy.³ Individuals with smaller social networks are more socially isolated and experience fewer socially rewarding interactions. Engagement in socially rewarding experiences predicts reduction in depression in psychotherapy for late-life depression.⁴ In psychotherapy, a positive therapist-patient relationship may enhance the experience of social rewards, and in turn lead to better outcomes. This study aimed to investigate whether the size of social network and the quality of the working alliance predict perceived valence of social rewards during psychotherapy. Our group developed a novel social reward task that measures response to social rewards during psychotherapy. We hypothesized that individuals with a larger social network and a stronger therapeutic alliance would demonstrate higher perceived social reward valence of feedback from their therapist. <h3>Methods</h3> Forty-five older adults with major depression (75.56% female; mean age = 69.44, SD = 9.60) participated in two randomized controlled psychotherapy trials for late-life depression. Participants received 9 weeks of either psychoeducation, Supportive Therapy, or "Engage" behavioral activation therapy⁵. Inclusion criteria included a primary diagnosis of major depression as determined by the SCID DSM-V and a depression severity score of ³20 on the Montgomery Asberg Depression Rating Scale (MADRS). At weeks 0, 3, 5, and 9, we measured the strength of working alliance with the Working Alliance Inventory (WAI) and the size of social network with the Social Network subscale of the Duke Social Support Index (DSSI). Perceived valence of social rewards was measured at weeks 0, 5, and 9 using our novel Social Task for Assessment of Reward (STAR), where participants received either positive or negative feedback from their therapist or an unfamiliar person and then rated the valence of the social feedback they received. We used mixed effects regression models, with fixed effect for the predictor (alliance or social network), time*predictor interaction, and random intercept for participant, to test whether social factors predicted participants' rated valence of the social reward feedback they received during the task. <h3>Results</h3> We found that a stronger working alliance predicted higher perceived valence of positive feedback from the therapist, <i>F</i>(1, 60.81) = 7.99, <i>p</i> = .006. We also found a significant interaction effect between size of the social network and time predicting perceived valence of positive feedback from the therapist, <i>F</i>(2, 17.98) = 5.06, <i>p</i> = .018. A larger social network predicted higher perceived valence of the feedback, and post-hoc analyses showed that this relationship increased over the course of the interventions. No significant effects were found for the feedback from the unfamiliar person. <h3>Conclusions</h3> Social factors – a strong working alliance and a larger social network – contribute to increased perceived value of positive feedback from a therapist and that this association increases during treatment. The strength of working alliance and size of social network were not associated with feedback from an unfamiliar person, suggesting that the working relationship with the therapist may serve as a socially rewarding experience. Our results underscore the importance of assessing social isolation as it may predict individuals' experience of socially rewarding interactions. Interventions could focus on increasing social rewards during psychotherapy by expanding patients' social networks and building a positive alliance. Future studies will focus on investigating the association between social factors, the perceived valence of social rewards, and treatment outcome. <h3>This research was funded by</h3> This study was funded by the National Institutes of Mental Health grants K23MH123864 and K23MH116105.