• View in gallery

    Mean ± SD glucose concentrations on days 1, 3, 4, and 5 measured via point-of-care (POC) glucometer, 2 glucose-monitoring systems (ie, Dexcom and Libre), and standard laboratory enzymatic chemistry methods (CHEM) in 8 healthy horses. *Significant difference (P < 0.05) for Dexcom vs CHEM values.

  • View in gallery

    Panels A through E are Bland-Altman plots describing the degree of agreement between 2 glucose-measuring techniques on days 1 through 5 in 8 healthy horses. The solid line shows the mean difference, whereas the upper dashed line represents the upper limit of agreement (difference + 1.96 X SD) and the lower dashed line represents the lower limit of agreement (difference + 1.96 X SD).

  • View in gallery

    Mean ± SD glucose concentrations in dextrose-induced hyperglycemia on day 2 measured via POC glucometer, 2 glucose-monitoring systems (ie, Dexcom and Libre), and CHEM in 8 healthy horses. **P < 0.01 compared with time 0. #P < 0.05 for POC vs CHEM.

  • View in gallery

    Panels A through E are Bland-Altman plots describing the degree of agreement between 2 glucose-measuring techniques on day 2 during dextrose-induced hyperglycemia in 8 healthy horses. The solid line shows the mean difference, whereas the upper dashed line represents the upper limit of agreement (difference + 1.96 X SD) and the lower dashed line represents the lower limit of agreement (difference + 1.96 X SD).

  • 1.

    McFarlane D. Diagnostic testing for equine endocrine diseases: confirmation versus confusion. Vet Clin North Am Equine Pract. 2019;35(2):327338. doi:10.1016/j.cveq.2019.03.005

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2.

    Carr EA. Enteral/parenteral nutrition in foals and adult horses practical guidelines for the practitioner. Vet Clin North Am Equine Pract. 2018;34(1):169180. doi:10.1016/j.cveq.2017.11.012

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3.

    Boshuizen B, Ploeg M, Dewulf J, et al. Inflammatory bowel disease (IBD) in horses: a retrospective study exploring the value of different diagnostic approaches. BMC Vet Res. 2018;14(1):21. doi:10.1186/s12917-018-1343-1

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4.

    Lee TF, Drake SM, Roberts GW, et al. Relative hyperglycemia is an independent determinant of in-hospital mortality in patients with critical illness. Crit Care Med. 2020;48(2):e115e122. doi:10.1097/CCM.0000000000004133

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5.

    Whitcomb BW, Pradhan EK, Pittas AG, Roghmann MC, Perencevich EN. Impact of admission hyperglycemia on hospital mortality in various intensive care unit populations. Crit Care Med. 2005;33(12):27722777. doi:10.1097/01.ccm.0000189741.44071.25

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6.

    Krinsley JS, Chase JG, Gunst J, et al. Continuous glucose monitoring in the ICU: clinical considerations and consensus. Crit Care. 2017;21(1):197. doi:10.1186/s13054-017-1784-0

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7.

    Edelman SV, Argento NB, Pettus J, Hirsch IB. Clinical implications of real-time and intermittently scanned continuous glucose monitoring. Diabetes Care. 2018;41(11):22652274. doi:10.2337/dc18-1150

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8.

    Ajjan R, Slattery D, Wright E. Continuous glucose monitoring: a brief review for primary care practitioners. Adv Ther. 2019;36(3):579596. doi:10.1007/s12325-019-0870-x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9.

    Aleppo G, Webb K. Continuous glucose monitoring integration in clinical practice: a stepped guide to data review and interpretation. J Diabetes Sci Technol. 2019;13(4):664673. doi:10.1177/1932296818813581

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10.

    Ida S, Kaneko R, Murata K. Utility of real-time and retrospective continuous glucose monitoring in patients with type 2 diabetes mellitus: a meta-analysis of randomized controlled trials. J Diabetes Res. 2019;2019:4684815. doi:10.1155/2019/4684815

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11.

    Corradini S, Pilosio B, Dondi F, et al. Accuracy of a flash glucose monitoring system in diabetic dogs. J Vet Intern Med. 2016;30(4):983988. doi:10.1111/jvim.14355

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12.

    Wiedmeyer CE, Johnson PJ, Cohn LA, Meadows RL. Evaluation of a continuous glucose monitoring system for use in dogs, cats, and horses. J Am Vet Med Assoc. 2003;223(7):987992. doi:10.2460/javma.2003.223.987

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13.

    Hug SA, Riond B, Schwarzwald CC. Evaluation of a continuous glucose monitoring system compared with an in-house standard laboratory assay and a handheld point-of-care glucometer in critically ill neonatal foals. J Vet Emerg Crit Care (San Antonio). 2013;23(4):408415. doi:10.1111/vec.12072

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14.

    Johnson PJ, Wiedmeyer CE, LaCarrubba A, et al. Clinical assessment of blood glucose homeostasis in horses: comparison of a continuous glucose monitoring system with a combined intravenous glucose and insulin test protocol. J Vet Intern Med. 2011;25(1):162165. doi:10.1111/j.1939-1676.2010.0643.x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15.

    Cunneen A, Wood KA, Mathison K, Herndon AM, Bertin FR. Comparison of a continuous indwelling glucometer with a point-of-care device in healthy adult horses. Vet Rec. 2020;187(3):e21. doi:10.1136/vr.105607

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16.

    Roberts MC, Hill FW. The oral glucose tolerance test in the horse. Equine Vet J. 1973;5(4):171173. doi:10.1111/j.2042-3306.1973.tb03219.x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17.

    Altman DG. Practical Statistics for Medical Research. CRC press; 1990.

  • 18.

    Hackett ES, McCue PM. Evaluation of a veterinary glucometer for use in horses. J Vet Intern Med. 2010;24(3):617621. doi:10.1111/j.1939-1676.2010.0481.x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19.

    Rendle DI, Armstrong SK, Heller J, Hughes KJ. Precision and accuracy of a point-of-care glucometer in horses and the effects of sample type. Vet J. 2019;252:105359. doi:10.1016/j.tvjl.2019.105359

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20.

    Collicutt NB, Garner B, Berghaus RD, Camus MS, Hart K. Effect of delayed serum separation and storage temperature on serum glucose concentration in horse, dog, alpaca, and sturgeon. Vet Clin Pathol. 2015;44(1):120127. doi:10.1111/vcp.12217

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 21.

    Ferrante PL, Kronfeld DS. Effect of sample handling on measurement of plasma glucose and blood lactate concentrations in horses before and after exercise. Am J Vet Res. 1994;55(11):14971500.

    • Search Google Scholar
    • Export Citation

Advertisement

Comparison of two glucose-monitoring systems for use in horses

View More View Less
  • 1 Virginia-Maryland College of Veterinary Medicine, Virginia Polytechnic Institute and State University, Blacksburg, VA
  • | 2 College of Veterinary Medicine, Iowa State University, Ames, IA
  • | 3 College of Veterinary Medicine, North Carolina State University, Raleigh, NC

Abstract

OBJECTIVE

To determine the accuracy of 2 interstitial glucose-monitoring systems (GMSs) for use in horses compared with a point-of-care (POC) glucometer and standard laboratory enzymatic chemistry method (CHEM).

ANIMALS

8 clinically normal adult horses.

PROCEDURES

One of each GMS device (Dexcom G6 and Freestyle Libre 14-day) was placed on each horse, and blood glucose concentration was measured via POC and CHEM at 33 time points and compared with simultaneous GMS readings. An oral glucose absorption test (OGAT) was performed on day 2, and glucose concentrations were measured and compared.

RESULTS

Glucose concentrations were significantly correlated with one another between all devices on days 1 to 5. Acceptable agreement was observed between Dexcom G6 and Freestyle Libre 14-day when compared with CHEM on days 1, 3, 4, and 5 with a combined mean bias of 10.45 mg/dL and 1.53 mg/dL, respectively. During dextrose-induced hyperglycemia on day 2, mean bias values for Dexcom G6 (10.49 mg/dL) and FreeStyle Libre 14-day (0.34 mg/dL) showed good agreement with CHEM.

CLINICAL RELEVANCE

Serial blood glucose measurements are used to diagnose or monitor a variety of conditions in equine medicine; advances in near-continuous interstitial glucose monitoring allow for minimally invasive glucose assessment, thereby reducing stress and discomfort to patients. Data from this study support the use of the Dexcom G6 and Freestyle Libre 14-day interstitial glucose-monitoring systems to estimate blood glucose concentrations in horses.

Abstract

OBJECTIVE

To determine the accuracy of 2 interstitial glucose-monitoring systems (GMSs) for use in horses compared with a point-of-care (POC) glucometer and standard laboratory enzymatic chemistry method (CHEM).

ANIMALS

8 clinically normal adult horses.

PROCEDURES

One of each GMS device (Dexcom G6 and Freestyle Libre 14-day) was placed on each horse, and blood glucose concentration was measured via POC and CHEM at 33 time points and compared with simultaneous GMS readings. An oral glucose absorption test (OGAT) was performed on day 2, and glucose concentrations were measured and compared.

RESULTS

Glucose concentrations were significantly correlated with one another between all devices on days 1 to 5. Acceptable agreement was observed between Dexcom G6 and Freestyle Libre 14-day when compared with CHEM on days 1, 3, 4, and 5 with a combined mean bias of 10.45 mg/dL and 1.53 mg/dL, respectively. During dextrose-induced hyperglycemia on day 2, mean bias values for Dexcom G6 (10.49 mg/dL) and FreeStyle Libre 14-day (0.34 mg/dL) showed good agreement with CHEM.

CLINICAL RELEVANCE

Serial blood glucose measurements are used to diagnose or monitor a variety of conditions in equine medicine; advances in near-continuous interstitial glucose monitoring allow for minimally invasive glucose assessment, thereby reducing stress and discomfort to patients. Data from this study support the use of the Dexcom G6 and Freestyle Libre 14-day interstitial glucose-monitoring systems to estimate blood glucose concentrations in horses.

Introduction

Evaluation of blood glucose is a vital component of the diagnostic evaluation of various disease processes, such as inflammatory and infiltrative bowel diseases1 and insulin dysregulation associated with equine metabolic syndrome and pituitary pars intermedia dysfunction2 in horses. Serial monitoring of blood glucose concentration in horses receiving carbohydrate-containing fluids IV (eg, those administered to anorectic or hyperlipemic horses) to avoid hyperglycemia3 is commonly performed in hospitalized equine patients. Various studies4,5 in critically ill people suggest that prolonged hyperglycemia is associated with higher mortality rates. Thus, glucose monitoring is a common and important measured variable for both clinical and research purposes in equine medicine.

Glucose-monitoring systems (GMSs) are widely used in people, primarily to monitor people with type 1 diabetes mellitus but also in the critical care setting because of agreement of their measurements with traditional methods of measuring blood glucose concentration, ease of use, and ability to provide continuous or near-continuous glucose data.610 Two types of systems are available: continuous GMSs (CGMS) and flash GMSs (FGMS). Both systems report interstitial glucose concentrations every 5 minutes, but the continuous system automatically uploads glucose concentration measurements to a reader, whereas flash glucose systems require direct scanning of the sensor with a reader at periodic intervals. Data may then be uploaded to online software to view daily and multiday graphic representation of glucose measurements.

Instead of utilizing finger pricks or venipuncture for obtaining blood samples, these systems measure glucose concentrations within the interstitial space of the subcutaneous tissue. The interstitial glucose reacts with glucose oxidase in a semipermeable membrane in the sensor, which converts glucose into gluconic acid and hydrogen peroxide. This reaction generates an electric signal proportional to glucose concentration and is translated into a milligram-per-deciliter value. Multiple studies have indicated that measurements of glucose concentrations within the interstitial space are comparable to measurements of whole blood glucose concentrations in both humans and dogs.7,11 Studies in horses have been limited, but an early study12 determined that interstitial glucose concentrations measured by CGMS correlated well with whole blood glucose concentrations, were sensitive to abrupt changes in glucose concentration, and provided a detailed and accurate representation of an animal’s glycemic status over an extended time period in dogs, cats, and horses. A more recent study11 evaluated an FGMS in diabetic dogs and noted that it was a valid alternative for glucose monitoring when compared with traditional methods. However, other studies13,14 have raised concerns regarding the accuracy of these devices, necessitating the need for further investigation for clinical use. Additionally, although these devices are marketed for use for up to 10 to 14 days, no studies have evaluated their accuracy in adult horses past 31 hours (1.29 days) of use.1215 Advancements in CGMS technology have produced more readily available and affordable devices, making their use a potentially viable tool for clinical application.

The objectives of this study were 1) to determine the accuracy of 2 GMSs in horses compared with a point-of-care (POC) glucometer and a standard laboratory enzymatic chemistry method (CHEM) and 2) to determine the accuracy of the devices during dextrose-induced hyperglycemia using an oral glucose absorption test (OGAT). We hypothesized that GMS would provide acceptable agreement with POC and CHEM and provide diagnostically useful OGAT glucose curves.

Materials and Methods

Eight privately owned animals were enrolled, following informed owner consent, with a mean age of 11 years (range, 2 to 21 years) and a mean weight of 474 kg (range, 373 kg to 521 kg). Four geldings, 3 mares, and 1 stallion were used, and breeds included American Quarter Horse (n = 3), Paint (2), Appaloosa (1), Trakehner (1), and Arabian (1). All horses were determined to be healthy based on physical examinations, CBCs, and serum biochemistry profiles within acceptable reference intervals prior to initiation of the study. This study was approved by the university institutional animal care and use committee.

Horses were housed at least 12 hours prior to commencement of the study to allow for acclimatation to a new environment and were provided free choice hay and water throughout the study period, except for a 12-hour fast prior to performance of the OGAT. Physical examinations were performed every 12 hours over the 5-day study period. A jugular catheter was aseptically placed in each horse, and catheter patency was maintained by flushing with 6 mL of heparinized saline (0.9% NaCl) solution every 6 hours.

Two 3 X 3-inch areas were clipped using a No. 40 clipper blade over the lateral aspects of the neck and hindquarters, lateral to the tail head. The clipped areas were cleaned with isopropyl alcohol and allowed to dry prior to device application. Each device was applied according to its manufacturer’s instructions: the FGMS (FreeStyle Libre 14-day; Abbott) was placed on the neck, and the CGMS (Dexcom G6; DexCom Inc) was placed on the hindquarters. Adherence of the sensor pad to the skin was reinforced with cyanoacrylate adhesive placed on the periphery of the sensor pad. The devices were calibrated according to manufacturer’s recommendations, which included a 2-hour calibration period for the CGMS (ie, Dexcom) and a 1-hour period for the FGMS (ie, Libre), starting on day 1. The Dexcom system was recalibrated as prompted by the device using the value from the handheld glucometer, whereas the Libre operated on a factory calibration system.

Following calibration on day 1, a 5-mL blood sample was collected in sodium fluoride tubes every hour for a total of 5 samples and then every 3 hours over the remainder of the 24-hour period. Readings from the Dexcom and Libre were recorded, just prior to blood collection. Blood glucose concentration was immediately measured from each sample using a POC handheld glucometer previously validated for horses (AlphaTRAK 2; Zoetis). The remainder of the sample was submitted for standard laboratory assay as the gold standard using the hexokinase method (CHEM; AU480 Chemistry Analyzer; Beckman-Coulter). Blood samples acquired between 8 PM and 7 AM were collected in similar fashion but stored at 4 °C for blood glucose measurement via standard laboratory assay the following day. To evaluate both GMSs in the hyperglycemic range, horses were fasted overnight, starting at 7 PM on day 1, to allow for an OGAT the following day.

On day 2, baseline glucose measurements were recorded from glucose-monitoring devices, and blood samples were drawn and measured via POC and CHEM. Horses were sedated with xylazine (0.4 mg/kg, IV) via an IV catheter to facilitate nasogastric intubation associated with the OGAT. Subsequently, the OGAT was performed using a previously described protocol with a 20% dextrose solution, administered at a dose of 1 g/kg by nasogastric intubation.16 Blood samples (5 mL) were obtained every 30 minutes following dextrose administration for a total of 4 hours. Blood glucose concentration was measured immediately after collection via a POC glucometer, with the reminder of the sample submitted for CHEM. Dexcom and Libre measurements were recorded immediately prior to the collection of each blood sample.

For the remainder of the study period (days 3 to 5), blood samples were obtained every 4 hours for a total of 6 samples/d for blood glucose measurement. Corresponding Dexcom and Libre measurements were recorded immediately prior to sampling. Blood glucose concentration was measured immediately after collection via the POC glucometer with the remainder of the sample submitted for CHEM. Samples collected between 8 PM and 7 AM were stored at 4 °C overnight and submitted for CHEM the following morning. Following the final sample collection, the glucose-monitoring devices and IV catheters were removed.

Statistical analysis

Data were tested for normality by a Shapiro-Wilk test and were noted to be normally distributed. Data were presented as mean ± SD. To compare measured glucose concentrations between the CHEM (considered the reference standard), POC glucometer, and GMS (Dexcom and Libre) methods, the following paired comparisons were made: Dexcom-CHEM, Dexcom-POC, Libre-CHEM, Libre-POC, Dexcom-Libre, and POC-CHEM in both phases of the study.

Glucose concentrations on days 1 through 5 were compared among analyzers using the Pearson linear correlation. Agreement between glucose concentrations for each method of glucose measurement was determined using the Bland and Altman and Lin concordance analyses. The bias was calculated as the mean difference between the Dexcom-CHEM, Dexcom-POC, Libre-CHEM, Libre-POC, Dexcom-Libre, and POC-CHEM. A positive bias reflected overestimation of glucose concentration as compared with CHEM. Likewise, when comparing the Dexcom and Libre measurements with POC measurements, a positive bias reflected overestimation of the POC measurements as compared with CGMS analysis. The limits of agreement were reported as bias ± (1.96 X SD of the bias).

The mixed model for 2-factor repeated-measures ANOVA was performed between glucose concentrations obtained by the Dexcom device, Libre device, POC glucometer, and CHEM to assess the effect of time or assay on glucose concentrations. When multiple comparisons between time points were performed, a Bonferroni correction was used to determine differences in glucose concentrations. Commercial statistics software programs such as SPSS Statistics version 24 (IBM Corp), Graph Pad Prism version 8 (GraphPad Software), and StatsToDo (statstodo.com/Agreement_Pgm.php) were used. Significance was set at P ≤ 0.05.

An additional method used to compare the CGMS and POC measurements with CHEM was the number of measurements that were observed within 15% of CHEM values.

Results

Glucose concentration on days 1, 3, 4, and 5

For glucose concentrations, blood samples were collected at time 0 and 1, 2, 3, 4, 7, 10, 13, 16, 19, and 24 hours on day 1; at time 0 and 30, 60, 90, 120, 150, and 180 minutes after glucose administration on day 2; at time 0 and 4, 8, 12, 16, and 20 hours on days 3 and 4; and time 0 and 4 and 8 hours on day 5. There was an overall significant effect of time and method on glucose concentration on days 1, 3, 4, and 5 (P < 0.05; Figure 1). Glucose concentration was significantly higher when measured by the Dexcom device compared with CHEM analysis at time points 0 on day 1 and 12 hours on day 3. No other differences in glucose concentrations between time points and devices reached the Bonferroni-adjusted P value. The measurements from Dexcom, Libre, and POC were compared with CHEM, and the percent of total values within 15% of CHEM values were calculated, resulting in 65.5%, 68.6%, and 35%, respectively.

Figure 1
Figure 1

Mean ± SD glucose concentrations on days 1, 3, 4, and 5 measured via point-of-care (POC) glucometer, 2 glucose-monitoring systems (ie, Dexcom and Libre), and standard laboratory enzymatic chemistry methods (CHEM) in 8 healthy horses. *Significant difference (P < 0.05) for Dexcom vs CHEM values.

Citation: American Journal of Veterinary Research 83, 3; 10.2460/ajvr.21.05.0068

Glucose concentrations were significantly correlated with one another (Table 1) between all devices on days 1 through 5 (ie, Dexcom-CHEM, Dexcom-POC, Libre-CHEM, Libre-POC, Dexcom-Libre, and POC-CHEM). Acceptable agreement was observed between Dexcom and Libre when compared with CHEM with a combined mean bias of 10.45 and 1.53 mg/dL, respectively. The mean bias (95% limits of agreement) among the other devices on days 1 through 5 was as follows: Dexcom-POC, 7.66 mg/dL (–27.7 to 43.05); Libre-POC, 16.48 mg/dL (–26.2 to 60); Dexcom-Libre, 8.9 mg/dL (–34.41 to 52.3); and POC-CHEM, 18.13 mg/dL (–10.5 to 46.8; Figure 2).

Table 1

Bland-Altman and Lin coefficient analysis of glucose concentration (mg/dL) comparisons between point-of-care (POC) glucometer, 2 glucose-monitoring systems (ie, Dexcom and Libre), and standard laboratory enzymatic chemistry methods (CHEM) in 8 healthy horses.

VariableDexcom/CHEMDexcom/POCLibre/CHEMLibre/POCDexcom/LibrePOC/CHEM
Bland-Altman analysis
Bias (glucose, mg/dL)10.457.661.5316.488.918.13
95% limits of agreement–22.68 to 43.58–27.7 to 43.05–34.91 to 38–26.2 to 60–34.41 to 52.3–10.5 to 46.8
Lin coefficient (95% CI)0.80.850.820.740.750.79
(0.76 to 0.84)(0.81 to 0.88)(0.78 to 0.85)(0.68 to 0.78)(0.69 to 0.8)(0.75 to 0.82)
Pearson correlation0.860.880.820.820.77 0.94
coefficient (r; 95% CI)(0.82 to 0.88)(0.85 to 0.9)(0.78 to 0.86)(0.78 to 0.86)(0.72 to 0.82)(0.92 to 0.95)
Figure 2
Figure 2

Panels A through E are Bland-Altman plots describing the degree of agreement between 2 glucose-measuring techniques on days 1 through 5 in 8 healthy horses. The solid line shows the mean difference, whereas the upper dashed line represents the upper limit of agreement (difference + 1.96 X SD) and the lower dashed line represents the lower limit of agreement (difference + 1.96 X SD).

Citation: American Journal of Veterinary Research 83, 3; 10.2460/ajvr.21.05.0068

Lin concordance correlation coefficient tests how well bivariate pairs of observations conform relative to a gold standard. This test measures both precision and accuracy. The Lin coefficient between Dexcom-CHEM, Dexcom-POC, Libre-CHEM, Libre-POC, Dexcom-Libre, and POC-CHEM on days 1 through 5 demonstrated strong agreement according to Altman et al (Table 1).17

Dextrose-induced hyperglycemia on day 2

During the OGAT, blood samples were collected at time 0 and 30, 60, 90, 120, 150, and 180 minutes after dextrose administration. There was a significant (P < 0.01) effect of dextrose administration on glucose concentration from 30 to 150 minutes for all 4 methods of glucose measurement (Figure 3). Glucose concentration was higher when measured by POC compared with CHEM at time 0 and 30 minutes (P < 0.05). There were no other differences noted in glucose concentrations between time points and devices (P > 0.05).

Figure 3
Figure 3

Mean ± SD glucose concentrations in dextrose-induced hyperglycemia on day 2 measured via POC glucometer, 2 glucose-monitoring systems (ie, Dexcom and Libre), and CHEM in 8 healthy horses. **P < 0.01 compared with time 0. #P < 0.05 for POC vs CHEM.

Citation: American Journal of Veterinary Research 83, 3; 10.2460/ajvr.21.05.0068

Glucose concentrations were significantly correlated with one another (Table 2) between all devices on day 2. During dextrose-induced hyperglycemia on day 2, mean bias values for Dexcom (10.49 mg/dL) and Libre (0.34 mg/dL) showed good agreement with CHEM. The mean bias (95% limits of agreement) among the other devices on day 2 was as follows: Dexcom-POC, 20.16 mg/dL (–12.8 to 53.16); Libre-POC, 30.33 mg/dL (–25.7 to 86.4); Dexcom-Libre, 10.16 mg/dL (–41.3 to 61.6); and POC-CHEM, 30.6 mg/dL (–10.6 to 71.9; Figure 4). The Lin coefficient between Dexcom-CHEM, Dexcom-POC, Libre-CHEM, Libre-POC, Dexcom-Libre, and POC-CHEM in on day 2 demonstrated moderate to strong agreement (Table 2).

Table 2

Bland-Altman and Lin coefficient analysis of glucose concentration (mg/dL) comparisons between POC glucometer, Dexcom, Libre, and CHEM in 8 healthy horses during dextrose-induced hyperglycemia on day 2.

VariableDexcom/CHEMDexcom/POCLibre/CHEMLibre/POCDexcom/LibrePOC/CHEM
Bland-Altman analysis
Bias (glucose, mg/dL)10.4920.160.3430.3310.1630.6
95% limits of agreement–23.5 to 44.47–12.8 to 53.16–49 to 49.7–25.7 to 86.4–41.3 to 61.6–10.6 to 71.9
Lin coefficient (95% CI)0.870.850.8210.680.80.71
(0.8 to 0.92)(0.78 to 0.9)(0.78 to 0.88)(0.55 to 0.78)(0.68 to 0.87)(0.61 to 0.79)
Pearson correlation0.910.940.830.820.820.93
coefficient (r; 95% CI)(0.86 to 0.95)(0.91 to 0.97)(0.72 to 0.89)(0.71 to 0.89)(0.7 to 0.88)(0.88 to 0.95)
Figure 4
Figure 4

Panels A through E are Bland-Altman plots describing the degree of agreement between 2 glucose-measuring techniques on day 2 during dextrose-induced hyperglycemia in 8 healthy horses. The solid line shows the mean difference, whereas the upper dashed line represents the upper limit of agreement (difference + 1.96 X SD) and the lower dashed line represents the lower limit of agreement (difference + 1.96 X SD).

Citation: American Journal of Veterinary Research 83, 3; 10.2460/ajvr.21.05.0068

Discussion

In this study, both the Dexcom and Libre showed good agreement and were significantly correlated with the CHEM with the mean bias of 10.45 and 1.53 mg/dL, respectively, indicating that the Dexcom and Libre overestimated blood glucose by 10.45 and 1.53 mg/dL, respectively. Despite the close correlation with CHEM, the 95% confidence interval for the Dexcom and Libre was –22.68 to 43.58 mg/dL and –34.91 to 38 mg/dL, indicating a notable range in which the results of the device measurements could fall compared with CHEM. In comparison, the POC glucometer, which is the most frequently used POC glucose-measuring device in clinical practice, had the largest mean bias (18.13 mg/dL) and wide 95% confidence interval (–10.5 to 46.8 mg/dL). Because 65.5% of Dexcom values and 66.8% of Libre values were within 15% of the CHEM values, compared with only 35.9% of POC values within 15% of CHEM, the authors suggest that either GMS is an acceptable clinical tool to help guide therapy.

The OGAT is a long-standing and simple method to evaluate small intestinal absorption and was used to induce hyperglycemia and determine how the GMS performed at high glucose concentrations.16 To safely administer the solution via nasogastric tube, all study horses were sedated with xylazine (0.4 mg/kg, IV). While the use of sedation may have induced hyperglycemia, it should not have affected the comparison of the values attained using different measurement methods on the same blood sample. Dextrose-induced hyperglycemia did not impact the accuracy of the devices used with the mean bias of Dexcom and Libre of 10.45 and 0.34 mg/dL, respectively. The mean bias for the POC device was the highest among the devices tested at 30.6 mg/dL. In addition, both GMSs provided more detailed OGAT glucose curves created by near-continuous (every 5 minutes) measurements of interstitial glucose during the OGAT period.

Conventional blood glucose monitoring only allows for spot glucose determinations at set intervals, producing a snapshot of glucose concentrations at 1 point in time. This limits the amount of information available to base treatment decisions and increases workload on hospital staff. Frequency of venipuncture is also associated with patient discomfort and stress, which can cause transient hyperglycemia, complicating the interpretation of a single glucose value. Utilizing interstitial GMS eliminates the need for venipuncture, lessens workload on staff, and provides dynamic information about glucose concentrations over time for clinician interpretation. Human subjects report little to no discomfort associated with GMS application, and the same was subjectively observed in the equine population in this study. Once applied, the Libre obtains glucose readings each minute while the Dexcom obtains readings every 5 minutes, resulting in 1,440 and 288 readings/d, respectively. Both devices have the ability to measure glucose readings within the range of 40 to 400 mg/dL. Slightly different technologies between real-time continuous devices (eg, Dexcom G6) and flash devices (FreeStyle Libre 14-day) affect the way each device reports measurements. The Dexcom G6 automatically uploads readings to the reader via Bluetooth technology every 5 minutes. The FreeStyle Libre also generates readings every 5 minutes, but the patient’s sensor must be scanned with a receiver or other compatible device to obtain readings at least every 8 hours due to limits of sensory memory. Both of these technologies allow for more comprehensive glucose data to be obtained with less patient disturbance and technical support.

Depending on clinical application, serial blood glucose monitoring by traditional methods can result in significant cost, but recent developments in GMS technology have increased their cost-effectiveness, making them a potential option for clinical use. Both devices have the option to purchase receivers that cost approximately $75 (FreeStyle Libre 14-day) to $365 (Dexcom G6). The use of a compatible smartphone may negate the need to purchase a receiver. The FreeStyle Libre 14-day requires the purchase of an approximately $57 sensor. A sensor and transmitter are required for Dexcom G6 use, costing approximately $117 and $250 respectively. While the sensors for either device are limited to 1-time use, the Dexcom G6 transmitter is reusable up to 90 days.

A recent study evaluated the use of the Libre in horses,15 comparing the Libre with the AlphaTRAK POC glucometer. Sensors were left in place for a maximum of 31 hours, whereas the study reported here evaluated sensors for approximately 108 hours. In the previously reported study, samples were collected every 5 minutes to evaluate a lag effect between change in blood glucose concentration and interstitial fluid glucose concentration under normal conditions, during insulin-induced hypoglycemia, and during dextrose-induced hyperglycemia. A lag effect was observed with significant differences detected in different phases of the study. A 60-minute lag time between glucose values generated by POC and Libre was observed with resting glucose values, while lag times of 10 and 20 minutes were observed with insulin-induced hypoglycemia and dextrose-induced hyperglycemia, respectively. In the previous study, the mean bias (95% limits of agreement) for dextrose-induced hyperglycemia was –0.03 (–2.46 to 2.52) mmol/L as compared with 1.69 (–1.43 to 4.80) mmol/L in the current study. The previous study suggested that because of conflicting bias between phases of hypoglycemia and hyperglycemia, use in practice may be limited due to potential delay of clinical decisions and detriment to the patient. Although the POC represents the standard use of glucose measurement in clinical settings, it is not the gold standard of measurement and likely has inaccuracies, producing some degree of bias, and therefore affecting assessment of accuracy. The current study also compared both the Dexcom and Libre to the gold-standard laboratory assay, the results of which showed smaller mean bias and limits of agreement.

There are several limitations to the study reported here. Assessment of glucose concentrations during a hypoglycemic state was not evaluated in this study. In a previously reported study,15 when hypoglycemia was induced by administration of 0.10 U/kg of regular insulin, the GMS (FreeStyle Libre) measured higher glucose concentrations than POC during periods of hyperglycemia. Another limitation of this study was that the sensors were not investigated for the marketed longevity of 10 or 14 days for the Dexcom and Libre, respectively. Sensors were left in place for a maximum of approximately 108 hours. A total of 10 Dexcom sensors and 13 Libre sensors were used for the 8 horses over the 5-day study period due to dislodgement or “sensor error” readings. Additionally, several variables have been shown to interfere with glucose measurement using POC glucometers, including oxygen pressure, hematocrit, pH, temperature, and medications such as mannitol, dopamine, and ascorbic acid.18 These variables were not measured in the current study because the horses used were apparently healthy. Further study is needed to investigate the effect of these variables on GMS performance in horses.

Veterinary medicine lacks a consensus regarding criteria for evaluation of glucometers for clinical use. Additionally, influences of sample type, glucose concentration, and delays in analysis on results in horses are largely unknown.19 In the current study, all blood samples submitted to the laboratory for CHEM were collected in sodium fluoride collection tubes. During business hours, these samples were immediately submitted to the laboratory for analysis. Any samples collected after hours were stored at 4 °C until submission. Fluoride inhibits enolase, an enzyme in the glycolytic pathway, and sodium fluoride tubes have historically been used to prevent glycolysis by erythrocytes. Although the use of sodium fluoride within collection tubes prevents glycolysis, studies have shown their use can artifactually decrease plasma glucose concentration through loss of intracellular water from RBCs, therefore diluting the plasma in the sample.20 Ferrante et al21 showed a 6% to 10% decrease in blood glucose concentration in samples obtained from horses collected in sodium fluoride tubes. Recently, Rendle et al19 evaluated the effect of sample type and storage on glucometer precision and accuracy, finding that blood stored in tubes containing EDTA and fluoride oxalate resulted in improved glucometer repeatability and agreement with laboratory standard, although the glucose concentration was underestimated. The effect of delayed serum separation and storage temperature was evaluated by Collicutt et al,20 who found that the storage of whole blood at 4 °C limits serum glucose concentration decline in equine blood samples for up to 8 hours of storage. A combination of the use of fluoride collection tubes and the length of time of sample storage due to unavailability of after-hours laboratory testing represents a limitation in this study.

GMSs have multiple potential clinical applications in equine medicine. Serial monitoring of glucose concentrations is necessary for the diagnosis of several equine diseases, such as small intestinal malabsorptive disorders or insulin dysregulation.1,3 Glucose concentrations are also measured in horses with altered glucose homeostasis during critical illness or during IV administration of parenteral nutrition.2 Some of these applications require sampling as often as every 30 minutes, which can cause discomfort to the patient and require added technical support. GMSs offer the ability to monitor glucose in these patients noninvasively while obtaining near-continuous glucose data. Reducing fluctuation and variability in glucose concentrations has been shown to decrease morbidity and mortality in human patients in intensive care units.4,5 The results of this study supported the use of these devices for application in clinical equine medicine.

Acknowledgments

Funded by the Haroldson Family Foundation. Funding sources did not have any involvement in the study design, data analysis and interpretation, or writing and publication of the manuscript.

The authors declare that there were no conflicts of interest regarding this study.

The authors thank the Haroldson Family Foundation and all the individuals who allowed their horses to be included in this study.

References

  • 1.

    McFarlane D. Diagnostic testing for equine endocrine diseases: confirmation versus confusion. Vet Clin North Am Equine Pract. 2019;35(2):327338. doi:10.1016/j.cveq.2019.03.005

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2.

    Carr EA. Enteral/parenteral nutrition in foals and adult horses practical guidelines for the practitioner. Vet Clin North Am Equine Pract. 2018;34(1):169180. doi:10.1016/j.cveq.2017.11.012

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3.

    Boshuizen B, Ploeg M, Dewulf J, et al. Inflammatory bowel disease (IBD) in horses: a retrospective study exploring the value of different diagnostic approaches. BMC Vet Res. 2018;14(1):21. doi:10.1186/s12917-018-1343-1

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4.

    Lee TF, Drake SM, Roberts GW, et al. Relative hyperglycemia is an independent determinant of in-hospital mortality in patients with critical illness. Crit Care Med. 2020;48(2):e115e122. doi:10.1097/CCM.0000000000004133

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5.

    Whitcomb BW, Pradhan EK, Pittas AG, Roghmann MC, Perencevich EN. Impact of admission hyperglycemia on hospital mortality in various intensive care unit populations. Crit Care Med. 2005;33(12):27722777. doi:10.1097/01.ccm.0000189741.44071.25

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6.

    Krinsley JS, Chase JG, Gunst J, et al. Continuous glucose monitoring in the ICU: clinical considerations and consensus. Crit Care. 2017;21(1):197. doi:10.1186/s13054-017-1784-0

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7.

    Edelman SV, Argento NB, Pettus J, Hirsch IB. Clinical implications of real-time and intermittently scanned continuous glucose monitoring. Diabetes Care. 2018;41(11):22652274. doi:10.2337/dc18-1150

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8.

    Ajjan R, Slattery D, Wright E. Continuous glucose monitoring: a brief review for primary care practitioners. Adv Ther. 2019;36(3):579596. doi:10.1007/s12325-019-0870-x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9.

    Aleppo G, Webb K. Continuous glucose monitoring integration in clinical practice: a stepped guide to data review and interpretation. J Diabetes Sci Technol. 2019;13(4):664673. doi:10.1177/1932296818813581

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10.

    Ida S, Kaneko R, Murata K. Utility of real-time and retrospective continuous glucose monitoring in patients with type 2 diabetes mellitus: a meta-analysis of randomized controlled trials. J Diabetes Res. 2019;2019:4684815. doi:10.1155/2019/4684815

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11.

    Corradini S, Pilosio B, Dondi F, et al. Accuracy of a flash glucose monitoring system in diabetic dogs. J Vet Intern Med. 2016;30(4):983988. doi:10.1111/jvim.14355

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12.

    Wiedmeyer CE, Johnson PJ, Cohn LA, Meadows RL. Evaluation of a continuous glucose monitoring system for use in dogs, cats, and horses. J Am Vet Med Assoc. 2003;223(7):987992. doi:10.2460/javma.2003.223.987

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13.

    Hug SA, Riond B, Schwarzwald CC. Evaluation of a continuous glucose monitoring system compared with an in-house standard laboratory assay and a handheld point-of-care glucometer in critically ill neonatal foals. J Vet Emerg Crit Care (San Antonio). 2013;23(4):408415. doi:10.1111/vec.12072

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14.

    Johnson PJ, Wiedmeyer CE, LaCarrubba A, et al. Clinical assessment of blood glucose homeostasis in horses: comparison of a continuous glucose monitoring system with a combined intravenous glucose and insulin test protocol. J Vet Intern Med. 2011;25(1):162165. doi:10.1111/j.1939-1676.2010.0643.x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15.

    Cunneen A, Wood KA, Mathison K, Herndon AM, Bertin FR. Comparison of a continuous indwelling glucometer with a point-of-care device in healthy adult horses. Vet Rec. 2020;187(3):e21. doi:10.1136/vr.105607

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16.

    Roberts MC, Hill FW. The oral glucose tolerance test in the horse. Equine Vet J. 1973;5(4):171173. doi:10.1111/j.2042-3306.1973.tb03219.x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17.

    Altman DG. Practical Statistics for Medical Research. CRC press; 1990.

  • 18.

    Hackett ES, McCue PM. Evaluation of a veterinary glucometer for use in horses. J Vet Intern Med. 2010;24(3):617621. doi:10.1111/j.1939-1676.2010.0481.x

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19.

    Rendle DI, Armstrong SK, Heller J, Hughes KJ. Precision and accuracy of a point-of-care glucometer in horses and the effects of sample type. Vet J. 2019;252:105359. doi:10.1016/j.tvjl.2019.105359

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20.

    Collicutt NB, Garner B, Berghaus RD, Camus MS, Hart K. Effect of delayed serum separation and storage temperature on serum glucose concentration in horse, dog, alpaca, and sturgeon. Vet Clin Pathol. 2015;44(1):120127. doi:10.1111/vcp.12217

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 21.

    Ferrante PL, Kronfeld DS. Effect of sample handling on measurement of plasma glucose and blood lactate concentrations in horses before and after exercise. Am J Vet Res. 1994;55(11):14971500.

    • Search Google Scholar
    • Export Citation

Contributor Notes

Corresponding author: Dr. Malik (cmalik@vt.edu)