Comparison of the accu-chek aviva point-of-care glucometer with blood gas and laboratory methods of analysis of glucose measurement in equine emergency patients

Background: More information is needed regarding accuracy of commonly used methods of glucose measurement in the critically ill horse. Hypothesis: Glucometry will have good agreement with a laboratory standard. Glucometry with plasma will have better agreement than when performed with whole blood. Animals: Fifty sequentially admitted equine emergency patients, aged > 1year. Methods: Venous blood was collected at admission and immediately analyzed by point-of-care glucometry on both whole blood (POC/WB) and plasma (POC/PL), a multielectrode blood gas analyzer with whole blood (BLG), and a standard laboratory method with plasma (CHEM). Paired data were compared using Lin's concordance correlation, Pearson's correlation, and robust regression. Bias and limits of agreement were tested by the Bland-Altman technique. Bivariate regression analysis was used to explore confounding factors. Results: Concordance was significant for all comparisons, and was strongest for CHEM-POC/PL (0.977) and weakest for POC/WB-POC/PL (0.668). Pearson's correlation was excellent for all comparisons except those with POC/WB. All comparisons had excellent robust regression coefficients except those with POC/WB. Conclusions and Clinical Importance: POC glucometry with plasma had excellent agreement with a laboratory standard, as did blood gas analysis. POC glucometry with whole blood correlated poorly with a laboratory standard. These differences may be clinically important, and could affect decisions based on glucose concentrations.