Clinical trial of a direct reading glucose biosensor in the neonatal intensive care unit

Objective To evaluate glucose concentration measured with a direct reading glucose biosensor (ABL 625), because fast and reliable measurement of blood glucose is essential in neonatal care. Methods In 56 neonates, 100 capillary blood samples were analyzed in three ways: by the glucose strip method and duplicate analyses with the ABL 625 a nd Hitachi 917 (hexokinase method). Blood glucose values were converted to plasma values by multiplication by 1.14. Results The ABL showed a higher repeatability coefficient (0.29 mmol/l) tha n the Hitachi (0.14 mmol/l). The bias, -0.13 mmol/l, was low. The limits of agreement (-0.91 to +0.65), however, were 50% wider than could be explaine d by the combined effects of the repeatability coefficients of the two meth ods. None of the recorded clinical variables appeared to contribute systema tically to the lack of agreement. Surprisingly, the limits of agreement of the strip values to the Hitachi unconverted values were close (-0.75 to +0. 8 mmol/l). Conclusion The ABL 625 was not as precise as the standard laboratory method , but similar to previous experience with direct reading sensors in neonate s.