Sleep Guide, Part 2 of 4
The Night You Have, Not the Night You Wanted
Nocturnal hypoglycaemia, the alarms that wake the house, and the bidirectional loop where fragmented sleep and unstable glucose feed each other. What predictive low-glucose suspend changed in 2015. What advanced hybrid closed-loop changed again in 2022. What the work feels like even when the algorithm is doing it.
Ask Grace
Want to know what AID does overnight in your specific system, or what the pre-AID nocturnal-hypo benchmark looks like? Ask Grace.
When the algorithm reaches first
The alarm goes at 02:47. Someone in the house wakes, but only just. The treatment box is in reach because everyone knows where it lives. From the doorway of my own home and from the conversations I have in clinic, the work that families and adults living with T1D are doing overnight is one of the most under-acknowledged forms of healthcare labour in the country. It is invisible in the morning, and the morning glucose number rarely captures it.
The pre-AID benchmark
Before continuous glucose monitoring and before automated insulin delivery, this work was uncountable. It is now countable. The Juvenile Diabetes Research Foundation’s CGM study group analysed 36,467 nights of unblinded sensor data in 176 children and adults with T1D on intensive treatment. Hypoglycaemia events occurred on 8.5 percent of nights, with a median duration of 53 minutes; on 23 percent of those nights, glucose stayed below 3.3 mmol/L (60 mg/dL) for at least two hours (JDRF CGM Study Group 2010, Diabetes Care). That is the pre-AID benchmark. Lower baseline HbA1c and any hypo during a baseline blinded CGM week were the strongest predictors of how often nocturnal hypo recurred. Pump or multiple daily injections did not change the rate.
What the algorithms changed
The algorithms changed where this hypo could be intercepted. Buckingham’s predictive low-glucose suspend trial in 81 children with T1D ran 3,420 randomised in-home crossover nights. On suspend-active nights, time below 3.9 mmol/L (70 mg/dL) overnight halved across both age groups, with no rise in morning ketones and no severe hypo or DKA across the trial (Buckingham 2015, Diabetes Care).
Modern advanced hybrid closed-loop has gone further. The MiniMed 780G pivotal trial in 157 adolescents and adults saw HbA1c fall from 7.5 to 7.0 percent, time-in-range rise from 68.8 to 74.5 percent, and time below 3.9 mmol/L (70 mg/dL) fall from 3.3 to 2.3 percent, with closed-loop time averaging 94.9 percent and zero severe hypos or DKA across the study phase (Carlson 2022, Diabetes Technology and Therapeutics). The same direction repeats across thirty-one papers on the MiniMed 780G in children, adolescents, and young adults.
The pivot most families and adults describe is not “the system fixes it”. It is “the system reaches first”. The alarm still wakes someone, sometimes, but the treatment is smaller and the line returns faster. The under-2s in our paediatric clinic, where a 0.1-unit error matters and where the child cannot tell anyone they feel low, are the population for whom this shift matters most. AID does not remove the night. It changes what shape the night has.
The bidirectional loop, and what AID does not solve
What AID does not do is reach the people for whom it is not yet available. Provision in the UK is uneven; eligibility under NICE TA943 is real but slow; many families and adults are still on multiple daily injections or sensor-augmented pump without the closed-loop layer. Sleep is more fragmented in T1D than in non-diabetic peers, even on the most modern technology, because the alarms that protect against hypo are the same alarms that wake the house (Reutrakul 2016, Sleep Medicine). The bidirectional loop is real: poor glucose breaks sleep, broken sleep makes the next day’s glucose harder, which breaks the next night.
The other thing the night can be hiding
Obstructive sleep apnoea is more common in adult T1D than in non-diabetic peers (Banghoej and colleagues, 2017, Journal of Diabetes and its Complications, found around forty-six percent of a Danish adult T1D cohort had at least mild OSA on home polygraphy, with cardiac autonomic neuropathy the strongest predictor). The signs are loud snoring, witnessed pauses in breathing, daytime sleepiness, and unexplained morning rises that the algorithm cannot explain. If any of those describe a night in your house, that is a clinic conversation; STOP-BANG screening and an overnight study are the next steps your team can arrange.
Name the pattern, then take it to clinic
If your nights have a recurring pattern (the same window, the same shape, the same morning rebound), that pattern is the most useful thing you can bring to your diabetes care team. They can review the overnight basal, the timing of evening insulin, the relationship to evening exercise and alcohol, and where AID provision sits for you. The conversation is much more useful with the pattern named than with a general feeling that “something is wrong overnight”.
Part 2 of 4
The Night You Have, Not the Night You Wanted