Web-based cognitive behavioral therapy for insomnia outperforms other formats in adolescents

The analysis used a frequentist random-effects model to evaluate sleep outcomes: total sleep time (TST), sleep onset latency (SOL), wake after sleep onset (WASO), sleep efficiency (SE), and insomnia severity scores. Web-based CBTi consistently outperformed other formats across most measures. Compared with usual care, web-based CBTi improved total sleep time by 33 minutes, reduced time to fall asleep by 23 minutes, increased sleep efficiency by 7 percentage points, and reduced insomnia severity by 5 points on the severity scale. In ranking analysis, web-based CBTi ranked highest for total sleep time (86% probability), sleep onset latency (79%), sleep efficiency (74%), and insomnia severity (99%). No significant differences emerged for wake after sleep onset across modalities. The study found no meaningful differences between most other formats, suggesting web-based delivery is the key driver of benefit.

The network meta-analysis ranked interventions using P scores, a probabilistic metric: web-based CBTi achieved 86% probability of being best for total sleep time and 99% for insomnia severity—notably higher than face-to-face (which scored in the 40–60% range for most outcomes). The 33-minute gain in sleep and 23-minute reduction in sleep onset latency represent meaningful clinical improvements for adolescents whose cognitive development and mood depend on adequate sleep. The mechanism likely involves the scalability and accessibility of web delivery, the structured nature of digital CBTi protocols, and reduced barriers (no travel, scheduling flexibility, privacy). Interestingly, mobile-app-based and self-help formats did not match web-based CBTi's efficacy despite similar digital accessibility, suggesting that guided or semi-structured web delivery outperforms purely standalone approaches. The analysis included only RCTs, strengthening causal inference, though heterogeneity in treatment intensity, duration, and adolescent age ranges (not specified in the abstract) may have influenced relative rankings.