Summary: In children and young adults (aged 7 to 25) with type 1 diabetes and suboptimal glycaemia (HbA1c at least 69 mmol/mol), automated insulin delivery (AID) over 13 weeks was compared with usual care for its effect on sleep. AID markedly improved glycaemia (HbA1c mean difference -34 mmol/mol) and one subjective measure (sleep-related impairment, MD -4.7), but objectively reduced sleep efficiency (MD -4.2 percentage points) and increased wake after sleep onset by 21 minutes. Sleep was a secondary outcome.
PICO Summary
| Element | Detail |
|---|---|
| Population | 80 children and young adults (aged 7 to 25 years) with type 1 diabetes and suboptimal glycaemia (HbA1c at least 69 mmol/mol); open-label randomised controlled trial, New Zealand. Sleep was a secondary outcome (64 provided subjective and 40 provided objective sleep data). |
| Intervention | Automated insulin delivery (advanced hybrid closed-loop) system for 13 weeks. |
| Comparison | Usual care (standard insulin delivery). |
| Outcome | Subjective sleep-related impairment improved more with AID (MD -4.7; 95% CI -8.7 to -0.8; p = 0.02), with no between-group difference in other subjective measures. Objectively (accelerometry), the AID group had 21 more minutes of wake after sleep onset (MD 21; 95% CI 1 to 41; p = 0.04) and a 4-percentage-point fall in sleep efficiency (MD -4.2; 95% CI -8.0 to -0.3; p = 0.03). HbA1c improved substantially (MD -34 mmol/mol; 95% CI -43 to -25; p < 0.001). No clear relationship was found between glycaemic improvement and sleep change. |
AID and sleep in youth with very high HbA1c (CO-PILOT)
RCT · type 1 diabetes · 13 weeks
AID sharply lowered HbA1c but did not improve sleep: objective sleep efficiency fell and time awake after sleep onset rose, while only one subjective measure (sleep-related impairment) improved.
Expert Commentary
The honest verdict from these prespecified secondary outcomes is that AID delivered a large and expected glycaemic benefit but did not translate into better measured sleep, and on the objective measures sleep was modestly worsened. Self-reported sleep-related impairment improved, yet every other questionnaire domain was unchanged, while accelerometry showed reduced sleep efficiency and more time awake after sleep onset. The plausible explanation is that earlier device pump and sensor alerts and overnight micro-adjustments disturb sleep even as glucose control tightens, and the authors are appropriately candid that the glycaemia-sleep relationship remains unclear. The chief limitation is that sleep was secondary and statistical power was modest, with objective accelerometry available in only 40 of 80 participants, so these signals are exploratory and the confidence intervals brush against the null. The trial was necessarily open-label, since a closed-loop device cannot be masked against usual care, which can colour subjective reports. Can I use this with my patients? Yes for the central message: when starting a young person on AID, glycaemic gains are reliable, but I would set expectations that sleep may not improve and could feel more fragmented at first, and I would ask directly about alarms and overnight alerts. Sleep in young people with type 1 diabetes looks multifactorial, and clinicians should pair device optimisation with behavioural sleep support rather than assume better glucose fixes rest. Dedicated, adequately powered sleep trials are needed.
References
Michaels VR, Boucsein A, Haszard JJ, Galland B, Meredith-Jones KA, Zhou Y, et al. Impact of automated insulin delivery on subjective and objective sleep in children and young adults with very high HbA1c: Sleep outcomes from the CO-PILOT randomised controlled trial. Diabet Med. 2025;42(8):e70093. doi:10.1111/dme.70093
