LBSUN214 Accuracy Of A Continuous Glucose Monitor In The Intensive Care Unit

Studies have shown that hyper/hypoglycemia and glycemic variation are associated with adverse outcomes in critically ill patients. Currently, frequent blood point-of-care (POC) glucose measurements from an arterial or capillary sample is the only technology available to minimize glycemic excursions in the ICU. Continuous glucose monitoring (CGM) is becoming the standard of care for outpatient diabetes care and has shown improved glycemic control in the non-ICU inpatient setting. The Dexcom G6 sensor (G6) is the first CGM device approved by Health Canada for outpatient diabetes management without the need for calibration, but it has not yet been approved for inpatient use. We collected data from 23 adults who were on an insulin infusion in a medical-surgical ICU in Vancouver, British Columbia to evaluate the accuracy of uncalibrated CGM in the ICU. A blinded G6 was attached to the patient's arm and collected glucose measurements every five minutes without calibration. Nursing staff continued POC arterial glucose measurements using the AccuChek Inform II machine per standard of care. Excluding four outliers (with mean absolute relative difference (MARD) ≥ 25%), the overall MARD was 13.24% (SE 0.43) over 649 matched CGM and arterial glucose values. A Clarke Error Grid demonstrated 99.1% of CGM measurements within zones A and B. The MARD using three-point rolling averages of CGM measurements in five-minute intervals was 13.49% (SE 0.68). There were 573 matched pairs between glucose ranges of 3.9-13.9 mmol/L with two pairs <3.9 and 74 pairs >13.9 mmol/L. Eleven patients had renal replacement therapy and twelve had vasopressor use. There was no significant difference in MARD with renal replacement or across glycemic ranges ≥3.9. The MARD for patients with vasopressors was lower than for patients without (13% vs 13.55%, p<0. 01), a finding of doubtful clinical relevance. There is no expert agreement yet about the acceptable accuracy for CGM use in hospital. Our overall MARD meets the 2013 Critical Care expert consensus recommendations for MARD <14%. The FDA guidance on standard of accuracy for conventional POC glucometers require 98% of values within 15% for BG ≥ 75mg/dL. Our data showed 65.18% of values were within 15%. Previous studies using CGM in the ICU, even those using G6, all used calibration. The majority were non-blinded, and none met FDA criteria for inpatient use. Our study is the first uncalibrated and blinded study that was able to demonstrate acceptable accuracy in a large sample size. Since CGM accuracy may be affected by various factors, our comprehensive data can potentially identify specific interferences and quantify a calibration or correction criteria to improve CGM accuracy in critically ill patients. Overall, our results show that CGM shows strong potential to be an accurate, resource-efficient, and intuitive alternative to POC glucose monitoring to meet glycemic targets in the ICU. Presentation: Sunday, June 12, 2022 12:30 p.m. - 2:30 p.m.