We compared the analytical performance of a new point-of-care glucose meter specifically designed to automatically correct for hematocrit level and other interference factors with 3 commonly used glucose meters. We evaluated all 4 hospital-based glucose meter technologies for inaccuracy, imprecision, and analytical interferences likely to be encountered in patients with critical illness. In addition, we compared the performance of this new meter with one currently in use in our neurosurgical intensive care unit (ICU) to evaluate conformity to the International Organization for Standards 15197 criteria.
Imprecisions, both within run and day to day, were evaluated on all the 4 glucose meters. Inaccuracy (bias) of the meters and analytical interference were evaluated by comparing results from whole blood specimens with results from plasma samples obtained from these whole blood specimens run on a hexokinase reference method. Analytical accuracy was assessed with glucose specimens spiked with different interference factors. Clinical accuracy was assessed on 95 whole blood specimens collected from patients admitted to a neurosurgical ICU ward.
Imprecision and method correlation studies demonstrated slight differences in the degree to which the meters correlated with the hexokinase reference method. Paracetamol showed significant interference with 1 of the 4 meters tested. Ascorbate showed significant interference with 3 of the 4 meters. Hematocrit level also affected the correlation between whole blood and plasma hexokinase glucose results on 3 of the 4 meters tested. Assessment of clinical accuracy demonstrated that only 1 of the meters met the International Organization for Standards 15197 criteria when applied to neurosurgical ICU patients.
Correlation with plasma hexokinase values and hematocrit level interference are the main variables that differentiate glucose meters. A meter that directly correlates with the plasma hexokinase glucose reference method over a wide range of glucose concentrations and minimizes the effects of hematocrit level provides more reliable data as an adjunct to intensive insulin therapy in patients with critical illness. Of the 2 meters studied in the neurosurgical ICU, only the StatStrip (Nova Biomedical Corporation, Waltham, Mass) demonstrated a total error that was within the acceptable limits.