Technical note: Validation of a handheld meter for measuring β-hydroxybutyrate concentrations in plasma and serum...

The use of handheld meters for cow-side tests for β-hydroxybutyrate (BHB) concentrations in whole blood has become common. The aim of this study was to compare serum BHB (sBHB) and plasma BHB (pBHB) concentrations analyzed using either a "gold standard" enzymatic laboratory method (LM; Randox Laboratories Ltd., Antrim, UK; cat. no. RB1007) or a handheld meter (PX; Precision Xtra, Abbott Diabetes Care Inc., Alameda, CA). Results from 374 (187 serum and 187 plasma) samples taken from Holstein cows from 11 d before (52 samples) to 5 d after parturition (137 samples) were used for the analysis. Statistical analysis was performed using the MIXED, REG, and LOGISTIC procedures of SAS (v9.4; SAS Institute Inc., Cary, NC). A linear mixed model with repeated measures was fitted for LM and PX. Regression and correlation analyses were completed to estimate the relationship and agreement between the 2 methods. Lin's concordance correlation coefficient (CCC) and Bland-Altman plots were used to evaluate agreement between LM and PX. Cross-validation by randomly splitting the data in model-building and validation sets was performed to estimate and validate the equation that predicted the LM results using PX. Receiver operating characteristic (ROC) curves were used to estimate the sensitivity (Se) and specificity (Sp) of PX at different threshold levels. The CCC was 0.74 for pBHB and 0.68 for sBHB. The 95% confidence interval of agreement of the Bland-Altman plot encompassed 96% of the difference between LM and PX for pBHB and 95% for sBHB. The mean difference for pBHB was -0.50±0.25mmol, and that for sBHB was -0.63±0.41mmol. The highest Se and Sp for PX were achieved when the threshold for ketosis was set to 1.8mmol/L for pBHB and 2.1mmol/L for sBHB. The area under the ROC curve was 0.97 for pBHB and 0.96 for sBHB. The negative bias shown by the Bland-Altman plots suggested that PX yielded higher pBHB and sBHB concentrations than the LM. However, the excellent test characteristics and area under the ROC curve close to 1 indicated that PX at the adjusted thresholds was able to accurately classify between samples <1.2mmol/L and those ≥1.2mmol/L based on the gold standard test.