Understanding engagement with a novel smoking cessation smartphone application: an exploratory analysis of usage data

Abstract

Background: Engagement with smoking cessation smartphone applications (apps) is essential for them to be effective; however, the process of identifying appropriate parameters for measuring engagement is at an early stage. This study aimed to: 1) assess how far a range of engagement parameters discriminate between the active and control versions of a novel smoking cessation app; 2) explore participant characteristics that prospectively predict engagement. Methods: A subsample of 184 UK-based adult smokers, enrolled in the first wave of the BupaQuit trial, was randomly assigned to receive the intervention or control version of the BupaQuit app. Secondary outcomes included number of logins, time spent using the app, and content used. Negative binomial regression models were fitted to estimate relationships between app version and engagement parameters over 28 days. Findings: App version did not predict number of logins. Intervention app users spent more time using the app (IRR = 1.46, p < 0.001, 95% CI = 1.43-1.48) and used more content (IRR = 1.08, p = 0.008, 95% CI = 1.02-1.14). Being male, having higher educational attainment, and low nicotine dependence independently predicted engagement on all parameters assessed across experimental arms (all p’s < 0.001). Discussion: Using number of logins as the sole measure of engagement in randomised trials may not suffice to discriminate between apps. Time spent using apps and content used should additionally be considered ‘gold standard’ measures of engagement. The finding of baseline predictors of engagement suggests that the tailoring of content according to individual differences may improve engagement.