Despite being used commonly in bovine medicine, information on reliability of point-of-care (POC) lactate meters is limited.
Despite being used commonly in bovine medicine, information on reliability of point-of-care (POC) lactate meters is limited.
To determine the validity of 4 commercially available POC lactate meters in cattle.
Cattle with various diseases (n = 120).
Blood samples collected from the jugular vein were processed immediately on POC lactate meters. Plasma l-lactate concentration was measured by the enzymatic-colorimetric method (ELISA). Data were subjected to Friedman's test for comparison, Passing-Bablok regression and Bland-Altman plot analyses for reliability, and receiver operating characteristics analysis for sensitivity (Se) and specificity (Sp).
The POC lactate meters were highly correlated with ELISA (r = 0.98–0.99) despite disagreements among units. When regressed on ELISA, blood l-lactate concentrations generated from Accutrend Plus and Lactate Pro were linear up to 16.6 and 15.7 mmol/L, respectively, whereas those generated from i-STAT and Lactate Scout were linear up to 19.5 and 19.7 mmol/L, respectively. All POC lactate meters had a Se of 100% with Sp of 95.7–98.6% at a plasma l-lactate cut-off concentration of 4 mmol/L. i-STAT had the best accuracy (99.0%) and precision (99.8%), the best linear fit (y = −0.13 + 1.04X) yielding the lowest bias (−6.6%) as well as the highest Se (100%) and Sp (98.6%).
Despite high correlation with the reference method, dilution is needed for Accutrend Plus/Lactate Pro and i-STAT/Lactate Scout if concentrations >15 and 20 mmol/L, respectively. i-STAT provided the most accurate and precise results.
enzyme-linked immunosorbent assay
packed cell volume
receiver operating characteristics
residual standard deviation
L-Lactate is produced from anaerobic glycolysis in the body. Increased blood l-lactate concentration has been used as an important therapeutic and prognostic marker in human and veterinary medicine.[1-7] For example, mean initial plasma l-lactate concentrations were reported be 6.2 ± 3.2 and 10.3 ± 3.2 mmol/L in survivor and nonsurvivor dogs with gastric dilatation-volvulus, respectively. Serial blood l-lactate concentration measurements in conjunction with clinical examination have been used in equine neonatal intensive care to determine prognosis. It also has been shown to be a useful prognostic indicator in cattle with respiratory disease and right displaced abomasum.[4, 9]
Laboratory-based methods to measure l-lactate concentration are available at hospitals, which may not be convenient to many field practitioners. In recent years, point-of-care (POC) lactate meters have become available to instantly measure blood l-lactate concentration in veterinary medicine. Commercially available, inexpensive, and portable lactate meters (Accutrend Plus,1 i-STAT,2 Lactate Pro,3 Lactate Scout4) are simple to operate and require only a small amount of blood. The i-STAT especially has been extensively used in veterinary hospitals and ambulatory clinics. In ruminants, studies reporting validation of POC lactate meters are limited.[9, 10] The objective of this study was to determine the validity of 4 commercially available POC lactate meters (Accutrend Plus,1 i-STAT,2 Lactate Pro,3 Lactate Scout4) for assessing blood l-lactate concentration as compared to plasma l-lactate measured by the enzymatic-colorimetric method (ELISA) in cattle.
One hundred and twenty cattle diagnosed with various diseases were admitted to the study. All procedures were approved by the Firat University Animal Experimentation Ethics Committee. Cattle with respiratory distress, dehydration, or diarrhea were selected to obtain a wide range of plasma l-lactate concentration and packed cell volume (PCV).
Blood samples were collected from the jugular vein from all cattle. A portion of the blood sample was transferred to test tubes5 including sodium fluoride and potassium oxalate for plasma separation. The remaining blood was used immediately for PCV determination and blood l-lactate determination with the POC lactate meters.
After centrifugation of test tubes at 3,000×g for 15 minutes, plasma samples were stored at −20°C for plasma l-lactate measurement within 2 months. Briefly, plasma l-lactate concentration was determined with 96-well microtiter plates by ELISA and employing lactate dehydrogenase specific to l-Lactate.6 Resultant color intensity was quantitated at 490 nm with μ-Quant microplate spectrophotometer.7 All POC lactate meters were calibrated and used in accordance with the manufacturer's specifications.
Single-use reagent strips were used for Accutrend Plus, Lactate Pro, and Lactate Scout, whereas cartridges were used for i-STAT. The measurement range is 0.8–21.7, 0.3–20, 0.8–23.3, and 0.5–25.0 mmol/L for Accutrend Plus, i-STAT, Lactate Pro, and Lactate Scout lactate meters, respectively. Although Accutrend Plus is based on an enzymatic spectrophotometry system, the other POC lactate meters (i-STAT, Lactate Pro, and Lactate Scout) are based on an enzymatic amperometric system for l-lactate detection. The cartridge of i-STAT and the test strips of Lactate Pro and Lactate Scout employ l-lactate oxidase to catalyze the oxidation of l-lactate. The Accutrend Plus strip has 4 layers. After the blood cells are separated in the second layer, chemical reaction takes place with lactate oxidase in the third layer. Detailed information on principles of the POC lactate meters is described elsewhere.[11, 12]
Data were first subjected to the PROC MEANS and UNIVARIATE procedures to determine descriptive statistics parameters (mean, range, standard deviation, range, median, and percentiles) as well as to assess normality.8 One observation was an outlier and was deleted from the dataset.
Because of not being normally distributed, l-lactate data were analyzed by Friedman's test to determine differences among the units.9 The effect of PCV on l-lactate concentrations was determined by regressing l-lactate concentrations measured by ELISA and POC lactate meters on PCV by the PROC REG procedure.8
ELISA was considered the reference method, whereas the POC lactate meters were considered test methods in the validation process. Accuracy and precision of the POC lactate meters were determined by concordance and Pearson's correlation coefficients.10 Because the data were not normally distributed, Passing-Bablok regression analysis was employed to compare the POC lactate meters for determination of best linear fit.[13, 14] Systematic, proportional, and random differences (biases) between the POC lactate meters and ELISA were determined by Bland-Altman plot analysis.10 Moreover, l-lactate concentrations measured by the POC lactate meters were regressed on l-lactate concentrations measured by ELISA to determine curvilinearity.9 The inflection point (concentration at which curvilinearity begins) was determined by the simple moving average approach on a set of 4 observations for slope determination at each execution by a commercial math software.11
Finally, receiver operating characteristics (ROC) curves were developed at the cut-off plasma l-lactate concentration of 4 mmol/L by ELISA to determine sensitivity (Se), specificity (Sp), positive likelihood ratio, and negative likelihood ratio of the POC lactate meters. The Se and Sp were compared by the z-test as described elsewhere. In data analyses, statistical significance was set at P < .05.
There were 5 cattle breeds: Simmental (n = 62), Swiss Brown (n = 28), Holstein (n = 27), Angus (n = 2), and Hereford (n = 1). The cattle ranged from 1 day to 9 years of age. Tentative diagnoses of cattle included neonatal problems (n = 42, neonatal infection, premature birth, omphalitis, neonatal diarrhea), alimentary disorders (n = 41, traumatic reticuloperitonitis, prolapsed rectum, diarrhea, abomasal displacement-volvulus, vagus indigestion, simple indigestion), respiratory disorders (n = 28, pneumonia, laryngeal edema, aspiration pneumonia), traumatic pericarditis (n = 4), uterine torsion (n = 3), theileriosis (n = 1), and uroabdomen (n = 1).
The PCV ranged from 17 to 61% (mean, 36%; median, 35%, IQR, 41%). The PCV did not affect l-lactate concentrations measured by ELISA and the POC lactate meters. When regressed on PCV (Y = b0 + b1X), slopes of lactate concentrations were not different from 0 and were 0.03 (P = .51), 0.04 (P = .41), 0.05 (P = .27), 0.001 (P = .98), and 0.02 (P = .70) for ELISA, Accutrend Plus, i-STAT, Lactate Pro, and Lactate Scout, respectively (figures not shown).
Descriptive statistical parameters of plasma or blood lactate concentrations measured by the reference method and the POC lactate meters are shown in Figure 1. Plasma l-lactate concentrations measured by ELISA ranged from 0.32 to 24.59 mmol/L (mean, 4.72 ± 0.46 mmol/L). Among the POC lactate meters, the lowest (0.3 mmol/L) and highest (20.1 mmol/L) blood l-lactate concentrations were measured by i-STAT. The median blood l-lactate concentrations differed among the POC l-lactate meters as follows: Accutrend Plus (4.5 mmol/L) > Lactate Scout (3.1 mmol/L) > i-STAT (2.9 mmol/L) > Lactate Pro (2.3 mmol/L) (P < .001; Fig 1).
The PCV was not correlated with l-lactate concentrations measured by ELISA and POC lactate meters. Plasma l-lactate concentration measured by ELISA was positively correlated with blood l-lactate concentrations measured by all of the POC lactate meters (r = 0.98–0.99, P < .0001 for all). Concordance and Pearson's correlation coefficients indicated that i-STAT had the best accuracy (99.0%) and precision (99.8%) with ELISA, followed by Lactate Scout, Lactate Pro, and Accutrend Plus (Table 1).
|POC Lactate Meter||Concordance Correlation Coefficient||95% CI||Precision (Pearson ρ)||Accuracy (Bias Correction Factor, Cb)|
Among the POC lactate meters, i-STAT and Lactate Scout had slopes closer to 1 and residual standard deviation (RSD) closer to 0 in agreement with the reference method in Passing-Bablok regression (Table 2). Both i-STAT (P = .64) and Lactate Scout (P = .35) changed in a linear fashion with the reference method as reflected by no statistical difference with the identity line (Table 2; Fig 2). Both Accutrend Plus (Y = 1.34 + 1.13X − 0.02X2, R2 = 0.97, P < .0001) and Lactate Pro (Y = −0.35 + 1.11X − 0.01X2, R2 = 0.96, P < .0001) fitted better to a curvilinear line as reflected by improved R2 and deviation from the identity line. Based on the simple moving average approach, Accutrend Plus and Lactate Pro lost linearity at blood l-lactate concentrations higher than 16.6 and 15.7 mmol/L, respectively. The i-STAT and Lactate Scout maintained linearity up to blood l-lactate concentrations of 19.5 and 19.7 mmol/L, respectively (Fig 2).
|POC Lactate Meter||Differences||Cusum Test for Validity of Linearity|
|Intercept A (95% CI)||Slope B (95% CI)||RSD (± RSD Interval)|
|Accutrend Plus (Ymean = 5.9 ± 0.4 mmol/L)||1.57 (1.47/1.72)||0.95 (0.92/0.98)||0.78 (−1.53/1.53)||P = .07|
|i–STAT (Ymean = 4.6 ± 0.4 mmol/L)||−0.13 (−0.16/0.09)||1.04 (1.01/1.06)||0.57 (−1.12/1.12)||P = .64|
|Lactate Pro (Ymean = 4.3 ± 0.4 mmol/L)||0.04 (−0.06/0.17)||0.88 (0.82/0.92)||0.73 (−1.43/1.43)||P < .01|
|Lactate Scout (Ymean = 5.1 ± 0.4 mmol/L)||0.43 (0.35/0.50)||0.99 (0.96/1.03)||0.62 (−1.22/1.22)||P = .35|
Table 3 summarizes bias (difference as % of average) evaluation by Bland-Altman bias plots. Both i-STAT (−6.6%) and Lactate Pro (−7.5%) achieved the lowest difference among the test methods. The slope of % bias was not different from 0 only for Lactate Pro (Table 3; Fig 2). Standard error of estimate (Sy.x) was the lowest for i-STAT. Accutrend Plus had the highest Sy.x and largest bias range. Overall, except for the i-STAT, the other POC lactate meters underestimated plasma l-lactate concentration by varying levels (Fig 2).
|POC Lactate Meter (Y)||Difference (% of Average)|
|Mean (95% CI)||S y.x||Lower Limit||Upper Limit||Regression Equation|
|Accutrend Plus||49.9 (42.3/57.4)||41.6||−31.7 (−44.6/−18.7)||131.4 (118.5/144.3)||Y = 87.6 ± 3.6 − 7.1 ± 0.5X|
|i–STAT||−6.6 (−9.4/−3.8)||15.4||−36.8 (−41.6/−32.0)||23.6 (18.8/28.4)||Y = −11.0 ± 1.9 + 0.9 ± 0.3X|
|Lactate Pro||−7.5 (−12.1/−2.9)||25.5||−57.5 (−65.4/−49.5)||42.4 (34.5/50.4)||Y = −5.2 ± 3.2 − 0.5 ± 0.5X|
|Lactate Scout||20.9 (16.2/25.9)||25.8||−29.8 (−37.8/−21.7)||71.5 (63.5/79.6)||Y = 35.2 ± 2.8 − 2.9 ± 0.4X|
Based on the reference method, 49 cattle had <4 mmol/L plasma l-lactate concentration, whereas 70 cattle had >4 mmol/L plasma lactate concentration. At the cut-off plasma l-lactate concentration (4 mmol/L), the ROC curves developed for the POC lactate meters are presented in Figure 3 and summarized in Table 4. All POC lactate meters had a Se of 100%. i-STAT had the highest Sp (98.6%), whereas Lactate Scout had the lowest Sp (95.7%). Moreover, +LR was the highest for the i-STAT (70.%), with –LR of 0%. Although the Accutrend Plus and i-STAT exceeded the cut-off concentration, Lactate Pro and Lactate Scout had lower criterion than the cut-off concentration. However, area under the curve values was not different among the POC lactate meters.
|Value||95% CI||Value||95% CI||+LR||−LR||Mean||SE||95% CI||z||P <|
|Contrasts||Difference||SE||95% CI||z||P <|
|Accutrend Plus versus i-STAT||0.0020||0.0018||−0.0014 to 0.0055||1.17||.24|
|Accutrend Plus versus Lactate Pro||0.0003||0.0019||−0.0035 to 0.0041||0.15||.88|
|Accutrend Plus versus Scout||0.0004||0.0017||−0.0029 to 0.0038||0.26||.80|
|i-STAT versus Lactate Pro||0.0018||0.0017||−0.0016 to 0.0051||1.04||.30|
|i-STAT versus Lactate Scout||0.0025||0.0020||−0.0015 to 0.0064||1.23||.22|
|Lactate Pro versus Lactate Scout||0.0007||0.0011||−0.0014 to 0.0029||0.66||.51|
Plasma l-lactate concentration measurement gives important clues about the patient's prognosis and disease severity. Higher plasma l-lactate concentrations at admission indicate poorer prognosis in dogs with gastric dilatation-volvulus, babesiosis, and idiopathic immune-mediated hemolytic anemia[6, 17, 18] and in horses with surgical colic.[7, 19] Hyperlactatemia can be a result of increased lactate production, decreased clearance of lactate, or a combination. Clinically, the most common cause of hyperlactatemia is tissue hypoxia that occurs with hypoperfusion or poor oxygen delivery.[1, 20] Abomasal ischemia is the main cause for hyperlactatemia in cows with abomasal volvulus[21, 22] and plasma l-lactate concentration has been used as a prognostic indicator in cattle with abomasal displacement. The median plasma l-lactate concentration was reported to be 0.54 mmol/L (n = 34), 5.88 mmol/L (n = 18, range: 4.88–9.3), and 3.23 mmol/L (n = 36, range: 2.0–5.8) by the i-STAT in healthy lactating cows, cows with negative outcome right-sided displaced abomasum, and positive outcome right-sided displaced abomasum, respectively. Hypoxia causes increased plasma l-lactate concentrations in newborn calves and high l-lactate concentrations contribute to a progressive primary metabolic acidosis in hypoxic calves. Calves born with assistance were found to have higher plasma l-lactate concentrations than those born unassisted. Both l-lactate and, especially, d-lactate contributed to metabolic acidosis in diarrheic calves, but all POC lactate meters used in the present report measured only blood l-lactate concentration. Plasma l-lactate concentration may be used as a prognostic marker in cattle with respiratory disease as well. Plasma l-lactate concentrations of calves suffering from acute bronchopneumonia increased with severity of the disease, and a plasma l-lactate concentration >3.6 mmol/L measured by the enzymatic-photometric method was a reliable prognostic indicator for mortality within 24 hours. Increased blood l-lactate concentrations were reported in some case studies.[26, 27] The blood l-lactate concentration of a cow with cardiac lymphoma was determined to be 3.4 mmol/L using the Lactate Pro. A limited number of clinical studies have been performed using serial l-lactate concentration measurements to determine the prognosis in cattle.
Laboratory-based methods and expensive specific equipment for determination of blood l-lactate concentration seem to challenge the utility of lactate determination in bovine medicine, especially in field settings. However, the validity of the lactate measurement in bovine medicine should be further assessed. POC lactate meters have been validated in dogs and horses.[28-30] The i-STAT, Lactate Pro, Lactate Scout, and Accutrend12 were reported to be in agreement with the laboratory reference method to measure plasma lactate concentration in dogs. Accutrend was reported to accurately measure plasma lactate concentrations in adult horses and Lactate Scout was considered a reliable lactate meter in the measurement of plasma lactate concentration in equine neonatal intensive care cases. In ruminants, studies validating POC lactate meters are limited. In cows with respiratory disease, Accusport11 was highly correlated with the reference method, but significantly overestimated plasma l-lactate concentration. Similarly, despite a high positive correlation, Accutrend Plus tended to overestimate plasma l-lactate concentration in this study. The performance of Lactate Scout also was evaluated in dairy cows and calves. Blood samples with lithium heparin and sodium fluoride were used in that experiment. Concentrations of l-lactate measured in lithium heparin were correlated when analyzed by Lactate Scout and the reference method (r = 0.75 for dairy cows and r = 0.98 for calves). Concentrations of l-lactate measured in sodium fluoride were higher when analyzed by Lactate Scout (1.85 ± 0.6 mmol/L) compared with the reference method (0.72 ± 0.45 mmol/L), indicating that blood samples with sodium fluoride should not be used for Lactate Scout. Blood samples without any anticoagulant were used in the POC lactate meters in this study.
Despite differences in median values (Fig 1), plasma l-lactate concentration measured by the reference method was positively correlated with blood l-lactate concentrations determined by all of the POC lactate meters (r = 0.98–0.99, P < .0001). i-STAT had the highest accuracy (99.0%) and precision (99.8%) with ELISA, followed by Lactate Scout (Table 1). Nevertheless, correlation analysis is not enough to judge the reliability and reproducibility of POC lactate meters. Classical regression is based on the assumption that only the dependent variable has random errors, which is rarely the case. Thus, in model comparison, Deming or Passing-Bablok regression procedures are widely accepted because they consider random errors in both the independent and dependent variables. Although Deming regression procedure is suitable for parametric methods, where data are normally distributed, the Passing-Bablok procedure is suitable for nonparametric methods, where data are not normally distributed. In this experiment, data were not normally distributed (Fig 1), thus the degree of fit was subjected to Passing-Bablok regression analysis (Fig 2; Table 2). Deviation of the i-STAT (P = .64) and Lactate Scout (P = .35) from the identity line was insignificant. Moreover, their random errors were the lowest as compared with other POC lactate meters (Table 2). Moreover, the simple moving average approach ascertained that i-STAT and Lactate Scout maintained linearity up to blood l-lactate concentrations of 19.5 and 19.7 mmol/L, respectively, whereas Accutrend Plus and Lactate Pro became curvilinear at blood l-lactate concentration of 16.6 and 15.7 mmol/L, respectively (Fig 2).
To evaluate bias, a sophisticated approach was developed, in which differences between 2 methods expressed as percentage are plotted against the average of data produced from both methods. This allows identification of outliers as well as systematic, proportional, and random biases. Mean bias was lowest for the i-STAT (−6.6%) with the lowest Sy.x (15.4) and highest for the Accutrend Plus (49.9%) with the lowest Sy.x (41.6) (Table 3).
An increased PCV has been identified as a confounder when measuring lactate concentrations in horses using the Accusport,13  Similar to the present data, lack of correlation between lactate concentration measured by Lactate Scout and PCV has been identified in dogs. Moreover, changes in plasma/blood l-lactate concentrations were independent of PCV as determined by a regression coefficient not different from 0.
Different lactate concentration cut-off values have been used in dogs and horses. The cut-off for optimal prediction of outcome or gastric necrosis in dogs with gastric dilatation-volvulus was determined to be 7.4 mmol/L at admission to the hospital. Corley et al used a cut-off of 2.5 mmol/L in critically ill neonatal foals and Castagnetti et al chose a higher lactate concentration cut-off (5 mmol/L) in critically ill neonatal foals. There have not been enough studies performed in cattle to investigate the association between plasma l-lactate concentration and prognosis in different diseases. Figueiredo et al measured median plasma l-lactate concentration by the i-STAT as 5.88 mmol/L (n = 18; range 4.88–9.3) and 3.23 mmol/L (n = 36; range 2.0–5.8) in cows with negative outcome right-sided displaced abomasum and positive outcome right-sided displaced abomasum, respectively. Wittek et al determined mean plasma l-lactate concentration as 4.8 ± 2.8 mmol/L in cows with abomasal volvulus. Plasma l-lactate concentrations >3.6 mmol/L were determined to be a reliable prognostic indicator for mortality within 24 hours in calves with acute bronchopneumonia. A cut-off concentration of 4 mmol/L plasma l-lactate may be useful in cattle (Fig 3). All POC lactate meters had Se of 100% at the cut-off plasma lactate concentration (4 mmol/L). i-STAT had the highest Sp (98.6%), whereas Lactate Scout had the lowest Sp (95.7%) (Table 4).
Point-of-care lactate meters tested in this study are used in veterinary practice, and it is important to determine their accuracy and precision for comparison and obtaining reliable results. Plasma and blood l-lactate concentrations were independent of PCV. All 4 POC lactate meters produced results that were in agreement with the reference method (r = 0.98–0.99). All POC lactate meters had a Se of 100% with Sp of 95.7–98.6% at a plasma l-lactate cut-off concentration of 4 mmol/L. However, Accutrend Plus and Lactate Pro were no longer linear at blood l-lactate concentrations >16.6 and 15.7 mmol/L, respectively. i-STAT and Lactate Scout maintained linearity up to blood l-lactate concentrations of 19.5 and 19.7 mmol/L, respectively, when results were regressed on those generated by the reference method. i-STAT had the best accuracy (99.0%) and precision (99.8%), the best linear fit (y = −0.13 + 1.04X) yielding the lowest bias (−6.6%), and highest Se (100%) and Sp (98.6%).
This research was supported by The Scientific and Technological Research Council of Turkey (TUBITAK) project no: 111O045. We thank Asst. Prof. Bulent Cavusoglu (Department of Electrical and Electronics Engineering, Ataturk University) for MATLAB usage.
Conflict of Interest Declaration: Authors disclose no conflict of interest.
Roche Diagnostics, Mannheim, Germany
Abbott Point of Care, Abbott Laboratories, Chicago, IL
Arkray Inc, Kyoto, Japan
SensLab GmbH, Leipzig, Germany
BD Vacutainer, Becton, Dickinson and Company, Franklin Lakes, NJ
l-lactate assay kit; Eton Bioscience Inc, San Diego, CA
BioTek Instruments, Inc, Winooski, VT
SAS 2002 User's Guide Statistics, Version 9th Statistical Analysis System SAS Inst., Inc, Cary, NC
SPSS Software, Version 20; IBM SPSS, Somers, NY
MedCalc Software Version 126.96.36.199, Ankara, Turkey
MATLAB Software, Version, 2012a, MathWorks, Natick, MA
Accutrend, Roche Diagnostics, Mannheim, Germany
Accusport, Boehringer Mannheim, Germany