(PS13-C74) Disengagement from a Moderated Online Social Therapy Platform for Psychosis: A Time-to-event Analysis

Disengagement in digital mental health interventions is a common occurrence that disrupts a digital platform’s potential for behavior change. Despite the demonstrated potential for digital interventions to enhance the mental health and well-being of vulnerable populations like young adults with serious mental illnesses (Biagianti et al., 2017), research identifying when disengagement occurs and factors that heighten the risk of disengagement from a digital intervention is limited. Thus, we analyzed the usage activity of 37 young adults with a psychotic spectrum disorder participating in an ongoing six-month pilot study of the Horyzons digital platform. We aimed to (a) characterize the median time until first disengagement from the platform and to (b) identify predictors of the risk for disengagement from a set of baseline variables including demographic characteristics (e.g., age, sex), social-related factors (e.g., loneliness, perceived social support, dimensions of social media-related social anxiety), and self-anticipated engagement. We defined disengagement as a continuous period of at least two weeks of no logins or activity on the platform. Preliminary analyses using Kaplan-Meier curves and Cox proportional hazard models suggested that the median time to disengagement was approximately 47 days, with nearly half of the sample (44%) first experiencing a period of at least two weeks of disengagement by the end of the first month of the intervention. At any point in time, older age at study entry was associated with a decreased risk of disengagement (hazard ratio [HR] = .68, 95% CI .57 – .80; p < .001). Female participants, when compared to males, were approximately 10 times more likely to disengage from the intervention at any given time (HR = 10.15, 95% CI 2.77 – 37.18; p < .001). We also observed that, at any point in time, the risk of disengagement was reduced by higher baseline perceived social support (HR = .94, 95% CI .88 – .99; p = .037), higher baseline loneliness (HR = .92, 95% CI .84 – .99; p = .039), higher baseline social media interaction anxiety (HR = .83, 95% CI .73 – .95; p = .006), and higher anticipated engagement (HR = .14, 95% CI .04 – .50; p = .002). Additional analyses are forthcoming. We will discuss the implication of these findings for recruitment and retention of adults with psychotic disorders in digital interventions as well as efforts to provide additional support to subpopulations at risk of early disengagement.