• View in gallery

    Error grid analysis (A) and Bland-Altman plot (B) comparing blood BHB concentrations measured with a dual-purpose POC meter with serum BHB concentrations measured with an LBA in samples obtained from pregnant mixed-breed ewes with different feeding regimens. Sheep in group 1 (triangles; n = 12) were fed 100% of their nutrient requirements, those in group 2 (squares; 12) were fed 50% of their nutrient requirements from 28 through 78 dGA, and those in group 3 (circles; 12) were fed 50% of their nutrient requirements from 28 dGA through parturition. A—The error grid comprises 4 zones. Zone A represents POC meter BHB concentrations within 20% of the LBA concentrations as well as when the POC BHB concentration was < 0.8 mmol/L and the corresponding LBA BHB concentration was also < 0.8 mmol/L. Zone B contains POC meter values that were > 20% of LBA values but would result in a benign treatment decision. Zone C contained POC meter BHB concentrations that would result in not treating a sheep when the LBA BHB cocentration was above the acceptable range, and zone D contained POC concentrations that would result in treating a sheep when the LBA BHB concentration was within the acceptable range. B—Bland-Altman difference plots were prepared, accounting for multiple observations per test subject. The dashed lines indicate the 95% limits of agreement (± 1.96 SD of the mean difference), and the solid line indicates the mean difference between methods.

  • View in gallery

    Error grid analysis (A) and Bland-Altman plot (B) comparing blood glucose concentrations measured with a dual-purpose POC meter with serum glucose concentrations measured with an LBA in samples from pregnant mixed-breed ewes with different feeding regimens. A—The error grid comprises 4 zones. Zone A represents POC meter glucose concentrations within 20% of the LBA concentrations as well as when the POC concentration was ≤ 50 mg/dL when the LBA concentration was also ≤ 50 mg/dL. Zone B contains POC meter concentrations that were > 20% of LBA concentrations but would result in a benign treatment decision or no treatment decision. Zone C contained POC meter glucose concentrations that would result in treating a sheep when the LBA glucose concentration was within the acceptable range, and zone D contained POC glucose concentrations that would result in not treating a sheep when the LBA was lower than the acceptable range. See Figure 1 for remainder of key.

  • 1. Brozos C, Mavrogianni VS, Fthenakis GC. Treatment and control of peri-parturient metabolic diseases: pregnancy toxemia, hypocalcemia, hypomagnesemia. Vet Clin North Am Food Anim Pract 2011; 27:105113.

    • Search Google Scholar
    • Export Citation
  • 2. Bulgin M. Diseases of the periparturient ewe. In: Youngquist RS, Threlfall WR, eds. Current therapy in large animal theriogenology. 2nd ed. St Louis: WB Saunders Co, 2007; 695700.

    • Search Google Scholar
    • Export Citation
  • 3. Federici MO, Benedetti MM. Ketone bodies monitoring. Diabetes Res Clin Pract 2006; 74 (suppl 2): S77S81.

  • 4. Laffel L. Ketone bodies: a review of physiology, pathophysiology and application of monitoring to diabetes. Diabetes Metab Res Rev 1999; 15:412426.

    • Search Google Scholar
    • Export Citation
  • 5. Oetzel GR. Monitoring and testing dairy herds for metabolic disease. Vet Clin North Am Food Anim Pract 2004; 20:651674.

  • 6. Jones M, Miesner MD, Baird AN, et al. Diseases of the urinary system. In: Pugh DG, Baird AN, eds. Sheep and goat medicine. 2nd ed. St Louis: WB Saunders Co, 2012; 325360.

    • Search Google Scholar
    • Export Citation
  • 7. Iwersen M, Falkenberg U, Voigtsberger R, et al. Evaluation of an electronic cowside test to detect subclinical ketosis in dairy cows. J Dairy Sci 2009; 92:26182624.

    • Search Google Scholar
    • Export Citation
  • 8. Panousis N, Brozos C, Karagiannis I, et al. Evaluation of Precision Xceed meter for on-site monitoring of blood [beta]-hydroxybutyric acid and glucose concentrations in dairy sheep. Res Vet Sci 2012; 93:435439.

    • Search Google Scholar
    • Export Citation
  • 9. Voyvoda H, Erdogan H. Use of a hand-held meter for detecting subclinical ketosis in dairy cows. Res Vet Sci 2010; 89:344351.

  • 10. Precision Xtra blood ñ-ketone test strips [package insert]. Alameda, Calif: Abbott Diabetes Care Inc, 2006.

  • 11. Precision Xtra blood glucose test strips [package insert]. Alameda, Calif: Abbott Diabetes Care Inc, 2009.

  • 12. Bland J, Altman D. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat 2007; 17:571582.

    • Search Google Scholar
    • Export Citation
  • 13. Clarke W, Cox D, Gonder-Frederick L, et al. Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care 1987; 10:622628.

    • Search Google Scholar
    • Export Citation
  • 14. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 327:307310.

    • Search Google Scholar
    • Export Citation
  • 15. Tonyushkina K, Nichols JH. Glucose meters: a review of technical challenges to obtaining accurate results. J Diabetes Sci Technol 2009; 3:971980.

    • Search Google Scholar
    • Export Citation
  • 16. Morris JD, Fernandez JM, Chapa AM, et al. Effects of sample handling, processing, storage, and hemolysis on measurements of key energy metabolites in ovine blood. Small Ruminant Res 2002; 43:157166.

    • Search Google Scholar
    • Export Citation
  • 17. Ginsburg BH. Factors affecting blood glucose monitoring: sources of errors in measurement. J Diabetes Sci Technol 2009; 3:903913.

    • Search Google Scholar
    • Export Citation
  • 18. Hytten F. Blood volume changes in normal pregnancy. Clin Haematol 1985; 14:601612.

  • 19. Coldman MF, Good W. The distribution of sodium, potassium and glucose in the blood of some mammals. Comp Biochem Physiol 1967; 21:201206.

    • Search Google Scholar
    • Export Citation
  • 20. MacKay E. The distribution of glucose in human blood. J Biol Chem 1932; 97:685689.

  • 21. Zhang DJ, Elswick RK, Miller WG, et al. Effect of serum-clot contact time on clinical chemistry laboratory results. Clin Chem 1998; 44:13251333.

    • Search Google Scholar
    • Export Citation

Advertisement

Evaluation of a point-of-care glucose and β-hydroxybutyrate meter operated in various environmental conditions in prepartum and postpartum sheep

View More View Less
  • 1 Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins, CO 80523.
  • | 2 Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins, CO 80523.
  • | 3 Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins, CO 80523.
  • | 4 Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins, CO 80523.
  • | 5 Department of Animal Sciences, College of Agriculture Sciences, Colorado State University, Fort Collins, CO 80523.
  • | 6 Department of Animal Sciences, College of Agriculture Sciences, Colorado State University, Fort Collins, CO 80523.

Abstract

Objective—To compare β-hydroxybutyrate (BHB) and glucose concentrations measured with a dual-purpose point-of-care (POC) meter designed for use in humans and a laboratory biochemical analyzer (LBA) to determine whether the POC meter would be reliable for on-farm measurement of blood glucose and BHB concentrations in sheep in various environmental conditions and nutritional states.

Animals—36 pregnant mixed-breed ewes involved in a maternal feed restriction study.

Procedures—Blood samples were collected from each sheep at multiple points throughout gestation and lactation to allow for tracking of gradually increasing metabolic hardship. Whole blood glucose and BHB concentrations were measured with the POC meter and compared with serum results obtained with an LBA.

Results—464 samples were collected. Whole blood BHB concentrations measured with the POC meter compared well with LBA results, and error grid analysis showed the POC values were acceptable. Whole blood glucose concentrations measured with the POC meter had more variation, compared with LBA values, over the glucose ranges evaluated. Results of error grid analysis of POC-measured glucose concentrations were not acceptable, indicating errors likely to result in needless treatment with glucose or other supplemental energy sources in normoglycemic sheep.

Conclusions and Clinical Relevance—The POC meter was user-friendly and performed well across a wide range of conditions. The meter was adequate for detection of pregnancy toxemia in sheep via whole blood BHB concentration. Results should be interpreted with caution when the POC meter is used to measure blood glucose concentrations.

Abstract

Objective—To compare β-hydroxybutyrate (BHB) and glucose concentrations measured with a dual-purpose point-of-care (POC) meter designed for use in humans and a laboratory biochemical analyzer (LBA) to determine whether the POC meter would be reliable for on-farm measurement of blood glucose and BHB concentrations in sheep in various environmental conditions and nutritional states.

Animals—36 pregnant mixed-breed ewes involved in a maternal feed restriction study.

Procedures—Blood samples were collected from each sheep at multiple points throughout gestation and lactation to allow for tracking of gradually increasing metabolic hardship. Whole blood glucose and BHB concentrations were measured with the POC meter and compared with serum results obtained with an LBA.

Results—464 samples were collected. Whole blood BHB concentrations measured with the POC meter compared well with LBA results, and error grid analysis showed the POC values were acceptable. Whole blood glucose concentrations measured with the POC meter had more variation, compared with LBA values, over the glucose ranges evaluated. Results of error grid analysis of POC-measured glucose concentrations were not acceptable, indicating errors likely to result in needless treatment with glucose or other supplemental energy sources in normoglycemic sheep.

Conclusions and Clinical Relevance—The POC meter was user-friendly and performed well across a wide range of conditions. The meter was adequate for detection of pregnancy toxemia in sheep via whole blood BHB concentration. Results should be interpreted with caution when the POC meter is used to measure blood glucose concentrations.

Pregnancy toxemia, also known as pregnancy ketosis, is a metabolic disorder that can develop as a result of relative undernourishment or excessive body condition in late-gestation ewes, typically those bearing twin or multiple fetuses.1 A suspicion of clinical pregnancy toxemia is supported by a ewe's history and signalment as well as a progression of signs such as anorexia, lethargy, weakness, and neurologic abnormalities.2 When undetected or left untreated, the condition usually results in death of the ewe and its fetuses.

Treatment of pregnancy toxemia can have disappointing results and be expensive and time-consuming, particularly when multiple sheep are affected or when detection is delayed, resulting in an increase in the severity of clinical signs. Therefore, a rapid and accurate diagnosis of ketosis could increase the possibility of a successful treatment outcome while decreasing the likelihood of illness and death of affected ewes and lambs and lessening economic costs associated with delayed treatment and sheep loss.

Sheep producers and veterinarians typically use urine ketone test strips to evaluate ketone status in late-gestation ewes. The urine ketone strips can be used to semiquantitatively measure the amount of acetoacetate in urine but not of BHB, which is the primary circulating ketone body.3 Accuracy of the urine ketone test strips in dairy cows is reportedly variable, and results can be affected by environmental conditions (eg, ambient humidity or temperature), user characteristics (eg, nature of handling and interval from testing to reading), some medications, and pigmenturia.4,5 Urine collection can be difficult and is stressful for ewes because urination is most reliably induced by restraining the ewe and holding off the nares to prevent breathing.6 The ewe eventually becomes distressed and postures to urinate.

A POC meter designed to measure blood glucose and BHB concentration in humans has been evaluated for detection of ketosis in dairy cattle and dairy sheep.7–9 The only other method for measuring blood BHB concentrations in prepartum animals is with an LBA. A POC glucometer is available for veterinary use; however, it has not been evaluated or marketed for use in sheep.

Although blood glucose concentrations vary in sheep with pregnancy toxemia, they can be used to assist in evaluating response to treatment. Therefore, a reliable POC meter capable of measuring both BHB and glucose concentrations would be beneficial and economical for sheep producers and veterinarians. The objective of the study reported here was to compare blood BHB and glucose concentrations measured with a dual-purpose POC meter with those of an LBA in various environmental conditions and throughout gestation and lactation in healthy and nutritionally stressed ewes at high risk for developing pregnancy toxemia. The hypothesis was that the POC meter would yield results of sufficient accuracy to support its use for measurement of blood BHB and glucose concentrations in sheep on farms.

Materials and Methods

Animals—Thirty-six adult mixed-breed white-face ewes were used for the study. Each ewe was identified with a unique ear tag number. The sheep were concurrently involved in an unreported study of maternal nutritional energy restriction and were assigned equally into 3 study groups. Group 1 (control group) was fed 100% of their nutrient requirements, group 2 was fed 50% of their nutrient requirements from 28 through 78 dGA, and group 3 was fed 50% of their nutrient requirements from 28 dGA through parturition. Nutrient requirements were based on the National Research Council's recommendations for twin-bearing ewes in early and late gestation. The ewes were housed in individual sand pens with individual automatic waterers and feed pans to control their feed intake. All pens were located within a 3-sided barn that was not temperature regulated.

Ovulation in the ewes was synchronized, after which they were exposed to a ram at 12-hour intervals for a 2-day period. The second day of exposure to the ram was used as the breeding date for the study. Transabdominal ultrasonography was performed on ewes observed to have been bred at 45 days after breeding (45 dGA), and ewes confirmed to be pregnant with twins were selected for the maternal malnutrition study. The study protocol was approved by the Colorado State University Institutional Animal Care and Use Committee.

Blood sample collection—Blood samples were collected from each ewe throughout gestation at the following intervals: once at 3 time points from 20 through 22 dGA, 75 through 80 dGA, and 88 through 90 dGA; and weekly thereafter from 91 dGA through parturition (at approx 147 to 152 dGA). Postpartum samples were collected once per day until serum BHB concentrations were < 1.0 mmol/L, then every 2 weeks for 6 weeks, and again when lambs were weaned (approx 11 to 12 weeks after parturition). For sample collection, each ewe was restrained within its pen by 1 person while a second person obtained the blood sample via jugular venipuncture with an 18-gauge, 2.8-cm needle and 12-mL syringe.

BHB and glucose concentration measurement—Immediately after collection, blood samples were analyzed with each of 2 dual-purpose POC metersa: one calibrated for BHB measurement and the other calibrated for glucose measurement. Meter calibration was performed in accordance with the manufacturer's instructions by insertion of a calibrator key specific to the BHB or glucose test strips into the meter. A lot code was matched between the calibrator key and the test strips. Calibration was performed prior to each blood collection day and when lot numbers changed between test strips. The POC meters were designed to measure whole blood BHB concentrations between 0.0 and 8.0 mmol/L10 and glucose concentrations from 20 to 500 mg/dL.11 Samples were immediately retested with a new test strip whenever the meter displayed an error code or provided a result of high or low.

Before the study began, the precision of the POC meter had been tested by use of whole blood samples collected from a nonpregnant sheep and a ketotic cow. A whole blood sample (3 mL) collected by jugular or coccygeal venipuncture had been immediately tested 10 times with the 2 POC meters calibrated for study BHB and glucose measurements, and the CV had been calculated for each analyte.

The POC meters were evaluated in various conditions, which included environmental temperatures ranging from −10° to 25°C, dusty conditions, and wet (rain and snow) conditions given the open barn design. The 2 meters and test strips were stored in small, insulated coolers with chemical hand warmersb when environmental temperatures were lower than the meters' operating temperature range (10° to 50°C). Extra test strips were stored in their original packaging between 10° and 25°C.

The remainder of each blood sample was transferred into evacuated serum tubesc and stored on ice prior to transport to the Colorado State University Clinical Pathology Laboratory. Within 3 hours after sample collection, the serum tubes containing clotted blood were centrifuged for 5 minutes at 1,000 × g, serum was harvested, and serum samples were refrigerated overnight. The serum BHB and glucose concentrations were measured with an LBAd within 24 hours after blood sample collection. The LBA was calibrated in accordance with the manufacturer's instructions. The analyzer measured serum BHB concentration with a spectrophotometric method and glucose concentrations with a hexokinase method.

Statistical analysis—Data were analyzed with the aid of a commercial software package.e The D'Agostino-Pearson test indicated that the data from blood sample testing were not normally distributed, so nonparametric analyses were used. The Wilcoxon test was used to determine whether the POC meter results for blood BHB and glucose concentrations were significantly different from the LBA results for serum BHB and glucose concentrations. The Spearman rank correlation coefficient (ρ) was calculated to assess correlations between POC meter and LBA results. The percentage difference between the POC meter and LBA results was calculated for each paired whole blood and serum BHB or glucose value. Percentages of POC meter results for BHB and glucose that were within 5%, 10%, and 20% above or below the LBA values were then calculated.

Agreement between results of the POC meter and the LBA was assessed via Bland-Altman methods for multiple observations per test subject,12 and plots were constructed so that the identities of sheep could be discerned by feeding group. Bland-Altman plots describe the limits of agreement between 2 methods as the mean difference between measurements ± 1.96 SD. Through examination of the plotted data, one can detect systematic biases (mean difference) and proportional biases (positive or negative slope in the data). The LBA was considered the gold standard method in these analyses.

Error grid plots were developed to theoretically evaluate the results of clinical decisions regarding whether the POC meter BHB or glucose concentration should be used rather than the LBA concentration. To do this, the Clarke error grid analysis method and acceptance criteria were modified because the critical limits pertinent to sheep are different from those used in humans.13 On the BHB plot, zone A represented POC meter BHB concentrations within 20% above or below the LBA concentrations as well as when the POC BHB concentration was < 0.8 mmol/L and the corresponding LBA BHB concentration was also < 0.8 mmol/L. Zone B contained POC meter concentrations that were > 20% of LBA concentrations but would result in a benign treatment decision. Zone C contained POC meter BHB concentrations that would result in not treating a sheep when the LBA BHB concentration was higher than the acceptable range, and zone D contained POC concentrations that would result in treating a sheep when the LBA BHB concentration was within the acceptable range. These zones were established with the presumption that treatment would be deemed necessary in a clinical setting when the BHB concentration was ≥ 0.8 mmol/L. No lower limit for intervention was deemed necessary for this study.

For the glucose concentration error grid plot, zone A represented the POC meter glucose concentrations that were within 20% above or below the LBA concentrations as well as when the POC concentration was ≤ 50 mg/dL when the LBA concentration was also ≤ 50 mg/dL. Zone B contained the POC meter glucose concentrations that were > 20% of LBA concentrations but would result in a benign treatment decision or no treatment decision. Zone C contained the POC glucose concentrations that would result in treating a sheep when the LBA glucose concentration was within the acceptable range, and zone D contained POC glucose concentrations that would result in not treating a sheep when the LBA concentration was lower than the acceptable range. The zones were established with the presumption that treatment would be deemed necessary when the glucose concentration was ≤ 50 mg/dL. No upper limit for treatment intervention was deemed necessary for glucose concentrations. The meter would be considered acceptable for on-farm use if at least 95% of the BHB or glucose concentration were within zones A and B. The BHB and glucose concentration intervention guidelines were not implemented in this study. Values of P ≤ 0.05 were considered significant. Device results are summarized as mean ± SD.

Results

Blood samples—Four hundred sixty-four blood samples were obtained from the 36 ewes over the study period. Three blood samples were not used in BHB or glucose concentration statistical analyses because of gross hemolysis that occurred during collection. Additionally, 16 blood glucose concentrations were reported as low (< 20 mg/dL) by the POC meter and remained low despite retesting. The corresponding serum LBA glucose concentrations ranged from 10 to 71 mg/dL. These values were omitted from the statistical analyses. Error codes were randomly displayed by the meter but were not documented. The manufacturer's instructions recommended verification that the meter had been properly calibrated and retesting of samples with a new test strip. Following these recommendations led to resolution of the error code display and obtainment of valid measurements, which were used in the statistical analyses.

Precision testing—Precision testing of the POC meters revealed a smaller CV for BHB concentrations than for glucose concentrations, and the CVs were similar to those reported by the meter manufacturer.10 The high BHB concentration C V, obtained through use of bovine blood samples, was 3.56% (mean ± SD, 3.92 ± 0.14 mmol/L), and the low BHB concentration CV, obtained through use of ovine blood samples, was not relevant because the mean was close to 0 mmol/L (0.12 ± 0.042 mmol/L). The CV reported by the meter manufacturer was 3.8% for the mid BHB concentration range (2.36 ± 0.09 mmol/L) and 3.1% for the high range (6.32 ± 0.2 mmol/L), and a CV was not reported for the low BHB concentration range (0.34 ± 0.03 mmol/L). The CV for the glucose concentration (sheep blood sample) was 5.8% (52.5 ± 3.0 mg/dL), compared with the manufacturer's reported CV of 6% for the low glucose concentration range (43.9 ± 2.3 mg/dL).11

POC meter and LBA comparisons—The POC meter results for blood BHB and glucose concentrations differed significantly (P < 0.001) from the LBA results for serum concentrations overall and by feeding group (Table 1). The correlation between the POC meter and LBA values was 0.934 (P < 0.001) for BHB concentration and 0.620 (P < 0.001) for glucose concentration.

Table 1—

Median (interquartile range) blood concentrations of BHB and glucose as measured with a POC meter and serum concentrations as measured with an LBA in 36 pregnant mixed-breed ewes with different feeding regimens* (12 sheep/group).

 BHB (mmol/L)Glucose (mg/dL)
AnimalsNo. of samples testedPOCLBANo. of samples testedPOCLBA
All4610.30 (0.10–0.70)0.42 (0.28–0.73)44642 (34–54)54 (48–61)
Group 1*1610.30 (0.20–0.70)0.44 (0.27–0.74)15843 (35–54)55 (50–62)
Group 2*2000.30 (0.10–0.70)0.38 (0.27–0.69)19341 (33–53)52 (47–60)
Group 3*1000.30 (0.20–0.75)0.47 (0.30–0.76)9540 (33–55)53 (45–62)

Sheep in group 1 were fed 100% of their nutrient requirements, those in group 2 were fed 50% of their nutrient requirements from 28 through 78 dGA, and those in group 3 were fed 50% of their nutrient requirements from 28 dGA through parturition.

The LBA was considered the more accurate method. All median values differ significantly (P < 0.001) between devices.

The percentage difference in values obtained with the POC meter and LBA indicated the POC meter results for blood BHB and glucose concentrations were quite different from the LBA results. For BHB values, 13%, 21%, and 37% of POC meter concentrations were within 5%, 10% and 20% above or below the LBA values, respectively. Because meter performance criteria have not been established for blood BHB concentrations, samples were analyzed further when the LBA-measured BHB concentrations were > 0.8 mmol/L, which would be more clinically relevant for making treatment decisions. In that further analysis, 37%, 48%, and 71% of POC meter values were within 5%, 10%, and 20% of the LBA values, respectively. Patterns were similar to those of BHB for blood glucose concentrations as measured with the POC meter, for which 11%, 23%, and 44% of the POC meter glucose values were within 5%, 10%, and 20% above or below the LBA values.

The Bland-Altman difference plot revealed that the mean difference in BHB concentrations between devices was small, with narrow limits of agreement (Figure 1). Visual examination of the plot revealed a small negative constant bias in the BHB data and an increase in data scatter as BHB concentration increased. Eleven upper and 5 lower outlier values were identified, and the blood samples to which these pertained had been collected across multiple dates; however, 3 were from the same collection date and 3 others were from the same sheep over 3 weeks. When the plot was examined by feeding group, a small positive bias was evident in the data points for the group fed 50% of their nutrient requirements from 28 through 78 dGA and a negative bias was identified in the data points for the group fed 50% of their nutrient requirements from 28 dGA through parturition. The difference plot for glucose concentration showed a larger mean difference with wide limits of agreement, negative constant bias, and more scatter in the data than was evident for BHB (Figure 2). Twelve upper and 3 lower outliers were identified; 6 values pertained to blood samples collected on the same day, and 3 values pertained to samples collected on a day when severe hemolysis was noticed in 3 discarded samples. Outlier data points represented approximately 3% of the samples collected.

Figure 1—
Figure 1—

Error grid analysis (A) and Bland-Altman plot (B) comparing blood BHB concentrations measured with a dual-purpose POC meter with serum BHB concentrations measured with an LBA in samples obtained from pregnant mixed-breed ewes with different feeding regimens. Sheep in group 1 (triangles; n = 12) were fed 100% of their nutrient requirements, those in group 2 (squares; 12) were fed 50% of their nutrient requirements from 28 through 78 dGA, and those in group 3 (circles; 12) were fed 50% of their nutrient requirements from 28 dGA through parturition. A—The error grid comprises 4 zones. Zone A represents POC meter BHB concentrations within 20% of the LBA concentrations as well as when the POC BHB concentration was < 0.8 mmol/L and the corresponding LBA BHB concentration was also < 0.8 mmol/L. Zone B contains POC meter values that were > 20% of LBA values but would result in a benign treatment decision. Zone C contained POC meter BHB concentrations that would result in not treating a sheep when the LBA BHB cocentration was above the acceptable range, and zone D contained POC concentrations that would result in treating a sheep when the LBA BHB concentration was within the acceptable range. B—Bland-Altman difference plots were prepared, accounting for multiple observations per test subject. The dashed lines indicate the 95% limits of agreement (± 1.96 SD of the mean difference), and the solid line indicates the mean difference between methods.

Citation: American Journal of Veterinary Research 74, 8; 10.2460/ajvr.74.8.1059

Figure 2—
Figure 2—

Error grid analysis (A) and Bland-Altman plot (B) comparing blood glucose concentrations measured with a dual-purpose POC meter with serum glucose concentrations measured with an LBA in samples from pregnant mixed-breed ewes with different feeding regimens. A—The error grid comprises 4 zones. Zone A represents POC meter glucose concentrations within 20% of the LBA concentrations as well as when the POC concentration was ≤ 50 mg/dL when the LBA concentration was also ≤ 50 mg/dL. Zone B contains POC meter concentrations that were > 20% of LBA concentrations but would result in a benign treatment decision or no treatment decision. Zone C contained POC meter glucose concentrations that would result in treating a sheep when the LBA glucose concentration was within the acceptable range, and zone D contained POC glucose concentrations that would result in not treating a sheep when the LBA was lower than the acceptable range. See Figure 1 for remainder of key.

Citation: American Journal of Veterinary Research 74, 8; 10.2460/ajvr.74.8.1059

Modified error grid plots developed to hypothetically evaluate results of clinical decision making based on use of the POC meter versus the LBA results showed that 97% of BHB concentrations were within zones A (93%) and B (4%), 2% were within zone C, and 1% was within zone D (Figure 1). For glucose concentrations, 72% were within zones A (68%) and B (4%), 27% were within zone C, and 1% was within zone D (Figure 2).

Discussion

The present study was conducted to evaluate the performance of a dual-purpose POC meter for measurement of blood BHB and glucose concentrations in 3 groups of sheep during gestation and lactation. The maternal malnutrition study from which they originated provided an opportunity to evaluate the POC meter in various environmental conditions and ewe nutritional states, which would be difficult to do in a nonexperimental setting because most producers would intervene to reduce the effects observed in these ewes.

The POC meter was more appropriate for measuring blood BHB concentrations than blood glucose concentrations in sheep. The results were poorer than those in a previously reported study,8 but the differences between studies may have been attributable to the larger sample size in the present study, ewe metabolic effects from malnutrition during gestation, environmental conditions, differences in sample handling, statistical techniques, or LBAs used. Another explanation is potential variations between the POC meters used, given that they are marketed with different names depending on the country of origin.8 The LBA and sample handling details were not described for the earlier study,8 and blood samples in that study were obtained at only 1 point during gestation or lactation. In addition, the use of the Bland-Altman plots and error grid plots in the study provided more clinically relevant results regarding meter performance, compared with the other study.12–14

The precision of the POC meter was compared with the manufacturer's reported precision because we were unable to identify any reports of studies in which the meter had been evaluated. The CV was similar to the manufacturer's reported results.10,11 Glucose meter performance criteria have been established by several testing and health organizations and range from 5% to 20%.15 Similar criteria are not available for POC meter performance when measuring blood BHB concentration. When used as a POC glucometer, the results for the device did not meet the criteria established by the testing organizations. This was not surprising because the meter is manufactured for measurement of analytes in human blood samples. The BHB performance results were best when compared with the LBA results in a more clinically relevant range (> 0.8 mmol/L) rather than across the entire range. Some of the variation at the lower ranges is likely attributable to reporting differences, considering that the LBA results were provided to 2 decimal places, whereas the POC meter results were provided to only 1 decimal place.

The Bland-Altman difference plots also showed that the POC meter compared more favorably with the LBA when used to measure BHB concentration rather than glucose concentration. The systematic bias evident with both meters used could be addressed through application of a correction factor to the results or establishment of meter-specific reference limits. However, the significant proportional bias in the blood glucose results would not be correctable. The increase in scatter at higher BHB and glucose concentrations was likely due to the smaller number of blood samples in these ranges.

Modified error grid plots revealed that the POC meter was clinically acceptable for use according to our criteria for measuring blood BHB concentrations in sheep, whereas the measurement function for glucose concentrations was not acceptable. No differences were observed in values between nutritional status groups for either analyte. The plots were useful in identifying risk for over- or undertreatment or inappropriate treatment decisions. Use of the POC meter would result in a large number of decisions resulting in overtreatment when used to measure blood glucose concentrations. The risk of overtreatment would not likely increase illness severity but would increase the cost to the producer because low POC meter glucose concentrations would most commonly lead to oral or parenteral administration of glucose substrates to increase blood glucose concentration. Diagnosing hyperglycemia rather than normoglycemia would not likely result in different intervention strategies because it would be difficult to identify the cause of an isolated high glucose concentration. An increase in blood glucose concentration can result from stress induced by capture and blood collection, actual meter characteristics, medical disorders, or typical intraindividual variation.

The POC meter was robust and easy to use in various conditions; however, the low end of the operating range of 10°C would prevent its use by producers and veterinarians working with pregnant ewes in colder climates. We found that keeping the meter and test strips in a small cooler with a chemical hand warmer maintained the proper operating temperature. Because of the large number of samples tested at each time point, we used 2 identical POC meters to reduce the need to recalibrate between BHB and glucose testing. One POC meter would be adequate to measure both analytes in most farm or veterinary settings.

The choice to evaluate the meters in various conditions that many sheep producers would experience resulted in many potential sources of variation in the present study. Operator variables included differences in sample collection and handling technique and risk of test strip contamination from dirt, lanolin, blood, or other factors. Despite the use of warming containers for the meters and test strips, the temperature and environmental variations potentially could have affected meter and strip performance. Multiple lots of test strips were required, adding another potential source of variation in results. To reduce this variation, the meters were calibrated prior to every collection day and before new boxes of test strips were used. In addition, the expiration date of the test strips was heeded and extra test strips were stored in a climate-controlled environment in their original packaging.

The ewes were not accustomed to being handled nor confined, contributing to the difficulty in obtaining blood samples and increasing the risk of hemolysis or blood clots forming in samples prior to testing. Hemolysis is easy to produce in sheep blood samples and is known to increase serum glucose concentrations.16 The study blood samples had various degrees of hemolysis, which may have affected POC meter performance; however, the effect of hemolysis on the results was not evaluated.

Other factors such as anemia, hypertriglyceridemia, hyperproteinemia, handling technique, ambient temperature, altitude, and pregnancy-induced blood volume expansion could have also affected blood glucose results obtained with the POC meter.17,18 The effects of these factors on blood glucose concentration have not been evaluated in combination with gestational stage and fetal numbers in sheep. Hematocrit was not measured; however, an unreported preliminary nutritional energy restriction study revealed no significant effect on Hct in the sheep. All the ewes used in the study were visibly examined for signs of anemia (pale mucous membranes and sclera) during sample collection, and no gross evidence of anemia was observed; however, the presence of mild anemia may have been overlooked. Some serum samples had evidence of mild to moderate lipemia, but there was no apparent relationship with nutritional status or gestational stage because concentrations would fluctuate among blood collection points and sheep. Similar factors have not been reported as affecting blood BHB concentration, and the manufacturer reports that Hct, various medications, and high triglycerides concentration do not affect BHB results.10

The POC meter used in the present study was designed for use in humans, in which blood glucose is known to be distributed evenly between the RBCs and plasma; however, in sheep, 83% of the glucose is distributed in the plasma and the remainder in the RBCs.19,20 Although glucose utilization by the RBCs continues until the serum is separated from the cells,21 the low serum glucose concentrations observed in the study sheep were surprising. Blood samples were kept on ice to limit cell glucose utilization until the samples could be centrifuged, and serum was removed from the blood clot within 3 hours after collection. We identified a significant difference in LBA-measured glucose concentrations between the control sheep fed 100% of their nutrient requirements and those fed a restricted amount (data not shown); however, serum glucose concentrations in the control group were still quite lower than the lower reference limit of our laboratory (70 mg/dL). Given the study findings, the laboratory's reference limits for adult sheep may not be appropriate for pregnant ewes. Our methods for sample collection and storage after serum separation should not have dramatically altered the blood glucose and BHB concentrations.16

The 3 groups of sheep that differed in nutritional status were used to determine whether feed restriction would sufficiently alter ewe metabolism to affect blood glucose and BHB concentrations. No evidence of such an alteration was observed; however, the feed-restricted sheep had BHB and glucose results that would typically require treatment intervention to prevent development of pregnancy toxemia.

Overall, the dual-purpose POC meter appeared to be adequate for measurement of blood BHB concentration in sheep and useful in the diagnosis of ketosis and pregnancy toxemia. Its use could lead to earlier disease detection and intervention as well as lower treatment and management costs for producers. The meter was not as reliable for measurement of blood glucose concentration. Therefore, glucose results should be interpreted with caution to avoid inappropriate medical management of suspected hypoglycemic or hyperglycemic sheep.

ABBREVIATIONS

BHB

β-Hydroxybutyrate

CV

Coefficient of variation

dGA

Days of gestational age

LBA

Laboratory biochemical analyzer

POC

Point of care

a.

Precision Xtra Blood Glucose & Ketone Monitoring System, Abbott Diabetes Care, Abbott Park, Ill.

b.

Grabber Hand Warmers, YSC Inc, Grand Rapids, Mich.

c.

Serum Vacutainer glass blood collection tubes, Becton, Dickinson and Co, Franklin Lakes, NJ.

d.

Roche Hitachi 917 Blood Chemistry Analyzer, Hitachi Ltd, Tokyo, Japan.

e.

MedCalc, version 12.3, MedCalc Software, Mariakerke, Belgium.

References

  • 1. Brozos C, Mavrogianni VS, Fthenakis GC. Treatment and control of peri-parturient metabolic diseases: pregnancy toxemia, hypocalcemia, hypomagnesemia. Vet Clin North Am Food Anim Pract 2011; 27:105113.

    • Search Google Scholar
    • Export Citation
  • 2. Bulgin M. Diseases of the periparturient ewe. In: Youngquist RS, Threlfall WR, eds. Current therapy in large animal theriogenology. 2nd ed. St Louis: WB Saunders Co, 2007; 695700.

    • Search Google Scholar
    • Export Citation
  • 3. Federici MO, Benedetti MM. Ketone bodies monitoring. Diabetes Res Clin Pract 2006; 74 (suppl 2): S77S81.

  • 4. Laffel L. Ketone bodies: a review of physiology, pathophysiology and application of monitoring to diabetes. Diabetes Metab Res Rev 1999; 15:412426.

    • Search Google Scholar
    • Export Citation
  • 5. Oetzel GR. Monitoring and testing dairy herds for metabolic disease. Vet Clin North Am Food Anim Pract 2004; 20:651674.

  • 6. Jones M, Miesner MD, Baird AN, et al. Diseases of the urinary system. In: Pugh DG, Baird AN, eds. Sheep and goat medicine. 2nd ed. St Louis: WB Saunders Co, 2012; 325360.

    • Search Google Scholar
    • Export Citation
  • 7. Iwersen M, Falkenberg U, Voigtsberger R, et al. Evaluation of an electronic cowside test to detect subclinical ketosis in dairy cows. J Dairy Sci 2009; 92:26182624.

    • Search Google Scholar
    • Export Citation
  • 8. Panousis N, Brozos C, Karagiannis I, et al. Evaluation of Precision Xceed meter for on-site monitoring of blood [beta]-hydroxybutyric acid and glucose concentrations in dairy sheep. Res Vet Sci 2012; 93:435439.

    • Search Google Scholar
    • Export Citation
  • 9. Voyvoda H, Erdogan H. Use of a hand-held meter for detecting subclinical ketosis in dairy cows. Res Vet Sci 2010; 89:344351.

  • 10. Precision Xtra blood ñ-ketone test strips [package insert]. Alameda, Calif: Abbott Diabetes Care Inc, 2006.

  • 11. Precision Xtra blood glucose test strips [package insert]. Alameda, Calif: Abbott Diabetes Care Inc, 2009.

  • 12. Bland J, Altman D. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat 2007; 17:571582.

    • Search Google Scholar
    • Export Citation
  • 13. Clarke W, Cox D, Gonder-Frederick L, et al. Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care 1987; 10:622628.

    • Search Google Scholar
    • Export Citation
  • 14. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 327:307310.

    • Search Google Scholar
    • Export Citation
  • 15. Tonyushkina K, Nichols JH. Glucose meters: a review of technical challenges to obtaining accurate results. J Diabetes Sci Technol 2009; 3:971980.

    • Search Google Scholar
    • Export Citation
  • 16. Morris JD, Fernandez JM, Chapa AM, et al. Effects of sample handling, processing, storage, and hemolysis on measurements of key energy metabolites in ovine blood. Small Ruminant Res 2002; 43:157166.

    • Search Google Scholar
    • Export Citation
  • 17. Ginsburg BH. Factors affecting blood glucose monitoring: sources of errors in measurement. J Diabetes Sci Technol 2009; 3:903913.

    • Search Google Scholar
    • Export Citation
  • 18. Hytten F. Blood volume changes in normal pregnancy. Clin Haematol 1985; 14:601612.

  • 19. Coldman MF, Good W. The distribution of sodium, potassium and glucose in the blood of some mammals. Comp Biochem Physiol 1967; 21:201206.

    • Search Google Scholar
    • Export Citation
  • 20. MacKay E. The distribution of glucose in human blood. J Biol Chem 1932; 97:685689.

  • 21. Zhang DJ, Elswick RK, Miller WG, et al. Effect of serum-clot contact time on clinical chemistry laboratory results. Clin Chem 1998; 44:13251333.

    • Search Google Scholar
    • Export Citation

Contributor Notes

Supported with start-up funds provided to Dr. Byers.

Presented in abstract form at the American College of Veterinary Internal Medicine Forum, New Orleans, June 2012.

Address correspondence to Dr. Byers (stacey.byers@colostate.edu).