Parent and demographic predictors of participant engagement in a parental mHealth intervention: results from the Let’s Grow trial (Preprint)

Abstract

BACKGROUND Parents are integral in shaping early childhood health behaviors, and mobile health (mHealth) interventions offer an accessible method of supporting them in this role. Optimizing participant engagement is acknowledged to be key to mHealth effectiveness and impact; however, research examining personal predictors of engagement is largely lacking. OBJECTIVE The aims of this study were to i) describe participant engagement with a novel parental mHealth intervention (Let’s Grow) during the first 25 weeks of use, and ii) investigate whether engagement levels varied by family demographics and parent characteristics. METHODS This observational study utilized data from parents in the intervention group in the Let’s Grow trial (n=682). The intervention targeted toddlers’ movement behaviors and the program (a purposive-designed progressive web app) was delivered via modules that the parents worked through over time. The content was built around three main components (behavior change activities, information provision, and social support). Engagement data (web app analytics) collected across the first 25 weeks of the intervention were summarized as study specific (time using the app, proportion of accessed features and pages, clicks in the main parts of the app) and overall engagement measures (composite engagement index [EI]) and individual subindices [click depth, loyalty, recency, diversity]). Items measured at baseline included: family demographics (main carer, child and family characteristics, postcode) and parent characteristics (coping, concern, information seeking). Linear regression assessed associations between family demographics and parent characteristics and engagement measures. RESULTS Most parents logged in and used at least one app feature (89%). App access declined from 81% in the first week to 28% at 25 weeks. For users who engaged with the app during weeks 12-25, EI remained consistent and was virtually the same as the average EI over the assessed time period (28%, range: 3-50%). More work hours, parents living together, having siblings in the family and living in a regional/remote area were associated with lower engagement on 10 out of 12 indicators (all P≤.046). Higher education level was associated with higher engagement on 9 indicators (all P≤.020). Of the parent characteristics, only higher parent coping was positively associated with engagement (P=.003). CONCLUSIONS Our findings indicate that time and sociodemographic factors might be most relevant to predicting engagement and highlight the characteristics of parents who may benefit from more active strategies to support their engagement with digital interventions. The uptake and continued engagement with this app were higher than what is reported for apps more generally, but it is unknown whether this is sufficient for behavior change. Individual and composite measures of engagement yielded similar results, indicating that simpler metrics which might be more feasible for researchers to use compared to complex EIs, can be useful for reporting engagement in digital interventions. CLINICALTRIAL ACTRN12620001280998; U1111-1252-0599 INTERNATIONAL REGISTERED REPORT RR2-10.1136/bmjopen-2021-057521