Digital cannabis tools may help even without long-term engagement

Understanding and optimizing engagement in digital interventions remains a significant challenge. This study aims to identify different patterns of engagement among users of a digital intervention to reduce cannabis use. We conducted a secondary analysis of engagement data from the intervention group of a study on the effectiveness of a digital cannabis intervention (ICan). Engagement patterns were identified through a latent class analysis using eight engagement indicator variables. The bias-adjusted three-step approach was used to examine differences in baseline characteristics across classes and associations with average change-from-baseline scores in cannabis use frequency and quantity. Three latent classes were identified: Class 1 (32%), 'non-engagers', showed minimal to no engagement; Class 2 (41%), 'shorter-term engagers', showed moderate engagement in the intervention, but with a shorter duration than recommended; Class 3 (27%), 'long-term engagers', showed high engagement in the intervention with a duration meeting the recommendations. The proportion of males was significantly higher in the 'non-engagers' class compared to the others. The 'long-term engagers' class reported fewer tobacco use days at baseline compared to the others. No differences were found between classes regarding average change-from-baseline scores in cannabis use frequency and quantity. Users of digital interventions show distinct engagement patterns, and characteristics such as male gender and tobacco use may predict sub-optimal engagement. Importantly, also sub-optimal exposure to a digital intervention may be associated with changes in cannabis use, as higher engagement does not necessarily lead to greater effectiveness.