Accuracy of subcutaneous continuous glucose monitoring in critically ill adults receiving intravenous insulin: a prospective observational study
This prospective observational study of 55 critically ill adults receiving intravenous insulin found that subcutaneous continuous glucose monitoring demonstrated acceptable numerical and clinical accuracy compared to arterial and point-of-care reference methods across varying sepsis severities, though its reliability during hypoglycemia remains unconfirmed due to limited data.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are driving a car on a winding mountain road at night. Your headlights are bright, but they only show you the road right in front of the bumper. To drive safely, you need to know what's coming up a few seconds ahead, not just what's under your tires. In a hospital's Intensive Care Unit (ICU), patients are like those cars, and their blood sugar is the road. Doctors need to see the "road ahead" to adjust insulin, the fuel that keeps the engine running smoothly. Usually, they have to stop the car, pull out a test strip, and check the fuel gauge manually every few hours. This is like checking the road only when you hit a bump; you might miss a sudden cliff or a sharp turn.
Enter Continuous Glucose Monitoring (CGM). Think of this as a high-tech radar system that constantly scans the road ahead, giving a live feed of the terrain. Instead of a single snapshot, it offers a movie of the patient's sugar levels. But here's the catch: this radar doesn't look at the fuel in the tank (the blood); it looks at the fuel in the soil around the tank (the fluid under the skin). Sometimes, the soil takes a little longer to react than the tank does, especially if the ground is frozen or the engine is overheating. The big question for doctors is: Can we trust this radar when the patient is in a crisis, like when they have a severe infection called sepsis, which is like a storm battering the car? If the radar gets confused by the storm, doctors might make the wrong turn, which could be dangerous.
This study, conducted by a team at Southern Medical University, decided to put that radar to the test in a real-world storm. They watched 55 adults in the ICU who were already sick enough to need intravenous insulin to control their high blood sugar. The researchers wanted to see if the CGM radar matched up with the "gold standard" manual checks: the blood gas tests taken from an artery (the most accurate way to check the tank) and the quick bedside finger-prick tests. They tracked the patients for several days, comparing the radar's live feed against these manual checks thousands of times.
The results were mostly good news for the radar. When they compared the CGM numbers to the manual blood tests, the radar was surprisingly accurate. The average difference between the radar and the manual check was about 11% when compared to the artery test, and even better at 9.6% when compared to the bedside finger-prick test. To put that in perspective, if the manual test said the sugar was 100, the radar was usually very close. Even more importantly, when they plotted these numbers on a safety map (called a Clarke Error Grid), 99.7% of the readings landed in the "safe zones." This means that in almost every case, the radar gave a reading that would lead a doctor to make the correct decision, rather than a dangerous one.
However, the study also found where the radar starts to get fuzzy. The researchers looked at whether the type of sickness mattered. They compared patients with no infection, those with sepsis, and those in septic shock (the most severe form). Surprisingly, the radar performed just as well across all these groups. It didn't matter if the patient had sepsis or not; the radar stayed reliable. But, the study did find a specific condition where the radar struggled: severe metabolic chaos. When patients had very low blood pH (acidic blood) or extremely high lactate levels (a sign of severe stress on the body), the radar's accuracy dropped significantly. In these specific, very sick cases, the difference between the radar and the real blood sugar jumped up to nearly 33% in some instances.
The authors also tried to see how the radar handled low blood sugar (hypoglycemia), but there were so few low-sugar moments in the study that they couldn't draw a solid conclusion. It's like trying to test a car's brakes in a snowstorm when it never actually snowed during the test drive.
So, what's the takeaway? For most critically ill adults in the ICU who need insulin, this CGM radar is a fantastic tool that gives doctors a clear, continuous view of their patient's sugar levels, even if they are dealing with sepsis. It's accurate enough to be a trusted co-pilot. However, if the patient's body is in a state of extreme metabolic distress—specifically if their blood is very acidic or their lactate is sky-high—the radar might start to lag or misread the road. In those specific, chaotic moments, the study suggests doctors shouldn't rely on the radar alone; they need to pull over and check the manual gauge (the blood test) to be absolutely sure before making a move. The radar is a powerful helper, but it's not a replacement for a human check when the storm gets really wild.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.