Objective The purpose of this study was to evaluate the accuracy of a novel commercial glucose-oxidase (GO)-based glucometer in measuring glucose concentrations in arterial blood samples collected from premature infants.

Patients and methods A commercial glucometer, which is a plasma separation GO-based glucometer, was employed to compare the accuracy of the commercial blood glucometer with that of a standard glucose analyser for monitoring blood glucose levels in premature infants. Arterial blood samples were collected and analysed, including samples adjusted for specific glucose (3.6 mmol/L, 8.3 mmol/L and 16.7 mmol/L) and PaO₂ levels (range, 40–400 mm Hg).

Results In total, 159 samples were collected. Bland–Altman analysis showed good correlation between the commercial glucometer and standard glucose analyser. Error-grid analysis revealed that all of the results were within zone A (ie, the clinically accurate estimate zone). The biases between the two systems were low at different PaO₂ levels and haematocrits. Finally, the influence of different PaO₂ levels was within acceptable ranges.

Conclusions The GO-based glucometer evaluated in this study provides accurate results even when measured at high PaO₂ and different haematocrits and is suitable for measuring arterial glucose levels in premature infants.