Continuous Glucose Monitoring Using the Dexcom G6 in Cardiac Surgery during the Postoperative Period.

UNLABELLED: Cardiac surgery is associated with hyperglycemia, which in turn is associated with adverse post-surgical outcomes such as wound infections, acute renal failure, and mortality. This pilot study seeks to determine if Dexcom G6Pro continuous glucose monitor (Dexcom G6Pro CGM) is accurate during the postoperative cardiac surgery period when fluid shifts, systemic inflammatory response syndrome (SIRS), and vasoactive medications are frequently encountered, compared to standard glucose monitoring techniques.

METHODS: This study received IRB approval. In this prospective study, correlation between clinical and Dexcom glucose readings was evaluated. Clinical glucose (blood gas, metabolic panel, and point of care [POC]) dataset included 1428 readings from 29 patients, while the Dexcom G6Pro CGM data included 45,645 data points following placement to upper arm. Additionally, average clinical measurements of day and overnight temperatures and hemodynamics were evaluated. Clinical and Dexcom data were restricted to being at least 1 hour after prior clinical reading Matching Dexcom G6Pro CGM data were required within 5 minutes of clinical measure. Data included only if taken at least 2 hours after Dexcom G6Pro CGM insertion (warm-up time) and and analyzed only following ICU admission. Finally, a dataset excluding the first 24 hours after ICU admission was created to explore stability of the device. Patients remained on Dexcom G6Pro CGM until discharge or 10 days postoperatively.

RESULTS: The population was 71% male, 14% with known diabetes; 66% required IV insulin infusion. The Clarke error grid plot of all measures post-ICU admission showed 53.5% in Zone A, 45.9% in Zone B, and 0.6% (n=5) in Zones D or E. The restricted dataset that excluded the first 24 hours post-ICU admission showed 55.9% in Zone A, 43.9% in Zone B, and 0.2% in Zone D. Mean absolute relative difference (MARD) between clinical and Dexcom G6Pro CGM measures was 20.6% and 21.6% in the entire post-ICU admission dataset, and the dataset excluding the first 24 hours after ICU admission, respectively. In the sub-analysis of the 12 patients who did not have more than a 5-minute tap in the operating room, a Consensus error grid, demonstrated that after ICU admission, percentage in zone A was 53.9%, zone B 45.4%, and zone C 0.7%. Similar percentages were obtained removing the first 24 hours post-ICU admission. These numbers are very similar to the entire cohort. A consensus error grid created post-ICU admission demonstrated: (Zone A) 54%, (Zone B) 45%, (Zone C) 0.9%, and the following for the dataset created excluding the first 24 hours: (Zone A) 56%, (Zone B) 44%, (Zone C) 0.4%, which demonstrated very close agreement with the original Clarke error grid. No adverse events were reported.

CONCLUSIONS: Almost 100% of Dexcom G6Pro CGM and clinical data matching points fell within areas considered as giving clinically correct decisions (Zone A) and clinically uncritical decisions (Zone B). However, the relatively high MARD precludes its use for both monitoring and treatment in the clinical context. As technology evolves, interstitial glucose monitoring may become an important tool to limit iatrogenic anemia and mitigate glycemic fluctuations.