Performance of a Modern Glucose Meter in ICU and General Hospital Inpatients: 3 Years of Real-World Paired Meter and Central Laboratory Results

Objective: Due to accuracy concerns, the Food and Drug Administration issued guidances to manufacturers that resulted in Center for Medicare and Medicaid Services stating that the use of meters in critically ill patients is "off-label" and constitutes "high complexity" testing. This is causing significant workflow problems in ICUs nationally. We wished to determine whether real-world accuracy of modern glucose meters is worse in ICU patients compared with non-ICU inpatients. Design: We reviewed glucose results over the preceding 3 years, comparing results from paired glucose meter and central laboratory tests performed within 60 minutes of each other in ICU versus non-ICU settings. Setting: Seven ICU and 30 non-ICU wards at a 1,300-bed academic hospital in the United States. Subjects: A total of 14,763 general medicine/surgery inpatients and 20,970 ICU inpatients. Interventions: None. Measurements and Main Results: Compared meter results with near simultaneously performed laboratory results from the same patient by applying the 2016 U.S. Food and Drug Administration accuracy criteria, determining mean absolute relative difference and examining where paired results fell within the Parkes consensus error grid zones. A higher percentage of glucose meter results from ICUs than from non-ICUs passed 2016 Food and Drug Administration accuracy criteria (p < 10^-10) when comparing meter results with laboratory results. At 1 minute, no meter result from ICUs posed dangerous or significant risk by error grid analysis, whereas at 10 minutes, less than 0.1% of ICU meter results did, which was not statistically different from non-ICU results. Conclusions: Real-world accuracy of modern glucose meters is at least as accurate in the ICU setting as in the non-ICU setting at our institution.