Skip to main content
Fig. 3 | Annals of Intensive Care

Fig. 3

From: Accuracy, reliability, feasibility and nurse acceptance of a subcutaneous continuous glucose management system in critically ill patients: a prospective clinical trial

Fig. 3

Bland–Altman plot, Clarke error grid, Color-coded Surveillance Error-Grid. n = 532 comparative samples. a Bland–Altman plot. The mean bias indicates whether there is a systematic error. Upper and lower limits were calculated by mean bias ±1.96 × standard deviation of the difference between BG and sensor glucose and represent random variations around the mean bias. If there is a Gaussian distribution, 95 % of points are located between these limits. [22, 41]. b Clarke error grid. Distribution: A = 76.9 %, B = 21.6 %, C = 0.2 %, D = 0.9 %, E = 0.4 %. Zones A (CGM data ≤20 % deviation from BG) and B are considered as clinically acceptable zones, whereas values in zones C, D and E are increasingly dangerous for the patient, and zone E may lead to adverse therapeutic decisions. [23]. c Color-coded Surveillance Error-Grid. The Surveillance Error-Grid software is available at http://www.diabetestechnology.org/SEGsoftware/Surveillance-Error-Grid-Analysis.xlsm. Last Accessed: Dec 11 2015 [24]

Back to article page