Engagement in a web-based intervention for depression

Abstract

To date, much of the work into online health interventions has focussed on determining the effectiveness of the interventions or the development of new interventions for different populations. The current consensus is that these interventions can be effective, but little is known about the required level of exposure for an effective outcome to an intervention. Central to this is the concept of engagement, which covers the uptake and usage of the intervention once it has been offered. Whilst current research touches on engagement through reporting attrition rates or modules completed, it fails to provide insight into the relationship between these metrics and the intervention outcome. This thesis seeks to better understand this relationship through a mixed-methods approach using CREDO, a randomised controlled trial (RCT) of an online, cognitive behaviour therapy (CBT) based intervention to target depression in cardiovascular disease. For the purpose of this research, engagement was categorised into three areas: uptake, usage and persistence. Uptake refers to the users’ initial choice to engage with the intervention. It was assessed using quantitative analyses of users who chose to enter the trial versus those who did not, and through findings about the psychological drivers of uptake through the qualitative study. The role of usage (the level of user activity whilst online in the program) was studied through completing a systematic review of usage across online interventions, and through quantitatively analysing the association of participant usage with outcome in the CREDO RCT. Finally, factors that influenced persistence were explored in the qualitative study of user motivations for engaging with the program. A number of factors were found to positively influence uptake, usage and persistence in a web-based intervention for depression. Uptake was influenced by gender, language spoken at home, education and baseline level of depression. Baseline depression had the greatest influence on uptake, with those who had more severe depression being more likely to utilise the intervention. However, these factors explained little of the variance in the decision to take up the program in the trial. Ongoing use of the program was influenced by users’ ongoing analysis of its benefits, balanced against its ongoing costs or barriers to its use. In particular, users persisted with the program due to the hope that they, or someone else in the trial context, would obtain benefit from it. These beliefs were also found to significantly influence uptake. Findings from the systematic review indicated that a number of usage metrics were positively associated with outcome in web-based interventions. This thesis confirmed that a number of usage metrics were shown to be related, albeit weakly, to outcome. What appeared consistently is that a high level of program use will lead to the best outcomes, and that assuming there is a linear relationship between usage and outcome may be too simplistic. In summary, usage is best measured through mutliple metrics, with composite measures such as number of activities per login being the most meaningful to determine outcome. Individual factors and beliefs drive online behaviour and thus influence outcome; therefore, these must be considered in developing future programs. The thesis identifies how developers may need to fine-tune their design features to obtain optimal usage and to make their programs more effective.