Accuracy of Dexcom G6 Pro Continuous Glucose Monitor in the Intensive Care Setting: TIGHT ICU G6 Study

Objective: To evaluate the accuracy of the Dexcom G6 Pro continuous glucose monitoring (CGM) system in the intensive care unit (ICU) setting. Methods: We performed a prospective, observational, multicenter study in adult ICU patients with a known diagnosis of diabetes or stress hyperglycemia who were being treated with insulin. Two Dexcom G6 Pro sensors were placed. Sensor accuracy was assessed by pairing sensor and blood glucose (BG) measurements obtained as a part of usual ICU care. Accuracy of the G6 Pro also was assessed concurrently in non-ICU hospitalized individuals. Results: A total of 130 participants were enrolled, with mean (±SD) age of 62 ± 12 years, and preexisting diabetes was present in 73% and stress hyperglycemia in 27%. A total of 9120 sensor-BG pairs were analyzed. The mean relative absolute difference (RAD) was 23% (median 19%), with a mean difference of +25 mg/dL (median 25 mg/dL). Forty-one percent of sensor glucose values were within 15% of BG values for BG values ≥100 mg/dL or within 15 mg/dL of BG values for BG values <100 mg/dL, 53% within 20%/20 mg/dL, and 72% within 30%/30 mg/dL. For the two sensors worn simultaneously, the mean absolute difference was 24 mg/dL (median 19 mg/dL) and the mean RAD was 14% (median 11%). In the non-ICU setting (N = 60; 1318 sensor-BG pairs), the mean difference was 24 mg/dL (median 21 mg/dL) and the mean RAD was 21% (median 17%). Conclusions: Accuracy of the Dexcom G6 Pro sensor in the ICU setting was worse than has previously been reported for this sensor, with sensor values tending to be biased high. However, this appears to be a function of the G6 Pro sensor and not the setting as similar results were obtained in a non-ICU setting. Results should not be generalized to the real-time G6 or other sensors.