The aim of the study was to evaluate the performance of an electrochem
ical blood glucose sensor, the satellite G, in the low-normal and hypo
glycaemic range. Eighty-five venous blood samples with glucose concent
rations below 4.0 mmol l-1 from six patients with Type 1 diabetes unde
rgoing induced hypoglycaemia were measured in duplicate on the sensor
and a reference method, the Yellow Springs whole blood glucose analyse
r. The sensor values correlated well with the reference method (r = 0.
9, p < 0.001) with a mean absolute percentage bias of 10.7 %. The 95 %
confidence limits of agreement between methods were +0.7 and -0.6 mmo
l l-1 with a mean difference of 0.04 mmol l-1. More than 80 % of the s
ensor measurements were within 20 %, and 60 % were within 10 % of the
reference method. Measured over all samples, the sensor had a coeffici
ent of variation of 7.2 % for paired measurements. The coefficients of
variation of 20 measurements on each of two samples with glucose conc
entrations of 1.3 and 2.6 mmol l-1 were 9.1 % and 5.8 %, respectively.
We conclude that the satellite G measures blood glucose accurately an
d precisely in the low range.