Contextual Predictors of Engagement in a Tailored mHealth Intervention for Adolescent and Young Adult Cancer Survivors.

Abstract

Despite the promise of mobile health (mHealth), engagement is often too low for durable health behavior change, and little is known regarding why certain individuals abandon mHealth tools. Guided by a mHealth engagement framework, we evaluated contextual predictors of objective engagement with an app for adolescents and young adults (AYA) who survived cancer. One hundred and ten AYA survivors (M age = 20.5, 43% female, 30% racial/ethnic minority) were randomized to receive a disease self-management app that delivered 1-2 tailored messages/day for 16 weeks, and contained a survivorship care plan (SCP). Demographic, disease, psychosocial, and setting characteristics were examined as predictors of three objective engagement outcomes: (a) % of active app days, (b) % of messages read, and (c) viewed SCP in the app versus not. A subsample (n = 10) completed qualitative interviews to further assess engagement barriers. Self-reported uninterrupted app access (β = -0.56, p < .001), iPhone (vs. Android) ownership (β = 0.30, p < .001), and receiving the intervention in the summer (β = -0.20, p = .01) predicted more active days. Lower depressed mood (β = -0.30, p = .047) and uninterrupted app access (β = -0.50, p < .001) predicted more messages read. Qualitatively, technical glitches and competing priorities were described as engagement barriers, whereas certain types of messages (e.g., health goal messages) were perceived as engaging. Among participants who had uninterrupted app access (n = 76), higher baseline motivation to change, better health perceptions, using the app during the summer, and iPhone ownership predicted higher engagement. Findings demonstrate the importance of comprehensively assessing and planning for multi-level ecological determinants of mHealth engagement in future trials. NCT03363711.