Background: Tight blood glucose control is used extensively in perioperative and critically ill patients. Several studies, however, have shown contradictory effects on patient outcomes. A major problem of these studies has been inadequate control of the prime variable, blood glucose. This paper describes the validation of a new intravascular continuous blood glucose sensor.
Methods: The glucose sensor was placed in the superior caval vein of seven anesthetized pigs. Sensor readings were compared with arterial blood gas readings. Fluctuations in blood glucose were created using intravenous glucose and insulin. A total of 807 paired sensor and blood gas readings were obtained.
Results: The sensor was tested with a range of blood glucose values (0.63-15.75 mM [mean bias, 0.0131 mM]). Analysis using Bland-Altman plots yielded 95% limits of agreement at -0.908 and 0.934 mM. There were 121 paired measurements with a mean value below 2.2 mM, yielding 95% limits of agreement at -0.553 and 0.466 mM.
Conclusions: The performance of the sensor was in agreement with blood gas measurements in a wide range of glucose values. For the clinician, it is noteworthy that performance was equally good in the hypoglycemic area.