In trials of digital interventions (DIs) for anxiety and depression, some degree of user engagement is necessary for effectiveness. Evaluating engagement with such interventions is therefore essential. A recent model by Perski et al. (2017) provides a conceptual framework that can be used to identify features of engagement; however, this model has not yet been tested in DIs for anxiety and depression. Aims This study aimed to i) examine the effectiveness of DIs for anxiety and depression, ii) assess the effect of engagement on intervention effectiveness, and iii) identify and quantify the influence of intervention attributes on engagement using the Perski model. Methods: We reviewed primary (i.e. symptoms of anxiety and depression) and secondary (i.e. engagement) outcomes of RCTs of DIs for anxiety and depression, conduct a meta-analysis to examine overall intervention effectiveness and then conducted a range of sub-group and moderator analyses. Results A random-effects meta-analysis (N = 39 RCTs) showed an overall effect in favour of DIs for anxiety and depression over controls. A moderator analysis showed engagement was a moderator of intervention effectiveness. Sub-group analyses identified a small number of intervention attributes that influence effectiveness however none of the secondary intervention attributes influenced engagement. Conclusions Digital interventions for anxiety and depression are effective compared with controls, but engagement is suboptimal with 29% of users dropping out of treatment. Intervention attributes that can improve engagement are frequently missing. The findings are discussed in terms of implications and design recommendations for researchers, practitioners, developers and policy makers .
|Publication status||Submitted - 2019|