Rapid and reliable determination of blood glucose concentration is ess
ential during the neonatal period to prevent adverse neurodevelopmenta
l outcome from hypoglycaemia. Despite their unreliability, reagent str
ip methods continue to be used extensively in neonatal nurseries due t
o their rapidity and convenience. Recently, a new portable laboratory
standard technique has been introduced (HemoCue B-Glucose system) for
whole blood glucose determination. It is particularly suitable for nea
r-patient testing in neonatal units. This new method, as well as other
established methods of whole blood (Yellow Springs Instrument (YSI) a
nd a hexokinase method on Cobas Bio), and plasma (Kodak Ektachem) gluc
ose measurement, were therefore evaluated for their accuracy and conco
rdance of measurements taken in the neonatal period. There were substa
ntial discrepancies among the four methods of glucose measurement with
wide limits of agreement between these methods. The glucose concentra
tions measured by HemoCue and YSI (n=206), HemoCue and hexokinase (n=1
13), HemoCue and plasma glucose on Ektachem (n=69) and hexokinase and
Ektachem (n=66) were likely to differ by -29 to +61%, -23 to +56%, -36
to +65%, and -19 to +30%, respectively. Even the laboratory methods o
f blood glucose determination, therefore, can not be used interchangea
bly. Using a model based approach, the probabilities of ''discordant''
classification as hypo- or normo-glycaemia were estimated to be 6.8%,
6.5%, and 7.1% between HemoCue and YSI, HemoCue and hexokinase on Cob
as Bio, and HemoCue and Ektachem analysers, respectively. In view of t
hese low probabilities of discordant classification with other glucose
analysers, the HemoCue system may offer a reasonable compromise betwe
en bedside and laboratory blood glucose estimations in neonates.