Adolescent Engagement With a Multicomponent mHealth Tool: Identifying Usage Patterns, Determinants, and Health Behavior Change in an Intervention Trial.

Abstract

Background: Research about the engagement of adolescents with mobile health (mHealth) interventions is scarce, while it is generally assumed that the engagement affects the intervention efficacy.

Objective: Using an mHealth intervention that targets the general population of adolescents to promote healthy behaviors (physical activity, low sedentary time, adequate sleep, and taking breakfast) and mental health, we aimed to investigate (1) how adolescents engage with the intervention, (2) which engagement styles can be identified and how these differ according to personal characteristics, and (3) which style of engagement predicts behavior change. The intervention used, #LIFEGOALS, includes self-regulation techniques, a support chatbot, narrative videos, and gamification, brought together in an app coupled to an activity tracker.

Methods: Logged usage data and self-reports of experience with #LIFEGOALS were collected from 159 adolescents (mean age 13.54, SD 0.95 years) over a 12-week intervention period and used to describe behavioral and experiential engagement with the intervention components over time. Baseline data on sociodemographic variables, mental health, and behavioral determinants were explored as determinants of engagement and were used to characterize engagement styles that were identified through exploratory cluster analysis on the frequency of usage of the components. Linear mixed-effects regression was used to analyze the effect of engagement style on health behavior change.

Results: Average time in the app was 26 minutes (SD 26) over the 12-week period, with usage decreasing substantially after the first week. The use of self-regulation techniques and gamification was strongly interrelated (0.65

Conclusions: Different engagement styles were identified based on the frequency and type of components used. Findings support the relevance of tailoring mHealth to individual, interpersonal, and contextual characteristics. The overall low engagement with the intervention may have limited the detection of differences in health effects between engagement styles.