(PS5-A21) Patterns of Engagement with a Smartphone-delivered Interpretation Bias Intervention

Background: High attrition rates are common among mental health (mHealth) interventions, with most users discontinuing use in the first two weeks. Engagement with mHealth apps is an understudied, yet critical, construct to understand in the pursuit of more efficacious app interventions. Given the potential of app delivery to provide accessible and scalable interventions, we sought to enhance engagement with a novel interpretation bias intervention app by incorporating strategies based on empirical support and stakeholder input. We then examined patterns of use to better understand behavioral engagement with the app.

Methods: HabitWorks was administered to 33 people traversing the challenging transition from acute psychiatric care to daily life. Participants were asked to use the following features for 1-month after discharge: (1) Interpretation bias modification (IBM) exercises prompted by the app 3x/week, (2) User-initiated “bonus exercises” as desired, (3) Weekly symptom (PHQ-9 & GAD-7) surveys prompted by the app, (4) User-initiated symptom surveys as desired, (5) HabitDiary, a free response diary at least 1x/week. Through visual data analysis of app usage (i.e., adherence to prescribed “dose” of the app) the first, second, and last author discussed and reached consensus to identify patterns of engagement in the month post-discharge.

Results: We identified 8 strategies to enhance engagement in HabitWorks: human support; customization and notifications; personalization; novelty; mood and tracking features; HabitDiary; feedback; privacy and data security. We will present descriptions of how these features were applied in the study. We identified five patterns of engagement: Consistently low (0-2 exercises completed/week; n= 5 (15%)), Adherent (12-14 exercises completed during month; n=12 (36%)), Drop-off (Adherent initially, then dropout; n= 4 (12%)), High diary (higher diary use than exercise completion; n= 3 (9%)), and Super user (15+ exercises completed during month, n= 9 (27%)).

Conclusion: This study highlights various strategies to enhance engagement with mHealth apps. Further, we have identified several novel patterns of engagement that emphasize the heterogeneity of engagement patterns and necessity of studying individual differences in engagement. Notably, almost two thirds of participants were categorized as adherent or super user. Given the typical pattern of high attrition with mHealth interventions, our high usage rates may be related to the engagement strategies adopted in our study. Future work is needed to test the temporal relationship between engagement strategies and usage outcomes. Our exploration of engagement with HabitWorks provides an example of how to operationalize engagement for other mHealth apps, with the hope of designing highly utilized, accessible interventions.