(PS3-C64) Investigating Differential Treatment Response as a Function of Latent Topics in Treatment-seeking Youth’s Top Problems

Introduction: The Top Problems Assessment (TPA) is an idiographic measure that captures unique, person-centered information regarding treatment-seeking youth’s presenting concerns. The qualitative nature of the TPA presents methodological challenges that rely on heuristic coding systems and limit the applicability of this measure as a research tool. To address such concerns, the present study applies topic modeling to quantitatively extract common topics in youth’s presenting concerns. Further, we investigate if youth differentially respond to the Unified Protocols for Transdiagnostic Treatment of Emotional Disorders in Children/ Adolescents (UP-C/A) as a function of TPA topic.  

Methods: Participants were 172 treatment-seeking youth between the ages of 6-18 (M = 12.07, 50.58% female) with primary diagnoses of internalizing disorders. In the present study, we utilized Latent Dirichlet Allocation (LDA) to model the latent topics in youth-reported concerns identified by the TPA.  To examine treatment response, difference scores were first calculated for youth- and parent-reported symptoms of depression and generalized anxiety symptoms measured by the Revised Child Anxiety and Depression Scale. Following, we investigated differential treatment response as a function of TPA topic using analysis of variance and Bonferroni adjusted post-hoc group comparisons. 

Results: Results indicate that a LDA model with four topics yielded the highest coherence value. The youth topics were characterized as those relating to Worry, Avoidance, Emotional reactivity, and Context, with Worry being the most frequently endorsed topic (38.33%). Furthermore, results indicate significant differences in parent-reported symptom improvement as a function of the TPA topic for depression and generalized anxiety symptoms (p’s < .05). Specifically, post-hoc comparisons indicate that youth with Avoidance TPA topics exhibited a significantly greater reduction in depression symptoms compared to those with Worry TPA topics (Mdifference = 1.72, p = .05) following treatment. Similarly, youth with Avoidance TPA topics exhibited a significantly greater reduction in generalized anxiety symptoms compared to those with Emotional Reactivity TPA topics (Mdifference = 2.12, p = .03). There were no significant differences in youth-reported symptom reductions as a function of TPA topics. 

Discussion: Here we demonstrate not only the application of data-driven topic modeling to capture common topics in youth's presenting concerns but also the ability to leverage such topics to elucidate differential treatment responses. Here, we found that youth with Avoidance related top problems demonstrated the greatest reduction in symptoms of both depression and anxiety, which may speak to the exposure and behavioral activation components of the UP-C/A being particularly effective for such youth compared to those with fewer emotionally driven behaviors (e.g., Worry). Results of the present study lay the foundation for future work to further parse which specific treatment components may contribute to differential treatment responses in UP-C/A as a function of presenting concerns.