Abstract
Background: To assess glucose meter performance in a diabetes antenatal clinic, focussing on clinical and pre-analytical factors that might impact on the quantification of meter accuracy and precision.
Methods: The Freestyle Lite and the Performa glucose meters were assessed by trained researchers. Finger stick glucose was measured and compared with plasma venous glucose, obtained from a concomitantly collected antecubital fossa sample. Venous plasma was separated on-site then sent to the laboratory for measurement of glucose using the hexokinase method (comparative method). Additional data collected included: (i) timing of and also (ii) quantity of last carbohydrate intake; (iii) time periods between collection, preparation and analysis of the venous sample; (iv) the haemolysis index of the plasma sample and (v) haematocrit.
Results: There were 104 participants. Both meters fulfilled ISO 15197 standards, with 99% and 97% of Freestyle Lite and Performa results, respectively, falling within acceptable limits for this standard. Both meters showed minor proportional bias, reading low at higher glucose values. Consensus error grid analysis showed 100% of results from the Freestyle Lite and 99% from the Performa falling within Zone A, thus the observed differences in measured capillary and venous plasma glucose were sufficiently minor that they would have little effect on clinical action. No association was observed between [capillary-plasma] glucose difference and the five variables outlined above.
Conclusions: The two glucose meters tested showed a reassuringly acceptable level of performance, when assessed by a research team in the setting of a diabetes antenatal clinic.