Evaluation of the Freestyle Optium Neo H point‐of‐care device for measuring blood glucose concentrations in sick calves

Abstract Background Data on the performance of a glucometer in calves with different diseases are currently lacking. Objective The primary objective of this study was to evaluate the reliability of a point of care glucometer in calves affected by different diseases relative to a traditional bench‐top autoanalyzer. Animals One hundred ninety‐six calves with different disorders in a referral hospital. Methods Prospective study. Venous blood samples were used for the determination of glucose concentrations in blood and plasma using the Freestyle Optium Neo H and autoanalyzer, respectively. Data were subjected to Passing‐Bablok regression and Bland‐Altman plots. The Freestyle Optium Neo H was the test method and the autoanalyzer was the reference method. The diagnostic performance of the glucometer relative to the autoanalyzer was assessed using 3 different plasma glucose concentrations. Results The Passing‐Bablok regression for the glucometer against the reference method revealed the presence of both proportional bias (1.12; 95% confidence interval [CI], 1.07‐1.18) and constant bias (−11.25; 95% CI, −16.0 to −7.70). The glucometer yielded 92.2%‐100% sensitivity and 86.4%‐96% specificity for the assessing glucose concentration based on different concentration thresholds. Conclusions and Clinical Importance The Freestyle Optium Neo H showed proportional and constant biases relative to the reference method. The glucometer showed poor performance according to criteria recommended by the International Standards Organization and the American Society for Veterinary Clinical Pathology. However, the glucometer determined hypoglycemia with high sensitivity and specificity therefore it might be used to diagnose hypoglycemia in calves with different diseases until calf‐specific POC glucometers are developed.


| INTRODUCTION
Commercially available portable blood glucose meters are small, convenient, and cost effective devices to measure blood glucose concentrations. Glucometers can yield rapid results and can potentially assist clinicians to make short-term therapeutic strategies especially in hypoglycemia. Therefore, point-of-care(POC) glucometers may be used in ambulatory clinics and farm settings in bovine medicine. Most commercially available glucometers are designed for human use. In spite of the fact that glucometers employ conversion algorithms to report the result as a plasma value, they measure the glucose concentrations in whole blood. These algorithms used by glucometers are based on intraerythrocyte to plasma glucose ratio that varies with species and age. 1,2 Evaluation of different glucometers designed for human beings was made in horses, foals, sheep, and goats. [3][4][5][6] The accuracy of blood glucose concentrations from different glucometers was also assessed in dairy cows. [7][8][9] However, difference intraerythrocyte to plasma glucose ratio between adult cattle and calves might cause different glucometers' performance in calves. 1,9 Moreover, hematocrit (Hct) might affect the accuracy of glucometers. In human blood samples, increased Hct can cause decreases in blood glucose measurements and vice versa. 10,11 Changes in Hct seen in calves with different diseases such as diarrhea, and infection might therefore affect agreement of glucometer relative to reference method. Data on the performance of a glucometer in calves with different diseases are currently lacking.
Calves may have wide variation of their glycemia with diseases.
Hypoglycemia is particularly common in acute conditions in neonates and is significantly associated with nonsurvival in ill neonatal calves. 12

| Blood samples
Blood samples were taken from the jugular vein of all calves before starting the treatment. A portion of the blood sample was transferred to test tubes including lithium heparin for plasma separation. The glucometer readings at the calf-side were conducted on whole blood with no preservative immediately after the blood collection. After centrifugation of the test tubes at 3000g for 10 minutes, the plasma samples were stored at −20 C for plasma glucose and plasma TP concentration measurements within 7 days using an autoanalyzer. The interval from blood collection until plasma separation in the laboratory was maximum 20 minutes for all samples.

| Blood glucose measurements by the glucometer
Blood glucose measurements were performed on blood samples without preservatives by the Freestyle Optium Neo H. Single-use reagent strips were used for the Freestyle Optium Neo H. A drop of blood samples was placed on the test strip. All blood glucose measurements were made with 1 POC glucometer. The measurement range of the glucometer was 20 to 500 mg/dL. When blood glucose concentration was too low to be read by the glucometer, the glucometer gave LO message. Glucose Hct range was 15 to 65% according to manufacturer's manual for the glucometer. The glucometer displayed glucose in mg/dL in 5 seconds. The Freestyle Optium Neo H uses glucose amperometric technology to quantitatively measure the glucose concentration in whole blood samples. Briefly, the glucose biosensor recognizes the glucose present in blood samples by means of the glucose spesificity of the enzyme glucose dehydrogenase present on the glucose test strip. The electrons liberated by this reaction are transferred to the glucometer where they are read as a small electrical current.
The size of the current is directly proportional to the concentration of the glucose in the applied blood sample. When blood sample applied to test strip is too small the glucometer gives error message. and photometric method principles. The analytical measurement range of the autoanalyzer was 2 to 750 mg/dL. The intra-assay CV of the Cobas C501 was also calculated from 7 replicates of low (mean, 24 mg/dL) and moderate (mean, 117 mg/dL) blood glucose concentrations.
Hematocrit measurement was performed once using capillary microhematocrit tubes after centrifugation for 5 minutes. Plasma TP concentrations were determined by the Cobas C501. The Cobas C501 uses biuret method and colorimetric assay. The analytical measurement range of the autoanalyzer was 0.2 to 12 g/dL. Because it has been previously established that intracorpuscular to plasma glucose ratio is higher in calves younger than 1 week of age, 1,2 we further identify the calves as <1 week old or 7 days of age or more because we anticipate that this covariate could potentially impact the relationship between Freestyle Optium Neo H and autoanalyzer.

| Statistical analysis
Normality of the data was assessed with the Shapiro-Wilk test. Normally distributed data were expressed as mean ± SD and nonnormally distributed data were expressed as median and range. The analyses were reported in mean ± SD(TP and Hct) or median (range) for blood, plasma glucose concentrations, and total observed error (TEobs) accordingly. The autoanalyzer was considered the reference method, whereas the Freestyle Optium Neo H served as the test method.
Passing-Bablok regression analysis was performed to evaluate the relationship between blood glucose concentrations and plasma glucose concentrations. 17 Cumulative sum test (Cusum test) for linearity was also assessed with the Passing-Bablok regression analysis. The agreement between the glucometer and the reference method was evaluated by Bland-Altman plots. 18 Receiver operating characteristic (ROC) curves were developed at the different cutoffs plasma glucose concentration of <72, 16 <58, 14 and <40 mg/dL 13,15 (hypoglycemia) to determine Se and Sp of the Freestyle Optium Neo H at the highest Youden index, which minimizes the number of misclassificatied samples. 19 The Youden index was calculated as Se + Sp-1. The area under the curve (AUC) for ROC curves was calculated to measure diagnostic test performance of the Freestyle Optium Neo H. AUC > 0.90 was considered highly accurate, AUC between 0.70 and 0.90 was considered moderate accurate, AUC between 0.5 and 0.7 was considered low accurate, and AUC ≤ 0.5 was considered a chance result. 20 In order to explore the potential impact of the glucose concentration as measured by the reference method, Hct and plasma TP concentration on the dependent variable (blood glucose concentration measured by the Freestyle Optium Neo H) a multivariable linear regression was performed. The difference percentage (DGlu (%) = (Glu-POC-GluRef)/GluRef)) was determined as the dependent variable.
The initial model that was tested was: The interaction between TP and Hct was a priori put in the model for accounting on these parameter changes in many different calves' diseases. 9 The age (dichotomous <7j, versus ≥7j was also considered as a potential covariate. The regression modeling was performed using lme4 package from the R statistical software. 21 The covariates were removed sequentially until the final model having all remaining variables with P-value <.05. Distribution of the residuals was visually checked for normality. The fit of the model was determined using adjusted R-squared. The glucometer performance was also analyzed in compliance with the 2013 standard (ISO 15197-2013) from the International Standards Organization in which 95% of the individual glucose results for the glucometer should fall within 15 mg/dL of the reference measurement for glucose concentrations when plasma glucose concentration <100 mg/dL and within 15% of the reference measurement for glucose concentrations when plasma glucose concentration >100 mg/dL. 22 The performance of the Freestyle Optium Neo H was also assessed using a calculation of TEobs as follows: TEobs = 2 × intra-assay CV (%) + bias (%). 23
Three calves that previously received glucose solutions before collecting the blood samples and 13 calves that had blood glucose concentration below the detection limit of the glucometer (≤20 mg/dL) were excluded from the present study.
The intra-assay CV of the Cobas C501 for the mean of 24 and 117 mg/dL plasma glucose concentration was 3.7 and 1.2%, respectively. The intra-assay CV of the Freestyle Optium Neo H for the mean of 38 and 116 mg/dL blood glucose concentration was 3.0 and 3.9%, respectively.
Nonnormally distributed data were observed in plasma and blood glucose concentrations. Plasma glucose concentrations measured by the Cobas C501 ranged from 18 to 155 mg/dL (median, 85 mg/dL).

The distribution of the measurements between the Freestyle
Optium Neo H and the Hct value is shown in Figure 2. The results of the final multivariable model include is indicated in The Bland-Altman approach is also widely used in method comparison studies to quantify agreement between 2 methods and to evaluate a bias the mean differences. Bland-Altman analysis detected that One Touch Vita underestimated blood glucose concentrations in cattle. 5 The mean bias between the Precision Xtra and the reference method was −0.9 mg/dL in lactating dairy cows. 25 The FreeStyle Precision Neo glucometer in equine emergency patients. 4 In periparturient dairy cows, plasma glucose concentration was a significant predictor of blood glucose concentrations measured by the Precision Xtra. In that study, however, Se analysis using a spider plot showed that the percent error in blood glucose concentration measured by the glucometer was moderately dependent on the Hct. 9

Sébastien Buczinski serves as Consulting Editor for Experimental
Design and Statistics for the Journal of Veterinary Internal Medicine. He was not involved in review of this manuscript.

OFF-LABEL ANTIMICROBIAL DECLARATION
Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION
This study was approved by the Firat University Ethics Committee on Animal Experimentation.

HUMAN ETHICS APPROVAL DECLARATION
Authors declare human ethics approval was not needed for this study.