Background: The aim of this study was to elucidate the characteristics of accuracy of subcutaneous continuous glucose monitoring (SCGM) in the perioperative period for neurosurgical and cardiac surgery patients.
Methods: Forty-five subjects, including healthy volunteers (n = 15), neurosurgical patients (n = 15), and cardiac surgery patients (n = 15), were enrolled. A subcutaneous sensor of the MiniMed™ 620G SCGM system was inserted into the upper arm. On the day after sensor insertion, SCGM data and blood glucose data were collected simultaneously and compared. In cardiac surgery patients, data were continuously collected on postoperative day (POD) 1 and POD 3. Clarke error grid analysis and Bland–Altman analysis were performed to assess the accuracy of SCGM.
Results: Clarke error grid analysis showed clinical acceptance of the SCGM system with 82.7% and 86.8% of the data being within zone A for healthy volunteers and neurosurgical patients, respectively. Mean biases were −2.1 mg/dL in healthy volunteers and −8.3 mg/dL in neurosurgical patients. In cardiac surgery, although Clarke error grid analysis showed clinical acceptance, 65.3% of the data were within zone A and mean bias was −23.5 mg/dL. Changes in accuracy of SCGM in individuals occurred during cardiopulmonary bypass (CPB), and SCGM tended to show a lower glucose level. On POD 1 and POD 3, the accuracy improved, and 85.0% and 86.3% of the data were within zone A.
Conclusions: Although the accuracy of the SCGM system was clinically acceptable in the perioperative period, sensor accuracy was affected by CPB and showed lower glucose levels.