0374 Does adding a chatbot and human support enhance treatment adherence to and efficacy of digital CBTi? A randomized controlled trial

Abstract

Abstract Introduction Digital cognitive behavioral therapy for insomnia (dCBTi) has been shown to be efficacious; however, treatment adherence and efficacy are suboptimal compared to CBTi delivered in person. The present study evaluated whether the addition of different types of coaching support (i.e., phone coaching by a therapist trained in CBTi, phone coaching by a research assistant not trained in CBTi, and virtual coaching by a chatbot that simulates human dialogue) would improve treatment adherence to and efficacy of dCBTi. It was hypothesized that the addition of any form of coaching will have beneficial effects, with therapist coaching achieving the best outcomes. Methods A single-blind, parallel, randomized controlled trial was conducted. One-hundred-fifty adults with insomnia, indicated by Insomnia Severity Index (ISI) >=10, were randomly assigned to one of the five conditions: dCBTi with three 20-min therapist phone coaching sessions (C1), dCBTi with three 20-min non-therapist phone coaching sessions (C2), dCBTi with chatbot virtual coaching (C3), dCBTi without any coaching (C4), and the sleep hygiene and sleep diary control (C5). dCBTi is a smartphone application adopting a six-session CBTi with components of sleep hygiene, sleep restriction, stimulus control, relaxation, and cognitive therapy. Participants completed sleep diaries, ISI, dysfunctional beliefs about sleep (DBAS), sleep-related behavior questionnaire (SRBQ), and measures of mood, fatigue, and quality of life at pre-intervention, post-intervention, and 4-week follow-up. Results Mixed modeling analysis showed that participants in C1, C2, C3, and C4 had greater improvements in insomnia symptoms, sleep efficiency, DBAS, SRBQ, mood, and fatigue than C5. Participants in C1 had significantly greater treatment adherence and improvement in DBAS and SRBQ than those in C4; those in conditions with any form of coaching had greater treatment adherence than those in C4. However, improvements in insomnia symptoms were not significantly different between conditions with or without coaching. Conclusion Adding human support, especially therapist phone coaching, improved treatment adherence to dCBTi but not efficacy. Nonetheless, adding therapist coaching led to greater improvements in the mechanisms of change, i.e., reductions in dysfunctional beliefs and behaviors related to insomnia, which may promote greater maintenance of treatment efficacy at longer follow-up. Clinicaltrials.gov registration#: NCT05136638 Support (if any) HKU Fund for Translation Research #104005654