Using Point of Care Glucose Meters in the Critically Ill: Assessing Meter Performance in the Clinical Context

Background Point of care (POC) glucose meters are widely used in hospitals to aid in monitoring blood glucose levels. We have evaluated the clinical performance of our meters in the critically ill population. Methods To evaluate the accuracy of glucose meters results, as obtained by our end users, we matched 6 months of POC results with corresponding laboratory data, minimizing the difference between the collection times and restricting the difference to 10 minutes. Using the Clarke and Parke's consensus error grids, we evaluated the differences between POC/laboratory results in the clinical context. We also evaluated meter performance in the critically ill, based on location (non-intensive care unit vs intensive care unit) and by derangement in other relevant laboratory results collected within 24 hours. Clarke error grids were also prepared for individual glucose meters by serial number as a quality assurance measure. Results Method agreement was far from ideal, as the correlation was described by the equation: y = 1.04x + 2.36, and r² = 0.73. However, when analyzed using consensus error grids, the overwhelming majority (98%) of these discrepancies were clinically insignificant. No relationship was found between severity of illness and degree of discrepancy, by hospital location (P > 0.5) nor abnormalities in other laboratory parameters (P > 0.1). We observed no significant biases for any given meter. Conclusions By incorporating the consensus error grids, we evaluated glucose meter performance in the clinical context, rather than a strictly analytical manner. This offers a more realistic determination of the accuracy necessary to effectively manage hospitalized patients, particularly in the absence of tight glycemic control.