0443 Time Course of Treatment Response in Digital CBT-I: Evidence for Rapid Early Improvement in Sleep Efficiency

Abstract Introduction Understanding the time course of treatment response in digital cognitive behavioral therapy for insomnia (dCBT-I) is critical for treatment optimization and patient expectations. Standard CBT-I protocols span 6-8 weeks, yet whether improvements occur gradually throughout treatment or concentrate in early sessions remains unclear. Identifying when therapeutic gains occur could inform optimal treatment duration and enable early non-responder identification. We characterized the week-by-week trajectory of sleep improvements across an 8-week dCBT-I program. Methods Adults (N=3,237) with insomnia enrolled in an 8-week dCBT-I program and completed daily sleep diaries. Sleep efficiency (SE) and total sleep time (TST) were measured weekly. Week-by-week changes were analyzed using repeated measures ANOVA. Response patterns were examined by calculating proportion of total improvement achieved by weeks 2, 4, and 6. Sample sizes varied by week due to progressive dropout (Week 1: n=2,230, Week 4: n=1,281, Week 8: n=528). Results Among all participants, SE improved from 74.7% at Week 1 to 87.0% at Week 8 (+12.3 percentage points total). The largest single improvement occurred between Week 1 and Week 2 (+9.0 percentage points, 73% of total improvement, p< 0.001), with subsequent weekly gains declining sharply (Week 2-3: +1.6%, Week 3-4: +1.5%, Week 4-5: +0.3%). By Week 4, participants achieved 98% of the total 8-week improvement (+12.1 of 12.3 points). TST followed a similar pattern, increasing from 6.13 to 6.84 hours over 8 weeks (+0.71 hours), with most gains (0.48 hours, 68%) occurring in the first 3 weeks. Post-Week 4, SE improvements plateaued with only +0.2 percentage points gained between Weeks 4-8. The rapid early response pattern was consistent across both objective (SE) and subjective (sleep duration) measures. Conclusion Digital CBT-I produces rapid early improvements in sleep efficiency, with 73% of total gains occurring in the first week and 98% by week 4. This rapid response pattern has important implications for treatment expectations, early non-responder identification, and potential for abbreviated treatment protocols. The early trajectory may also explain dropout patterns, as many participants achieve meaningful benefit before full program completion. These findings challenge the standard 6-8 week CBT-I duration and suggest opportunities for adaptive treatment approaches based on early response. Support (if any)