Glycemic control and blood gas sampling frequency during continuous glucose monitoring in the intensive care unit: A before‐and‐after study

BACKGROUND: Whether subcutaneous continuous glucose monitoring (CGM) can safely replace intermittent arterial blood gas glucose analyses in intensive care unit (ICU) patients remains uncertain. We aimed to compare CGM to blood gas glucose values and assess whether CGM use reduces blood gas sampling frequency and glucose variability in ICU patients with type 2 diabetes managed with liberal glucose control. METHODS: We used the FreeStyle Libre CGM in 15 ICU patients and compared their blood glucose metrics with a pre‐CGM control population of 105 ICU patients with type 2 diabetes. Both groups received insulin to target glucose range of 10–14 mmol/L. We used linear regression analysis adjusted for illness severity to assess the association of CGM use with blood gas sampling frequency and glucose variability. We used mean absolute relative difference (MARD) and Clarke error grid analysis to assess accuracy of matched CGM‐blood glucose values overall, across glucose stata (<10, 10–14, >14 mmol/L), and over time (≤48, 48–96, >96 h). RESULTS: We analyzed 483 matched glucose values. Overall MARD was 11.5 (95% CI, 10.7–12.3)% with 99% of readings in Clarke zones A and B. MARD was 15.5% for glucose values <10 mmol/L, 11.1% at 10–14 mmol/L, and 11.4% >14 mmol/L. MARD was 13.8% in the first 48 h, 10.9% at 48–96 h, and 8.9% beyond 96 h. CGM use was associated with 30% reduction in blood gas sampling frequency. CGM use was not associated with glucose variability as determined by glycemic lability index or standard deviation of blood glucose. CONCLUSIONS: In our cohort of ICU patients with type 2 diabetes receiving liberal glycemic control, CGM showed acceptable accuracy and was associated with a reduction in blood gas sampling frequency without compromising glucose control. Lowest accuracy was observed at glucose values below 10 mmol/L and during the first 48 h of CGM use.