Accuracy and reliability of a subcutaneous continuous glucose-monitoring system in critically ill patients.

Background: Continuous glucose monitoring has been proposed to optimize glucose control in critically ill patients. To achieve strict glucose regulation, accurate and reliable continuous glucose-monitoring systems are essential. Objective: Evaluation of a subcutaneous continuous glucose-monitoring system for use in critically ill patients. Design: Pooled-data analysis of two prospective, randomized, controlled trials. Setting: An eight-bed medical intensive care unit of a university hospital. Patients: A total of 174 critically ill patients on intensive insulin therapy. Interventions: Subcutaneous continuous glucose monitoring. Measurements: Two thousand forty-five continuous glucose-monitoring system sensor glucose values were compared with arterial reference blood glucose levels, determined by a blood gas analyzer. Continuous glucose monitoring data were recorded continuously for up to 72 hrs by using a subcutaneous continuous glucose-monitoring sensor. The correlation of both methods and differences between continuous glucose-monitoring systems and reference values were calculated, as well as the conformity of continuous glucose-monitoring values with the International Organization for Standardization criteria (<0.83 mmol/L [15 mg/dL] difference for glucose values <= 4.12 mmol/L [>= 75 mg/dL] and <20% difference for glucose values >4.12 mmol/L [>75 mg/dL]). Results: The Pearson correlation coefficient was 0.92, showing strong correlation between the two methods. The intraclass correlation coefficient was 0.92, indicating that 92% of the variability is due to subjects and measurement occasions. Mean difference between continuous glucose-monitoring system and reference values was -0.10 mmol/L (confidence interval: -0.13 to -0.07) (-2 mg/dL [confidence interval: -2 to -1]) (continuous glucose-monitoring system minus reference) and absolute difference 0.44 mmol/L (confidence interval: 0.41-0.47) (8 mg/dL [confidence interval: 7-8]). According to the insulin titration error grid analysis, 99.1% of continuous glucose-monitoring system values were in the acceptable treatment zone. No continuous glucose-monitoring system measurements were found in the life-threatening zone, and 92.9% of the continuous glucose-monitoring system glucose values met the International Organization for Standardization criteria. Conclusion: The subcutaneous continuous glucose-monitoring system is reliable for use in critically ill patients and showed glucose values with a strong correlation to arterial reference blood glucose levels, determined by a blood gas analyzer. (Crit Care Med 2011; 39: 659-664)