SCIENCE / SUMMARY / DIGITAL CBT-I

Do digital insomnia programs that run without a therapist help?

META ANALYSIS · RANDOMISED CONTROLLED TRIALS ■■■□ moderate evidence: 29 randomised trials, but most were open label, outcomes were reported by participants themselves and results varied greatly between trials

Who: 9,475 adults in 29 randomised controlled trials published up to March 2024 (4,847 assigned to an automated program, 73.3% of them women); mean age 45.7. 12 trials were run in Europe, 7 in the US, 6 in Oceania and 4 in Asia. People with sleep apnea, shift workers and people diagnosed with schizophrenia or bipolar disorder fell outside the inclusion criteria.
What they did: Digital programs without therapist guidance, with at least one cognitive and one behavioural component and lasting at least four weeks, were included. Comparisons were waitlist, placebo, sleep education, usual care and therapist supported CBT-I. The main outcome was an insomnia severity scale filled in by participants (mostly the Insomnia Severity Index). Risk of bias was rated with the Cochrane RoB 2 tool, and the protocol was registered in advance.
What they found: At the end of the program, the standardised mean difference in insomnia severity was -0.71 (95% CI -0.88 to -0.54; 32 comparisons), but variation between trials was very high (I² 91%). Against a waitlist the difference was -0.88. Therapist supported CBT-I did better than the automated programs by 0.61 (0.37 to 0.85; 3 comparisons). The effect on total sleep time in sleep diaries was small (0.19). The pooled completion rate across 19 trials was 55.9% (50.2% to 61.6%), ranging from 16.7% to 85.7% in individual trials. Completion rate did not significantly explain the size of the effect.
Limits: As the authors note: variation between trials was high for most outcomes and only partly explained by the type of comparison; of the 29 trials, 3 had low risk of bias, 19 some concerns and 7 high risk; most were open label and outcomes relied on what participants reported; adherence and quality of life were reported in only two trials and could not be analysed; the constant updating of software products could not be reflected. In addition: three papers from the same 1,149 person Australian trial appear as separate rows in the study table. Funding: Korea's Ministry of Food and Drug Safety and Dongguk University. No conflicts of interest reported.

What does this mean for you?

Structured, evidence based programs for insomnia can also be delivered digitally, and they helped in these trials; but therapist support widens the benefit and many people stop partway. If insomnia has gone on for a long time, talking through the options with a doctor can be a good start. This is not advice.

Link to Nefesly: Nefesly has no insomnia treatment program today. When considering a possible digital CBT-I module, it led us to keep in mind that only about half of people finish such programs and that therapist support widens the benefit.
Reference: Hwang JW, Lee GE, Woo JH, et al. Systematic review and meta-analysis on fully automated digital cognitive behavioral therapy for insomnia. npj Digital Medicine. 2025;8(1):157. doi:10.1038/s41746-025-01514-4 · Open access (CC BY-NC-ND 4.0).
Why chosen: One of 15 sources selected from 430 candidates on the digital CBT-I axis; a recent meta analysis looking at fully automated programs on their own.
Summarised by: Nefesly, from the full text, 11 October 2026. Awaiting clinician check.
Copyright: Summarised from the full text, not translated from the abstract.