The development and application of an engagement index on the participants’ use of an infant feeding app: the Growing healthy program

Author: 
Taki, S Lymer, S Ong, K.L. Campbell, K.J. Russell, C.G. Laws, R. Denney-Wilson, E.
Publication type: 
Conference presentation
Program: 
Stream 1 Families with young children
Year: 
2016
Behaviour Change Conference: Digital Health and Wellbeing, London

Background and Aims

Mobile health (mHealth) interventions have great potential to promote health (Denney-Wilson, 2015). To date, mHealth studies using apps have focused on intervention outcomes and have neglected to understand participants’ interactions with apps including how much, when and with which parts of apps participants engage. This is important in helping to determine the dose of the intervention they receive. We adapted an Engagement Index from the discipline of marketing to assess the levels at which mothers in a healthy infant feeding mHealth intervention (Growing healthy) interacted with the  Growing healthy app.

Methods/Results

The Engagement Index (EI) tool developed by Web analytics Demystified (Peterson & Carrabis, 2008) was adapted and used to measure how participants engaged with the Growing healthy app. The EI tool comprises five sub-indices designed to capture a range of participant behaviours: Click-Depth Index (Ci) describes the number of pages accessed each time participants visit the app (Ci= Sessions having at least ‘n’ page views / All Sessions); Recency Index (Ri) measures the days elapsed since the participant last accessed the app (Ri= 1/Number of days elapsed since the most recent session); Loyalty Index (Li) measures the frequency of app access over the program (Li= 1 - (1 / Number of visitor sessions during the timeframe); Interaction Index (Ii) measures the number of push notifications opened from those sent (Ii= Sessions where visitor completes an action / All Sessions); and Feedback Index (Fi) is a subjective indicator of the participant’s satisfaction with the app (Fi= number of positive responses/number of survey questions completed). Participants’ subjective satisfaction with the app was assessed from a quantitative survey (questions included: ease of navigation, readability, quality and usefulness of the content on the app) this score comprised the Fi. The total participant EI score was then calculated as the average across the five sub-indices, thus providing a scale ranging from disengaged through to highly engaged. Modelling will be done to establish the strength of the relationship between the EI and intervention outcomes, whilst controlling for co-variates such as parental age. Secondary analysis will be undertaken to consider the strength of associations between each sub-index and study outcomes. 

Conclusion 

MHealth interventions delivered by apps provide the opportunity to investigate participants’ engagement with the intervention and its constituent parts. The use of an Engagement Index may help researchers to understand how participants engage with such an intervention, the trends in engagement over the course of the program, and whether the levels of engagement affect intervention outcomes.