The objective of this study was to compare measurements of nonesterified fatty acids (NEFA) between the gold standard diagnostic laboratory method and a handheld NEFA meter (Qucare Pro meter, DFI Co. Ltd.). Three experiments were conducted to study the usability of the meter. In experiment 1 we compared results of the meter obtained from measurements in serum and whole blood with results of the gold standard method. Based on the results of experiment 1 we compared the results measured by the meter in whole blood with results obtained from the gold standard method on a larger scale, as we wanted to omit the step of centrifugation with the cow-side test. In experiment 3 we determined the influence of ambient temperature on measurements. Overall, blood samples of 231 cows were collected between 14 and 20 d in milk. The Spearman correlation coefficients (ρ) were calculated and Bland-Altman plots were created to compare the accuracy of the NEFA meter with the gold standard. In addition, in experiment 2 receiver operating characteristic (ROC) curve analyses were performed, to define thresholds for the NEFA meter to detect cows with a NEFA concentration above 0.3, 0.4, and 0.7 mEq/L. In experiment 1, there was a high correlation between NEFA concentrations in whole blood and serum determined by the NEFA meter and the gold standard (ρ = 0.90 for measurement in whole blood; ρ = 0.93 for measurement in serum). In experiment 2, the measurement in whole blood with the NEFA meter was compared with the gold standard. Despite a lower correlation (ρ = 0.79) the ROC curve analyses revealed a high specificity and a moderate sensitivity for lower cut-points (i.e., 0.3 and 0.4 mEq/L). The NEFA meter underestimated especially high concentration of >0.7 mEq/L. Considering thresholds of 0.3, 0.4, and 0.7 mEq/L measured by the gold standard test, sensitivity and specificity were 59.1% and 96.7%, 79.0% and 95.4%, and 86.4% and 95.6%, respectively, when using 0.3, 0.3, and 0.4 mEq/L as thresholds for the NEFA meter. Accuracy was 74.1%, 88.3%, and 93.8% for the 3 thresholds tested. Experiment 3 showed that measurements should be conducted at approximately 21°C (ρ = 0.73) as correlations were poor at 6.2°C and 15.1°C (ρ = 0.18 and 0.22, respectively).
