I read with interest the article by De Block et al. (1). Indeed, reliable devices recording continuously interstitial glucose concentrations (IGCs) may be an alternative to frequent glucose monitoring, especially in patients of intensive care units (ICUs) in whom normoglycemia has become a major target. However, all factors influencing the complex kinetics of IGCs should be understood before considering continuous glucose monitoring (CGM) as a valuable and accurate alternative to track hyperglycemia and adapt insulin therapy. In addition, to use CGM in an optimal way, the device should provide real-time glucose concentrations in order to quickly adjust insulin infusion rates according to ambient glucose levels. This requires initial rather than post hoc calibration. These two key issues deserve further comment.
First, the <3-min lag time between subcutaneous and arterial blood glucose concentrations emphasized by the authors might be questionable in ICU patients. Indeed, such lag time depends on physiological parameters responsible for a different glucose kinetic between interstitium and plasma (2). Such kinetic difference has been shown to lead to spurious hypoglycemia in the general diabetic population (3). Most importantly, in critically ill patients, the kinetics of IGC may be affected by alterations in hydric/ionic balance, as revealed by the presence of a third compartment and subcutaneous edema and probably by many other factors that are still unknown.
Second, the good accuracy of CGM was not assessed when using the device GlucoDay in its most optimal manner. Indeed, real-time glucose levels ideally should be obtained to adjust insulin therapy as rapidly as possible. This objective could only be achieved if the GlucoDay is calibrated 2 h after insertion of the microfiber in the subcutaneous tissue, provided that glucose levels are stable enough. De Block et al. used post hoc calibrations with two or six points. Accuracy was considered as excellent and, as expected, better with the option of more frequent calibrations (4). However, as all calibrations were performed a posteriori taking into account all points together, results of accuracy might be overoptimistic. Whether the results would be as good when using CGM to obtain real-time glucose values, i.e., using a single calibration after 2 h (as recommended by the manufacturer) and adjusting progressively thereafter thanks to later calibrations, remains an open question.
