Continuous glucose monitoring in the surgical intensive care unit: concordance with capillary glucose.

BACKGROUND: The role of intensive glycemic control (IGC) in the surgical intensive care unit (SICU) remains controversial. Continuous glucose monitoring systems (CGMSs) may mitigate the major risk of IGC, namely hypoglycemia, and improve clinical outcomes. METHODS: All patients admitted to the SICU requiring insulin infusion were eligible. CGMS (Medtronic Guardian REAL-Time CGM, Northridge, CA) was placed in the subcutaneous tissue of the abdomen or thigh and calibrated every 8 hours, based on capillary (fingerstick) blood glucose (CBG) readings. Monitors were changed every 72 hours until 144 hours of observation was complete or insulin infusion stopped. CGM data were compared with CBG at least every 2 hours. Other data collected included demographics, diagnoses, fluid balance, doses of vasopressors and/or steroids, and any intravenous or enteral glucose source. CGMS and CBG readings were compared (mean and median absolute difference, correlation coefficients, Bland-Altman plots, and Clarke error grids). RESULTS: Twenty-four patients were enrolled (11 men; mean [SD] age, 59 [14.1] years; mean [SD] body mass index 37.9 [10.1] kg/m; mean [SD] fluid resuscitation in the first 24 hours, 6.1 [3.5] L; 17 requiring vasopressor therapy). Correlation coefficient between CGMS and CBG was 0.61 (p < 0.001). The mean (SD) absolute difference was 22.0 [21.9] mg/dL and the median absolute difference was 16.0 mg/dL (interquartile rage, 7-31 mg/dL). The Bland-Altman plot did not identify any trends in accuracy. Clarke error grid analysis demonstrated that 98.92% of data points were in Zone A (71.30%), indicating agreement with CBG ± 20%, or Zone B (27.62%) (divergent but discrepancy would likely not lead to patient harm). Just 0.81% of data points were in Zone C (potentially dangerous overcorrection likely), and only 0.27% were in Zones D or E (potentially dangerous failure to detect hypoglycemia/hyperglycemia). CONCLUSION: CGMS seems reasonably accurate in the SICU, despite widespread use of pressors and large-volume resuscitation. Further investigation into the accuracy and precision of these devices to assist clinicians in achieving IGC is warranted. LEVEL OF EVIDENCE: Diagnostic study, level III.