Digital cognitive behavioural therapy for insomnia in women with cancer: The Sleepio After Cancer (SAC) randomised controlled trial

Sleepio After Cancer

Patient / Population Intervention / Exposure Comparison Outcome

In women with cancer and insomnia, digital cognitive behavioural therapy for insomnia improved the Sleep Condition Indicator score at week 24 compared with sleep hygiene education.

N

308patients

Design

Two-arm, open-label randomised trial in which 308 women with a current or prior cancer diagnosis and insomnia (Sleep Condition Indicator [SCI] score ≤16) were randomised 1:1 to digital cognitive behavioural therapy for insomnia (dCBT-I) delivered via Sleepio or sleep hygiene education (SHE), each plus treatment as usual (NCT05816460).

Endpoint

Co-primary endpoints at Week 24: change in SCI score and the proportion achieving SCI score >16 (remission).

ResultSCI scores improved significantly more with dCBT-I vs SHE at Week 24 (mean difference, 6.36 points; 95% CI 4.96-7.76; P<0.001; Cohen's d=0.99). Remission was achieved by 27% vs 12% (ITT, NNT=7) and 62% vs 19% of completers (NNT=3), robust to tipping-point sensitivity analysis. Treatment benefit was consistent across cancer subgroups (interaction P=0.922) and anti-endocrine therapy status (interaction P=0.313). Fatigue, depression, anxiety and hot flash interference improved significantly more with dCBT-I (effect sizes 0.48-0.77). In the dCBT-I arm, 126/156 (81%) commenced the programme and 61/156 (40%) completed six sessions. Week 24 completion was 69 (44%) and 52 (34%).

Brennan DJ, Treacy T, Zhao J, et al. Presented at ESMO Congress 2026, abstract LBA91, 26 October. Not yet published.

Discussion & critique

This open-label randomised trial shows a large improvement in SCI score with dCBT-I over sleep hygiene education at Week 24. For the co-primary remission endpoint, the abstract gives proportions and NNTs but no CI or P value. Week 24 completion was only 69 (44%) and 52 (34%), so the estimates depend heavily on mixed models under a missing-at-random assumption, although the remission result is reported as robust to tipping-point sensitivity analysis. Follow-up ends at 24 weeks, which limits how far the claim of sustained benefit can reach, no safety or adverse event data are reported, and one author is employed by Bighealth.