JMIR Publications

ABSTRACT

Background:

Mobile health (mHealth) interventions are increasingly used to reduce risk and promote health in real-time, real-life contexts. Engagement is critical to mHealth intervention effectiveness yet may be challenging for young people experiencing depressive symptoms.

Objective:

We examined engagement with the 4-week mHealth component of a counseling-plus-mHealth intervention to reduce sexual/reproductive health (SRH) risk among young people with depression (“MARSSI”) to determine 1) mHealth engagement patterns over time and 2) how sociodemographic characteristics, SRH risks, and depressive symptom severity were associated with these engagement patterns.

Methods:

We undertook secondary analysis of data collected 6/2021–9/2023 in a multi-state randomized controlled trial of MARSSI vs. a breast health podcast. Eligibility included age 16-21 years; ability to become pregnant; smartphone ownership; English fluency; past-3-month penile-vaginal sex ≥1x/week and ≥1 SRH risk; and PHQ-8 score≥8. Intervention participants received 1-on-1 telehealth counseling, then used an app for 4 weeks, responding to surveys (3 prompted at quasi-random, 1 scheduled daily) about affect, effective contraception and condom use self-efficacy, sexual and pregnancy desire, and recent sex, and receiving tailored messages reinforcing the counseling. We computed mHealth engagement days (responding to ≥1 app survey) by week and overall. Latent trajectory analysis identified engagement patterns over the four mHealth weeks among participants with any engagement. Using regression analysis, we examined associations of sociodemographic characteristics, SRH risks, and depressive symptom severity with mHealth engagement patterns (p<.05) and examined moderation by depressive symptom severity. Of 201 intervention participants, 194 (96.5%) enrolled in the app.

Results:

Among those responding to any app surveys (n=167, 86%), median (IQR) engagement was 14 days (4-23); 33% responded on ≥20 days. App engagement declined over weeks 1-4: median (IQR) days 5 (3-7), 3 (1-6), 3 (0-6), and 1 (0-5). On latent trajectory analysis, four patterns of any app engagement emerged: high-throughout (29%), high-then-declining (24%), mid-then-declining (28%), and low-throughout (20%). Participants identifying gender other than female and those perceiving higher socioeconomic status were more likely to have high-throughout and/or high-then-declining engagement. Asian or Black non-Hispanic participants and those using low-effectiveness contraception were more likely to have no engagement. In the multivariable model, Asian and Black non-Hispanic remained significantly associated with lower engagement and higher perceived SES with higher engagement. There were no differences in engagement patterns by depressive symptom severity and no significant moderation.

Conclusions:

Young people with depressive symptoms showed initial high engagement with an mHealth app during an intervention to reduce adverse SRH outcomes. Methods to increase and sustain mHealth engagement and differences in engagement by sociodemographic characteristics warrant further study to optimize reach of mHealth interventions. Clinical Trial: ClinicalTrials.gov NCT04798248