A vulnerable group that's often overlooked
Older adults make up a large share of hospital admissions. Among them, the overlap of diabetes and cognitive impairment (problems with memory, thinking, or decision-making) is significant. Cognitive impairment makes diabetes harder to manage because patients may forget to eat, not report symptoms of low blood sugar, or be unable to communicate how they're feeling.
Standard hospital glucose monitoring relies on point-of-care testing (POCT) — a finger prick that measures blood sugar at that moment in time. Done a few times a day, it misses everything in between. For someone who can't tell you they feel shaky or confused, those gaps can be dangerous.
Why continuous monitoring seems like a natural fit
Continuous glucose monitors (CGMs) are small wearable sensors — usually worn on the upper arm or abdomen — that measure glucose levels under the skin every few minutes, around the clock. They're widely used by people with diabetes at home.
But here's the twist: CGMs have mostly been studied in younger, cognitively intact outpatients. Almost no one had tested them in elderly hospital patients with cognitive impairment. Would they be accurate enough to be clinically useful in that setting?
Think of a CGM as a tiny lifeguard stationed just under the skin. Instead of waiting to be called, it constantly watches the water — sending a reading every few minutes, day and night. When glucose drops or spikes outside a safe range, it logs the event automatically.
Traditional finger-prick testing is more like a snapshot taken a few times a day. It catches what's happening right then, but misses the full story of what glucose levels are doing between checks.
What the study tested
Thirty-two older patients with both type 2 diabetes and cognitive impairment were recruited from a UK tertiary care hospital (a large specialist hospital). All were new to CGM. They wore blinded Dexcom G7 sensors — blinded meaning clinical staff couldn't see the CGM readings in real time — for up to 10 days, while continuing their usual care including standard finger-prick tests. Researchers then compared the CGM readings to the finger-prick results.
The CGM showed clinically acceptable accuracy. The mean absolute relative difference (MARD) — a standard measure of how close sensor readings are to actual blood glucose values — was 17.4%. That's within the range considered acceptable for clinical decision support. A strong correlation was found between the two methods (R²=0.82), and 99.3% of paired readings fell in the two safest accuracy zones.
Crucially, the CGM detected more episodes of low blood sugar (hypoglycemia), particularly overnight. These were events that the standard finger-prick schedule simply didn't catch. In a patient who can't reliably report symptoms, that undetected low blood sugar is a real safety risk.
These overnight lows could cause serious harm in patients who can't recognize or report symptoms on their own.
What the gap in standard testing looks like
The study found that CGM-based and finger-prick-based time-in-range (the proportion of time blood sugar stays in the healthy zone) were broadly similar: about 36% vs. 41%, respectively. But those overall numbers masked an important difference. CGMs revealed low-sugar events at night that were invisible to the daytime-only testing routine.
If you're a caregiver or family member of an older adult with diabetes and cognitive impairment — or if you're planning for your own care — this research matters. CGMs are already commercially available and approved for home use. Their use in hospitals is still evolving, but this study adds evidence that they may be especially valuable in patients who can't self-monitor or communicate symptoms.
If your loved one is admitted to hospital, it's worth asking whether continuous monitoring might be appropriate, especially overnight.
Limitations to be aware of
This was a small study — 30 participants who completed it — at a single hospital. The CGM sensors were blinded, meaning clinicians couldn't use the data in real time during the study. A larger trial where staff can act on CGM readings in real time would give a fuller picture of the benefit. Results may also vary with different CGM devices or in different hospital settings.
This study lays important groundwork for a clinical trial where CGM readings are actually used to guide care decisions in real time for this patient group. If larger trials confirm these findings, CGM could become a standard tool for managing glucose in vulnerable inpatients — reducing silent hypoglycemia and giving clinical teams a much clearer picture, day and night.