Defining the dose-response curve of digital CBT for insomnia: a model-based network meta-analysis of treatment duration and clinical effectiveness.

Digital CBT-I is efficacious for insomnia, but optimal intended dose and dose-response relationship are unclear. This study investigated the dose-response relationship and optimal dose of digital CBT-I for insomnia treatment. We included the randomized controlled trials (RCTs) investigating digital CBT-I and reporting Insomnia Severity Index (ISI). A model-based network meta-analysis was conducted to characterize the dose-response curve of digital CBT-I. The intended dose of digital CBT-I was defined as total treatment duration (number of sessions × minutes per session). The Emax function model was used to calculate the maximum efficacy and the median dose achieving 50% of maximum efficacy (ED50). The analysis (73 RCTs, 14,465 participants) identified a non-linear dose-response relationship between digital CBT-I duration and insomnia severity improvement, with maximum efficacy of -5.77 points on the ISI versus control (95%CrI, -7.51 to -4.68). ED50 was 44.55 (95%CrI, 3.20 to 137.23) minutes, while 250-min protocols demonstrated optimal effectiveness (mean difference -7.95; 95%CrI, -11.77 to -4.22). Effects plateaued beyond 300 min. Comorbid insomnia showed greater sensitivity (ED50 = 38.93 min) than isolated insomnia (ED50 = 62.26 min). Digital CBT-I at 250 min represents a promising candidate prescription for substantial insomnia improvement across populations, supporting its implementation as a preferred first-line low-intensity choice when performing adequately powered head-to-head trials.