A pilot study of continuous glucose monitoring in critically ill patients: Do they perform well enough for use in glycaemic control?

Abstract

Glycaemic control (GC) in the intensive care unit (ICU) has proven difficult and contentious. Continuous glucose monitors (CGM) have been mooted as a solution to provide better control with less clinical effort. The aim of this study was to assess the reliability of CGM devices in critically ill patients, as well as the impact of device type (inter-device) and sensor location (inter-site) on performance. Ten patients were enrolled in this pilot trial and each patient was monitored using 3 concurrent CGM devices: a Medtronic Guardian real-time on their abdomen and Medtronic iPro2 recorders on their abdomen and thigh. The Guardian real-time had an overall MARD of 24%, compared to ~12% for the iPro2 devices. Bland Altman plots showed Guardian SG errors were associated with BG level, but iPro2 SG errors were not associated with BG level. Inter-device SG discrepancies were larger than inter-site discrepancies, when comparing concurrent data and CGM device type, or calibration method, tends to have a larger impact on SG accuracy than sensor location. Three case studies showed several interesting findings regarding CGM behaviour in critically ill patients. CGM devices are capable of performing very well in critically ill patients, but certain illnesses/conditions, as well as drugs/therapies may impact SG data. These factors require further investigation before CGMs can be implemented as standard clinical practice and/or for GC.