The Role of Early Engagement in a Self-Directed, Digital Mental Health Intervention for Adolescent Anxiety: Moderated Regression Analysis.

Background: Digital mental health (dMH) interventions offer the ability to reach many more adolescents with anxiety than face-to-face therapy. While efficacious dMH interventions are available for adolescents, premature dropout and low engagement are common, especially if delivered on a self-help basis without any form of therapist guidance. This is concerning, given that higher engagement, in terms of the number of sessions completed, has been repeatedly associated with improved clinical outcomes. The reasons for poor adolescent engagement in dMH programs are unclear. A clear understanding of when and why disengagement occurs is important in order to seek to improve engagement rates. Contemporary models consider engagement as multifaceted, comprising both "use" (eg, amount of content completed, frequency of use, duration spent logged into the dMH program, and depth of use, such as word or character count) and "user experience" (eg, interest and satisfaction in the program and affect and attention whilst engaging in the program). Objective: This study investigated the role of demographic and early engagement (EE) factors, specifically program use, in predicting overall program engagement and continued engagement, respectively, in a self-directed, internet-based cognitive behavioral therapy program for adolescent anxiety, namely, BRAVE Self-Help. It examined multiple measures of program use, including task completion, homework completion, and depth of response (character count of responses typed into program tasks). It also examined the moderating role of baseline anxiety severity. Methods: Data collected between July 2014 and May 2020 from 2850 adolescents aged 12 to 18 years who participated in BRAVE Self-Help were analyzed via a series of moderated regressions. Results: Results showed that EE (in terms of program use) was associated with continued engagement, demonstrated by early tasks (tasks completed in the first two sessions; R2=0.035; P<.001) and early depth (characters written in the first two sessions; R2=0.08; P<.001) predicting continued depth of program response (total character count of responses typed into all program tasks from sessions 3 to 10). Demographic factors and anxiety severity did not directly impact adolescents' engagement in BRAVE Self-Help. Conclusions: These findings highlight the need to investigate ways to (1) enhance EE and (2) better understand how to measure and capture all aspects of program engagement.