(PS8-60) Differential Engagement with Digital CBT Impacts Depression Symptom Outcomes for Adolescents from Vulnerable Communities

Youth from minoritized backgrounds experience high rates of anxiety and depression, but most do not receive evidence-based treatment due to barriers like cost and stigma. Smartphone mental health apps offer low-cost, accessible interventions, but are rarely used after initial download, limiting benefits. SilverCloud, an app-based, clinician-guided, cognitive behavioral therapy (CBT) program, showed promising engagement among British youth, but its acceptability as an intervention for minoritized youth has not been evaluated. We examined feedback themes from adolescents from vulnerable communities who engaged with SilverCloud a high, average, or low amount to understand factors influencing use.

Method

Adolescents from New York City (aged 14-18) with moderate symptoms on the Patient Health Questionnaire-9 and/or the Generalized Anxiety Disorder-7 were recruited for an open trial of SilverCloud to assess acceptability and feasibility. Participants were interviewed about SilverCloud up to 4 times over their 8-week participation. Responses were coded, organized into themes, and analyzed using Dedoose, an app for qualitative and mixed methods research. Usage data were collected by SilverCloud. We used a data-driven approach to define user groups (low, average, and high engagement) based on pages completed on the app. 

Results

Participants (N=20) were mostly female (70%), Black (50%), and non-Hispanic (55%). On average, they viewed 45.45 pages and spent 104.04 minutes on SilverCloud. Low engagement users (n=6) completed < =13 pages (one module or less of 8 core modules), average engagement users (n=9) completed up to 55 pages (~ 5 modules), and high engagement users (n=5) completed >=93 pages (8 core modules, plus bonus content). On average, baseline PHQ-9 scores were 10.17 (low engagement), 14.89 (average engagement), and 17.2 (high engagement). At their final follow-up, PHQ-9 scores were 11.75, 10.89, and 7.6, respectively.

Three out of six low users did not complete a feedback call; all other participants completed at least one call. Themes from interviews included coach engagement, activities, content, functionality, perceived benefit, and engagement challenges. Almost all users believed the content was relevant to their concerns and felt the app was easy to use. Additionally, the majority of participants reported they did not engage with their coach and were unclear on the role of the coach or how to interact with the coach. About half of participants expressed desire for more interactive components, like games, to keep them engaged with the content. High engagement users were most likely to complete and enjoy the in-module activities and communicate with their coach.

Conclusion

Symptoms for the high engagement group improved more than the low and average engagement groups and they were the only group with a post-treatment PHQ-9 score under the clinical threshold of 10. Our results suggest that completing the in-module activities may be important for symptom improvement. Strategies to improve engagement with the activities and the coach may help to boost the clinical benefits of the program. Encouraging activity completion should be considered as an engagement strategy to improve participation and potential program benefits.