Personal CGMs accurate at measuring glucose in most ranges during hospitalization
Key takeaways:
- The mean relative absolute difference between CGM and other glucose measurements in the hospital was less than 14%.
- No correlation was observed with absolute relative difference and vital signs.
Personal continuous glucose monitors are accurate at measuring glucose levels in most ranges for hospitalized patients with diabetes, researchers report in a study published in Diabetes Technology & Therapeutics.
In a pilot study conducted at Mayo Clinic in Jacksonville, Florida, researchers assessed CGM data from 101 people with diabetes who were hospitalized at the center from October 2021 to November 2023 (median age, 62 years; 65.3% men). Researchers found CGM-measured glucose had a mean absolute relative difference (MARD) of 13.7% compared with glucometer readings and a 13.6% MARD compared with serum glucose values.
Adrian G. Dumitrascu, MD, an internist at Mayo Clinic in Florida, and colleagues wrote that there are no published standards for CGM accuracy in the hospital, but a review paper published in Endocrine Practice in 2023 stated the MARD for CGMs during hospitalization should be less than 14%.
“For glucose values above the hypoglycemia range, CGM accuracy is adequate for a safe use of these devices in inpatient setting,” the researchers wrote. “This supports their utility for glucose monitoring and insulin dosing under specific protocols, especially considering the convenience that CGMs offer to patients who are already familiar with their use.”
Researchers obtained data from adults aged 18 years and older with diabetes who were wearing transcutaneous CGM at hospital admission and admitted to noncritical care at Mayo Clinic in Florida. CGMs in the study included the Abbott FreeStyle Libre 14-day, Libre 2 and Libre 3; and the Dexcom G6 and G7. CGM data were downloaded at discharge. CGM-measured glucose was compared with glucometer readings and laboratory serum glucose.
There were 3,316 CGM glucose measures paired with glucometer readings conducted within 5 minutes of the CGM measure. The overall MARD between CGM and glucometer measures was 13.7%. MARD was 22.6% for glucose levels less than 70 mg/dL, 14.6% for glucose between 70 mg/dL and 180 mg/dL, 11.7% for glucose between 181 mg/dL and 250 mg/dL and 10.4% for glucose levels of 251 mg/dL and higher.
When 771 CGM measures were paired with laboratory serum glucose levels, the overall MARD was 13.6%. The MARD for glucose levels in the laboratory hypoglycemic range was 33.5%.
No correlation was observed between absolute relative difference for CGM-glucometer pairs and laboratory and vital signs.
“There have been concerns about the potential negative impact of laboratory values and vital signs on the CGM accuracy in hospitalized patients,” the researchers wrote. “However, our analysis did not find a correlation between CGM MARD and these factors. Along with the currently available evidence, these findings provide reassurance regarding the use of CGMs in the hospital setting for patients under various clinical conditions.”
Mean glucose was 172.4 mg/dL. The study group had a time in range with a glucose of 70 mg/dL to 180 mg/dL of 58.7%. Time below range with a glucose of less than 70 mg/dL was 2.2% and time above range with a glucose higher than 180 mg/dL was 39.9%.
The study examined data retrospectively, but the researchers said hospitals may be able to integrate personal CGM devices into hospital records and enhance glycemic management in the future.
“Such integration could enhance patient and staff satisfaction, streamline workflows, and reduce the frequency of traditional finger-stick measurements, which can be burdensome in busy inpatient settings,” the researchers wrote.
Reference:
Perspective
Back to Top
This study adds to the growing evidence supporting the use of personal CGMs in the hospital. As an endocrine hospitalist, I see the operational and clinical advantages these devices offer — but also the importance of rigor when integrating outpatient technology into inpatient workflows.
The findings are reassuring. CGMs brought from home demonstrated strong accuracy across a diverse hospitalized population, with more than 98% of values in clinically acceptable error grid zones. Accuracy remained stable regardless of lab or vital sign fluctuations — a common concern. However, accuracy dropped in hypoglycemia, reinforcing that CGMs should complement, not replace, point-of-care checks during low or rapidly changing glucose levels.
One important limitation is the authors did not have access to CGM alerts or trend data, and could not assess whether patients responded to warnings to prevent hypoglycemia. Trends — not just static glucose values — offer critical context and should not be underappreciated when evaluating inpatient CGM utility.
As we move toward broader CGM adoption, real-time data integration into the electronic health record is the logical next step and now feels increasingly within reach.
Collapse