Evaluation of a continuous glucose monitor in an unselected general intensive care population.

OBJECTIVE: To assess the accuracy of the Guardian REALTime continuous glucose monitoring system (Medtronic MiniMed, Northridge, Calif, USA) in critically ill adults compared with standard bedside point-of-care testing of capillary and arterial blood glucose levels.
SETTING: An 18-bed mixed university tertiary referral intensive care unit, July to December 2006. MAIN OUTCOME MEASURES: Glucose measurements by the Guardian subcutaneous monitoring system were compared with simultaneous measurements of capillary and arterial blood glucose levels. Correlation between the different methods was determined by linear regression, using capillary blood glucose values as the "gold standard", and was further analysed by Bland and Altman plots.
RESULTS: 17 emergency-admission mechanically ventilated adults were studied. A total of 1101 glucose measurements were available for analysis (from Guardian, 371; capillary blood samples, 373; and arterial blood samples, 357). With an a-priori acceptable difference of +/-1mmol/L, the Guardian system did not perform well enough to replace current methods of glucose measurement. There was also significant discrepancy between arterial and capillary blood glucose measurements. These common methods of intermittent blood glucose sampling are therefore not interchangeable.
CONCLUSION: The Guardian REAL-Time continuous glucose monitoring system cannot replace current methods of blood glucose measurement at a glucose threshold of +/-1mmol/L. It may have role as an early warning detection system for hypo- or hyperglycaemia, but this needs to be evaluated in a prospective study of extremes of blood glucose levels in a critically ill population. Glucose measurements in arterial and capillary blood samples with a point-of-care glucometer also showed wide discrepancies.