(PS3-88) Identifying Moderators of Treatment Response in a Single-session Intervention for Depression

Background. Digital single-session interventions (SSIs; internet-based interventions that individuals complete in one sitting) could reduce the burden of depression because they are brief, low-cost, and readily available. However, an SSI called the Common Elements Toolbox (COMET), which incorporates elements from cognitive behavioral therapy and positive psychology, was not associated with significant improvements in depression compared to a waitlist control in our randomized controlled trial of 828 adults. In the present study, we aim to identify baseline characteristics that are associated with or moderate response to COMET.

Methods. Online Prolific workers were randomized to receive either COMET or a waitlist control. Candidate predictors were identified from the literature. Participants completed questionnaires assessing demographic (e.g., age, race, marital status) and clinical characteristics (e.g, insomnia symptoms, childhood adversity, emotion regulation strategies) at baseline. After 2 weeks, participants also completed the PHQ-8, an 8-item self-report questionnaire that assesses the frequency of DSM-5 major depressive symptoms. We used linear regression to identify predictors and moderators of treatment response.

Results. The model identified statistically significant associations (R² = 0.6385, F(71, 738) = 21.13, p < .001). Better physical health (β = 0.128, p < .001), more adverse childhood experiences (β = 0.079, p = .03), higher levels of insomnia symptoms (β = 0.172, p < .001), higher levels of anxious distress (β = 0.359, p < .001), lower levels of dysphoric distress (β = -0.289, p < .001), higher levels of impairment (β = 0.253, p < .001), and lower endorsement of using behavioral activation strategies in the past two weeks (β = -0.083, p < .03) at baseline were associated with higher levels of depressive symptoms after 2 weeks. Endorsement of using behavioral activation strategies (β = 0.133, p = .01) and gratitude (β = -0.194, p = .008) at baseline moderated the relationship between treatment group and symptom improvement.

Discussion. By identifying moderators of treatment response, researchers may be able to develop a treatment selection process that is able to match individuals to the digital SSI that will work best for them. In general, statistical and actuarial approaches to treatment selection appear to be more effective than clinician judgments.