I recently collaborated with some neonatologist colleagues by comparing neonatal blood glucose measurements using the Bayer Rapidlab 860 blood gas analyser glucose electrode with those obtained using the YSI 2300 Stat and the Vitros 750 XRC.

As part of this study, we considered the technology of the glucose electrode (with assistance from Bayer).

Although capillary whole blood samples were presented to the analyser, the electrode membrane is impermeable to protein (and red cells) so only the plasma water is in contact with the electrode. As the glucose is measured in a fixed volume of solvent (temperature is constant) it is close to being a plasma glucose molality measurement.

The calibration is also crucial, the Bayer method is calibrated against the hexokinase/glucose-6-phosphate dehydrogenase plasma glucose reference method.

The end result is that there was extremely close agreement between the "whole blood" glucose electrode measurements and the plasma glucose method (Vitros 750). However there was a small discrepancy between the Bayer results and those for the YSI using whole blood, which was consistent with the difference in water content between plasma and whole blood.

We also looked at the effect of haematocrit and could find no effect at all between 20 and 80%, which is presumably a consequence of the membrane retarding the haemoglobin.

In short, certainly for this electrode, the results using whole blood were extremely close to those obtained using the laboratory plasma glucose method. This work has been accepted by the Annals, although we only looked at samples from neonates with low glucose concentrations.