In 2021, a research team at Vanderbilt published a paper in Clinical Diabetes with a title that reads less like an abstract and more like a confession: “I’ve Had an Alarm Set for 3:00 a.m. for Decades.”
That title was a quote. From a real person. With Type 1 diabetes.
The study found that the most commonly reported reason adults with diabetes wake up at night is responding to their continuous glucose monitor. Almost a third — 30.8% — reported being woken by a false alarm at least once a month. (Bergner et al., 2021) Not a real low. Not a meaningful event. Just a sensor doing what sensors do, in the middle of the night, while their body tried to rest.
This is the strange new shape of life with diabetes. The technology that keeps you safe is also the technology that won’t let you sleep.
The tool is doing its job. That’s the problem.
It’s worth saying clearly: continuous glucose monitors are a miracle of modern medicine. Before CGM, hypoglycemia at night was something you hoped your body would catch. People died in their sleep. Parents stayed up in shifts. Spouses slept with one ear open for years. CGM changed that. It saved lives. It still does.
But every technology has a shape. And the shape of a CGM is fundamentally an alerting tool. It watches. It tells you. It expects you to act.
That model works in daylight. It collapses at 3am.
In 2013, in the Journal of Diabetes Science and Technology, Shivers and colleagues introduced “alarm fatigue” into the diabetes vocabulary — borrowing a concept that hospital intensivists had been writing about for years. They named it as a primary cause of CGM discontinuation. (Shivers et al., 2013) Not the price. Not the wear. The alarms.
A decade later, the data has only sharpened. Kulzer’s 2024 work documented hypoglycemic events on roughly 8.5% of nights for CGM users — and observed something quietly devastating: many users start ignoring or silencing their overnight alarms. (Kulzer et al., 2024) The very feature that justifies the device gets turned off so its owner can survive the week.
What gets lost when sleep gets fragmented
The cost isn’t only fatigue.
Martyn-Nemeth and colleagues, writing in Brain & Behavior in 2022, traced the downstream effects in families of children with T1D. Roughly 75% of pediatric hypoglycemia happens at night. (Martyn-Nemeth et al., 2022) That single fact has reshaped the inner lives of parents and partners — driving anticipatory waking, insomnia, and a kind of low-grade vigilance that doesn’t switch off when the lights do.
You learn to wake up before the alarm. The anxiety wakes you first.
Bergner’s 2022 work in Pediatric Diabetes added another sobering finding: among parents of newly diagnosed children, 68.8% reported poor sleep quality, and Pittsburgh Sleep Quality Index scores did not differ significantly between families using CGM and those who weren’t. (Bergner et al., 2022) CGM didn’t rescue their sleep. It rearranged the sources of their worry.
And then there’s the trade. In 2023, Lin’s team in the Journal of Diabetes Science and Technology gave it a clean shape: patients who keep their alerts on tend to show better Time in Range. (Lin et al., 2023) Patients who turn them off sleep better — and lose control of their numbers. That’s the choice the current toolset asks people to make every night. Pick one.
Singh’s 2024 work suggested a third cost layered on top of the other two: in some users, constant glucose monitoring may actually increase anxiety. (Singh et al., 2024) The data stream that was supposed to reassure becomes another thing to monitor.
The gap between alerting and acting
None of this is a failure of CGM. CGM is doing exactly what it was designed to do. It’s a sensor and a messenger.
The gap is between what the tool can deliver — information — and what a person actually needs at 3am: not to be woken up unless something requires them to act.
For competent adults managing their diabetes well, most overnight alerts don’t require action. They require awareness. And awareness, for a sleeping human being, is a violence. It costs you the next day.
This is the real cost ledger of the alerting era: real lives saved, real sleep lost, real anxiety created, real adherence eroded — all from the same well-meaning device. The ledger doesn’t balance because it was never designed to. It was designed to inform.
Toward tools that act
The next chapter in diabetes care isn’t a louder alarm or a smarter notification. It’s a different kind of tool entirely: one that acts during the hours its user can’t, and steps back during the hours they can.
That’s what sleep-only automation is for. It’s a category built around a simple recognition — that the most dangerous hours are the ones you spend asleep, and the right response to that danger is automation, not alerts.
But you don’t need to be sold on a product to take the research seriously. The point of the literature is that the alarm-based model has reached its limit. The people living with these devices have been telling us for years, in surveys and journals and quotes that become paper titles. They’ve had alarms set for 3am for decades.
It’s worth asking what they could do with the night back.