Low-intensity behavioral activation smartphone app to support low mood: description, engagement and log data analysis

Estimates indicate that over 300 million people worldwide have depression, with prevalence increasing around 50% over the last 30 years. Despite the availability of evidence-based psychological therapies, barriers to access remain in areas such as treatment availability, the burden associated with accessing services, and treatment sessions often only available at times suitable to the service rather than the patient. Harnessing the potential of a mobile phone app to deliver a low-intensity behavioral activation intervention to support low mood presents a scalable solution to address these barriers. Located in the development phase of the MRC Complex Interventions Framework, this study describes the Iona Mind Wellbeing for Mood (IMWM) smartphone app and reports log data recognizing patterns of engagement with specific factor techniques associated with behavioral activation. Consistent with other research on other digital health interventions, there was a large difference between people downloading IMWM (1725), engaging with it to be considered an Engaged User (281; 16%), and completing more than one PHQ-8 and GAD-7 (65; 23%) following assessment. Paired samples t tests were undertaken to explore the potential to reduce PHQ-8 and GAD-7 scores and the relationship with engagement. For those completing the PHQ-8 and GAD-7 outcome measures, there was a significant reduction in PHQ-8 (t(48) = 5.2; p < 0.01) and GAD-7 (t(48) = 3.5; p < 0.01) scores, respectively, with reductions greater with higher baseline severity. Users were able to successfully engage with the majority of techniques associated with behavioral activation, as well as with most of the wider features associated with a cognitive behavioral therapy (CBT) approach, excluding reflection. The algorithmically driven conversational agent was successfully employed to assist the user in breaking down activities to reduce difficulty and retain engagement. Despite these levels of engagement, however, engagement did not predict changes in scores for either the PHQ-8 [F(2,62) = 2.71; p > 0.05] or GAD-7 [F(2,62) = 0.59; p > 0.05]. These results encourage the ongoing use of log data to further enhance usability and engagement prior to undertaking a pilot randomized controlled trial to explore potential effectiveness.