News · Sleep
Longer sleep steadied next-day blood sugar in teenagers
A SLEEP study paired wrist trackers with continuous glucose monitors across 2,245 days in 206 Danish 18-year-olds and found sleep and glucose pushing on each other in both directions.
Based on a peer-reviewed cohort study in SLEEP
- Researchers studied 206 Danish 18-year-olds from the COPSAC2000 birth cohort, publishing in SLEEP.
- 2,245 person-days of overlapping wrist-tracker sleep and continuous glucose monitoring, a median of 13 days each.
- Both measures are objective. Neither sleep nor glucose was self-reported.
- Each extra hour of sleep tracked with slightly higher next-day glucose, but lower variability and lower risk of extreme swings.
- The direction runs both ways: higher daytime glucose variability predicted shorter sleep the following night.
- Within-person night-to-night changes predicted next-day glucose, not just differences between people.
- A pre-wake glucose rise explained about 5% of the link between longer sleep and higher next-day glucose.
- These are healthy adolescents, not people with diabetes, and the effect sizes are small.
Sleep and blood sugar are known to be connected, mostly from laboratory studies where researchers deprive volunteers on purpose. A study in SLEEP watched it happen in ordinary life instead, strapping trackers and glucose monitors to teenagers for a fortnight.
Each additional hour of sleep was associated with higher next-day glycaemic concentration, lower variability, and reduced deviation risk.
The gap that laboratory work leaves
Controlled deprivation answers one question well and another badly. Sleep duration influences metabolic health, but the impact of daily sleep variation on next-day glycaemia in healthy individuals under real-world conditions is poorly understood.
Nobody in real life is randomized to four hours. They simply have a bad Tuesday, and until now the metabolic consequence of an ordinary bad Tuesday had not been measured across enough days to see.
Two sensors, thirteen days, no questionnaires
The team studied 206 adolescents from the Copenhagen Prospective Studies on Asthma in Childhood 2000 cohort, all around 18 years old.
The measurement is the strength here. There were 2245 person-days of overlapping accelerometer-derived sleep and continuous glucose monitoring recordings, a median of 13 days per person. Sleep came from wrist accelerometry and glucose from a continuous monitor, so neither number depends on anyone remembering anything.
Three measures, moving in different directions
The headline is a split result, and reading only the first part would mislead.
Each additional hour of sleep tracked with a slightly higher median glucose the next day. It also tracked with lower variability and reduced deviation risk, indicating lower risk of extreme glucose excursions.
So the average edged up while the swings settled down. For metabolic health the second usually matters more than the first, which is why reporting one without the other would give the wrong impression.
The loop, not the arrow
The more unusual finding is that the relationship runs in both directions within the same people.
Within-person deviations in sleep predicted next-day glycaemic concentration and deviation risk, meaning a person’s own bad night predicted their own next day, independent of what kind of sleeper they generally are.
And the reverse held. Higher daytime glycaemic variability predicted shorter subsequent sleep.
There is even a candidate mechanism for part of it: the early-morning pre-wake glucose rise partly mediated the link between longer sleep and higher next-day median glucose, accounting for about 5% of it.
What 206 healthy teenagers cannot show
The effects are small. A fraction of a milligram per decilitre per hour of sleep is a real signal in 2,245 days of data and not something anyone would feel.
Everyone here was a generally healthy adolescent, so nothing transfers automatically to adults, to people with diabetes, or to the clinical decisions either group faces.
And within-person comparison, while strong, cannot control what else differed on a given day. A late meal, a workout or an illness moves both sleep and glucose.
What the study establishes is that the coupling is dynamic and two-way in ordinary life, which is a better description than the one-way street the laboratory work implied.
People also ask
Longer sleep raised glucose. Is that bad?
Not straightforwardly, which is why the study reports three measures rather than one. Longer sleep tracked with a slightly higher median glucose but with lower variability and a lower risk of extreme excursions. Steadiness generally matters more for metabolic health than the average, so the overall pattern favours longer sleep even though one number moved the other way.
What is glycemic variability and why track it separately?
It is how much glucose swings across a day rather than where it sits on average. Two people can share an average and have very different days, one steady and one spiking and crashing. Continuous monitors made variability measurable outside a laboratory, and it tracks with metabolic risk somewhat independently of the average.
What does bidirectional mean here?
That the arrow runs both ways within the same people. Longer sleep predicted steadier glucose the next day, and a more variable glucose day predicted shorter sleep that night. It is a loop rather than a one-way effect, which is unusual to catch and only possible because both were measured repeatedly in the same individuals.
Does within-person analysis solve the confounding problem?
It helps considerably. Comparing a person's good-sleep nights against their own bad-sleep nights removes everything stable about them: genetics, income, body composition, habits. What it cannot remove is whatever else differed on those particular days, such as illness, exercise or a late meal.
Does this apply to adults or to people with diabetes?
Not directly. Everyone here was a generally healthy 18-year-old, and adolescence has its own sleep and metabolic physiology. The authors frame it as a critical period for establishing lifelong patterns rather than as a finding about diabetes management.