Accuracy of an Interstitial Continuous Glucose Management System in Critically Ill Patients: A Prospective Observational Study.

INTRODUCTION: Glycemic management in the intensive care unit (ICU) remains challenging because glucose levels fluctuate rapidly and monitoring is intermittent. Interstitial continuous glucose monitoring (CGM) could improve surveillance, but its performance during rapid glucose excursions remains uncertain. We assessed the accuracy of the Dexcom G7 CGM in critically ill patients under stable conditions and during intravenous glucose tolerance tests (IVGTT). METHODS: We conducted a prospective ancillary study of the Determination of Insulin Resistance in ICU (DIRICU) study. Twenty critically ill adults with stress hyperglycemia, no prior diabetes, and expected ICU stay ≥5 days were included. A Dexcom G7 sensor was compared with arterial blood glucose measurements. Accuracy was assessed over the first 10 ICU days using paired measurements, mean bias, mean absolute relative difference (MARD), linear regression (R2), and Clarke error grid analysis, both outside and during IVGTT. A post hoc 20-minute temporal shift analysis was performed to account for delayed CGM peaks. RESULTS: Outside IVGTT, 716 paired values showed moderate agreement with reference glucose (R2 = .58, mean bias +11 mg/dL, MARD 17%), with 98.9% of values in Clarke zones A/B. During IVGTT, accuracy deteriorated markedly (1018 paired values): R2 = .028, mean bias -34 mg/dL, MARD 24%, and only 80.3% of values in zones A/B, with 18.6% in zone D. After temporal shifting, performance improved (R2 = .34, MARD 19%, 94.7% in zones A/B). No sensor-related adverse events occurred. CONCLUSIONS: Dexcom G7 showed acceptable accuracy during stable ICU conditions, but performance deteriorated during rapid glucose excursions. Hybrid monitoring strategies combining CGM with intermittent blood glucose confirmation may therefore be preferable in critically ill patients.