Abstract
Objective
To evaluate the performance of a second-generation flash glucose monitoring system (FGMS; FreeStyle Libre 2) in nondiabetic dogs during rapidly induced hypoglycemia.
Methods
In a prospective, observational study, regular insulin was administered to 23 healthy dogs to induce hypoglycemia, with subsequent correction. Serial measurements of interstitial glucose with FGMS and blood glucose (BG) with a portable BG monitor (PBGM) were recorded every 10 minutes for 90 minutes. At each time point, blood was also collected for gold standard BG measurement via clinical chemistry analyzer. Analytical and clinical accuracy was determined by fulfillment of International Organization for Standardization 15197:2013 criteria.
Results
For BG values < 100 mg/dL, 67.2% of PBGM measurements and 50.9% of FGMS measurements were within ± 15 mg/dL of the reference method. In Parkes error grid analysis, 94.9% of PBGM and 88.6% of FGMS measurements fell within zones A and B. Readings within higher-risk zones were rare, with no PBGM values and 0.6% of FGMS values in zone D and neither producing readings in zone E.
Conclusions
In this model of rapidly induced hypoglycemia, the FGMS demonstrated limited analytical accuracy compared to the reference standard. However, it rarely produced glucose readings likely to lead to detrimental clinical decisions.
Clinical Relevance
The FreeStyle Libre 2 is clinically useful for monitoring glucose trends in dogs, though results should be interpreted with caution during periods of rapid glycemic change.
Continuous glucose monitoring systems (CGMSs) have revolutionized the management of diabetes mellitus in human patients. The development and implementation of flash glucose monitoring systems (FGMSs) in veterinary patients have in many ways become more convenient and clinically relevant than in-hospital blood glucose (BG) curves.1 The most widely available FGMS, the FreeStyle Libre 14-day system (Abbott), has previously been validated for use in diabetic dogs and was found to be user friendly, clinically accurate, and well tolerated.2 When compared with traditional BG curves via portable BG monitor (PBGM), FGMSs were shown to more accurately detect the glucose nadir, hypoglycemic events, and variability of day-to-day glycemic control.3 When placed on canine patients hospitalized with uncomplicated diabetic ketoacidosis, FGMSs were clinically accurate but not analytically accurate according to International Organization for Standardization (ISO) 15197:2013 criteria.4 The same criteria were used to assess the analytical accuracy of the 14-day system in nondiabetic dogs and found that the device performed less accurately in the hypoglycemic range (BG < 100 mg/dL).5 However, ISO criteria were specifically developed in order to validate PBGMs for use in type I diabetic humans, and no criteria dedicated specifically to interstitial glucose monitors currently exist.
As the use of FGMSs in veterinary patients evolves, so does the technology. As of January 2021, the FreeStyle Libre 14-day system (Abbott; the first commercially available FGMS) was discontinued and replaced by the FreeStyle Libre 2; shortly afterwards, the FreeStyle Libre 3 emerged on the market. The FreeStyle Libre 2 incorporates Bluetooth technology and offers optional alarms detecting values above and below preset glucose limits.6 The sensor can be paired with a proprietary reader or a smartphone using an application (FreeStyle Libre 2; available via Apple App Store or Google Play Store). Anecdotally, veterinarians using the Libre 2 system reported an increasing number of errors in a discussion on the American College of Veterinary Internal Medicine listserv, and there is a dearth of published evidence on this system’s accuracy in dogs.
This study aimed to analyze the analytical and clinical accuracy of the FreeStyle Libre 2 FGMS in nondiabetic dogs during a period of rapidly induced hypoglycemia.
Methods
Animals
Twenty-three apparently healthy mixed-breed dogs were subjects in a teaching laboratory for veterinary students. All dogs had a complete physical examination performed by a veterinarian (CAP) before the start of the laboratory. Institutional animal care and use committee approval was acquired for the study protocol (IACUC No. 2022-0172).
Glucose data collection
The protocol was similar to that of a previous study.5 Food was withheld from the dogs for 16 hours before data collection, but free access to water was available at all times. One hour before data collection, an FGMS sensor was placed on each dog. The skin between the scapulae was clipped and cleaned with isopropyl alcohol. In addition to the adhesive present on the sensor, 2 to 4 drops of tissue adhesive (Vetbond; 3M Animal Care Products) were added for additional adhesion. Per manufacturer guidelines, the sensor was allowed 1 hour to acclimate before any readings were obtained. Each dog had its own FGMS sensor and reader for the duration of the laboratory. A 22-gauge X 1.25-inch catheter (Cardinal Health Monoject) was placed into a cephalic vein for collecting blood samples and administration of insulin and dextrose solution.
At baseline, before insulin administration, approximately 2 mL of blood was collected via the cephalic IV catheter and discarded using a syringe designated the “dead space syringe.” One milliliter of blood was then drawn from the catheter, and 1 drop of the blood was used to measure glucose with a PBGM (AlphaTRAK 2; Zoetis Animal Health). An individual PBGM was used for each dog to ensure timely measurement of BG. The rest of the blood was placed into a lithium heparin blood tube and submitted within 60 minutes for analysis using a clinical chemistry analyzer (Vitros 4600; Ortho Diagnostics) with a BG reference interval of 60 to 135 mg/dL. An FGMS reader was waved over the SC glucose sensor to obtain a glucose reading from the FGMS at the same time the blood was drawn. After measurements of baseline glucose concentration, regular human insulin (Humulin R; Eli Lilly and Co; 0.3 U/kg) was injected through the catheter. Blood was then collected from the catheter in the described fashion at 10-minute intervals for the duration of the 90-minute study period for measurement via PBGM and clinical chemistry analyzer. At each of these time points, interstitial glucose was also recorded via FGMS. At baseline, and again at 90 minutes, PCV was measured using Hct tubes (Fisherbrand; Thermo Fisher Scientific).
In addition to sample collection and analysis, students were responsible for recognizing clinical and biochemical evidence of hypoglycemia and intervening if deemed necessary. Throughout the entirety of the 90-minute study period, heart rate and mentation were monitored in addition to the BG measurements. When hypoglycemia or suggestive clinical signs (dullness, anxiety, or ataxia) were noted, students could administer a bolus of 50% dextrose solution (Vedco) or feed a commercial dog food (Pro Plan Puppy Chicken and Rice; Purina). The dextrose was first diluted with sterile water to reach a 20% concentration and administered via a catheter at a replacement dose to correct a hypothetical glucose deficit of 60 mg/dL in the extracellular (intravascular and interstitial) space. The “60 mg/dL deficit” refers to a decrease in plasma glucose from 100 to 40 mg/dL. Estimated extracellular fluid volume was used for the calculation because glucose rapidly distributes into both the plasma and interstitial fluid when administered IV. A total of 100 mg/dL was used as the “normal” BG concentration, and 40 mg/dL was designated as “hypoglycemia needing correction.” Students administered the dextrose dose as a single bolus, not as a continuous infusion. All dogs received a single 5-mg dose of dexamethasone (VetOne) administered SC (median dose of 0.2 mg/kg body weight) at the first hypoglycemic episode after insulin administration. The selection and timing of these interventions were largely left to the discretion of the students while under supervision by CAP and SEW. Multiple interventions of dextrose and food, but not dexamethasone, could be administered if persistent hypoglycemia was noted after initial attempted correction. There were no set values for heart rate and mentation that would prompt intervention, but student groups were required to interpret trends in these parameters as they pertained to their specific dog and to act accordingly. Any glucose reading at or below 40 mg/dL required intervention.
After data collection, the IV catheter and glucose sensors were removed from each dog once their BG concentrations stabilized and they clinically appeared euglycemic.
Statistical analysis
Continuous data were assessed for normality using Anderson-Darling tests and visual inspection of quantile–quantile plots. Parametric data are presented as mean (± SD), and nonparametric data are presented as median (minimum to maximum).
Glucose concentrations were compared between 2 test methods (PBGM and FGMS) and a reference method (clinical chemistry analyzer). Bland-Altman plots were constructed, and bias and 95% limits of agreement were calculated using a statistical software package (Prism, version 5; GraphPad Software). Passing-Bablok regression compared the gold standard BG to FGMS and PBGM, respectively, and the 95% confidence bounds were calculated with the bootstrap (bias-corrected and accelerated) method using an online tool (https://bahar.shinyapps.io/method_compare/).
As previously described,4,6 the accuracy of the test methods was assessed using parameters defined in ISO 15197:2013 criteria (“ISO criteria”; BSI Standards Publication, in vitro diagnostic test system-Requirements for BG monitoring system for self-testing in managing diabetes mellitus; EN ISO 15197:2013). It is important to note that these were intended to assess the performance of BG monitors, and so they might not be directly relatable to IG monitors. The mean absolute relative difference, median absolute relative difference, mean relative difference, and mean absolute difference between the test methods (PBGM and FGMS) and the reference method were calculated to assess the analytical accuracy. The number of pairs of glucose concentrations for which the test reading was within ± 15 mg/dL of the reference BG reading for BG concentrations < 100 mg/dL and within ± 15% of the reference BG for reading ≥ 100 mg/dL were calculated. According to the ISO criteria for the assessment of BG monitors, at least 95% of the FGMS readings must adhere to these limitations for the device to be considered analytically accurate.
To assess clinical accuracy, Parkes consensus error grids for type 1 diabetes were constructed comparing BG values via clinical chemistry analyzer to those from both FGMS and the PBGM. These grids divide paired glucose measurements into 5 “zones” depending on the degree of difference and associated risk posed to the patient. The zones are as follows: A, no effect on clinical action; B, altered clinical action unlikely to affect the outcome; C, altered action likely to affect the outcome; D, altered action could carry substantial patient risk; and E, altered action could be dangerous to the patient. According to the ISO criteria for the assessment of BG monitors, 99% of measured glucose results must fall within zones A and B.7 This analysis was performed with an open access statistical software package in R (Error Grid Analysis package, version 2.0.0; The R Project for Statistical Computing).
To further assess the accuracy of measuring IG after a rapid drop of BG concentration induced by exogenous insulin administration, changes in glucose concentration were described over time.
Results
The dogs ranged in age from 2 to 6 years (median, 3.5 years), and their body weights ranged from 18.9 to 40.2 kg (median, 25.8 kg). Twelve dogs were castrated male, and 11 dogs were spayed female. Body condition score was assessed on a 9-point scale,8 and the median score was 5 (range, 4 to 6).
Out of a possible 230 readings, a total of 230 PBGM results (100%) and 180 FGMS results (78%) were available for comparison to a total of 230 values measured via the gold standard (clinical chemistry analyzer). The 50 missed FGMS readings represented a sensor error rate of 21.7%. All missed readings were attributed to a sensor error secondary to acclimation within the first 12 hours of placement, as stated in the user’s manual,9 and were treated as missing data in the analysis. At time point 50 minutes, a single PBGM reading (750 mg/dL) was markedly discordant with both the gold standard (146 mg/dL) and the FGMS (54 mg/dL), consistent with failure to adequately flush the IV catheter after administering a 50% dextrose solution. As the BG from the clinical chemistry analyzer could also have been affected, all measurements for this time point for this dog were excluded from analysis.
The changes in interstitial and BG after insulin administration were followed over time (Figure 1). The median glucose nadir was seen at the 20-minute time point for all methodologies. Notably, median IG is higher compared to both median BG measurements at the 10-minute time point, indicating slow equilibration of glucose between blood and interstitium (lag time).
Median glucose concentrations in 23 healthy dogs after administration of a 0.3-U/kg bolus of regular insulin, IV, at time point 0. Interstitial glucose was measured via flash glucose monitoring system (FGMS; red square), whereas blood glucose measurements were recorded both via portable blood glucose monitor (PBGM; blue circle) and a reference method (REF; black triangle). Glucose concentration nadir occurred at the 20-minute time point for all methodologies. The lag in interstitial glucose is demonstrated by the marked difference in median glucose measurement at the 10-minute time point.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.25.02.0055
Constant bias (95% limits of agreement) was estimated to be 6.3 mg/dL (−31.3.0 to 43.9 mg/dL) and 13.0 mg/dL (−36.3 to 62.3 mg/dL) for the PGMS and FGMS, respectively (Figure 2). For the FGMS, there was evidence of heteroscedasticity with increased variation between methods as the average value increased. Moderate positive correlations between glucose measurements from the PBGM (r = 0.796) and the FGMS (r = 0.516) and the clinical chemistry analyzer were observed (Figure 3).
Bland-Altman analysis of glucose measurements between (A) PBGM and (B) an FGMS and an REF. A—Constant bias was estimated to be 6.3 mg/dL (dashed line), and the 95% limits of agreement were −32.0 to 44.7 mg/dL (dotted lines). Increased variation and heteroscedasticity is evident at higher blood glucose concentrations. B—Constant bias was estimated to be 13.0 mg/dL (dashed line), and the 95% limits of agreement were −36.3 to 62.3 mg/dL (dotted lines). As the average glucose increased, the greater the variation between methods, and heteroscedasticity is the most notable finding.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.25.02.0055
Passing-Bablok regression analysis comparing glucose measurements between (A) PBGM and (B) an FGMS and an REF. The Pearson r indicates moderate positive correlation between PBGM (r = 0.796) and FGMS (r = 0.516) and the reference standard.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.25.02.0055
Parameters assessing analytical accuracy (mean absolute relative difference, mean absolute difference, median absolute relative difference, and mean relative difference) for both test methods are presented in Table 1. The number of readings in the low BG range (gold standard BG < 100 mg/dL) that were within 15 mg/dL of the gold standard were 133 of 198 for the PBGM (67.2%) and 78 of 153 for the FGMS (50.9%). The number of readings in the high BG range (gold standard BG ≥ 100 mg/dL) that were within 15% of the gold standard were 11 of 30 for the PBGM (36.7%) and 11 of 23 for the FGMS (47.8%).
Analytical accuracy of a portable blood glucose monitor (PBGM) and a flash glucose monitoring system (FGMS) as compared to a reference standard.
| PBGM | FGMS | |
|---|---|---|
| Low range (reference method BG < 100 mg/dL) | ||
| n | 198 | 153 |
| MAD (mg/dL) | 13.6 | 20.8 |
| Percentage of MAD values within ± 15 mg/dL of the BG value | 67.2% | 50.9% |
| High range (reference method BG ≥ 100 mg/dL) | ||
| n | 30 | 23 |
| MARD | 19.5% | 22.4% |
| mARD | 18.8% | 19.8% |
| MRD | −3.1% | −8.1% |
| Percentage of values within ± 15% of the BG value | 36.7% | 47.8% |
BG = Blood glucose. MAD = Mean absolute difference. MARD = Mean absolute relative difference. mARD = Median absolute relative difference. MRD = Mean relative difference.
Parkes error grid (Figure 4) illustrates the clinical accuracy for both testing methods. For the PBGM, 94.92% of values were within zones A and B, with 188 of 228 (82.5%) and 29 of 228 (12.7%) of values in zones A and B, respectively. For the FGMS, 88.6% of values were within zones A and B, with 112 of 176 in zone A (63.6%) and 44 of 176 in zone B (25%). According to the ISO 15197:2013 criteria for BG monitors in type I diabetes mellitus, ≥ 99.0% of the measured glucose results should fall within zones A and B, indicating they are unlikely to affect clinical outcome. Measurements in higher-risk zones were rare, with 0 PBGM (0%) and 1 of 176 FGMS (0.6%) measurements classified as zone D and no measurements from either device falling within zone E.
Parkes consensus error grid for type 1 diabetes comparing PBGM with REF (A) and comparing FGMS with REF (B). Zones are categorized as follows: A, no effect on clinical action; B, altered clinical action unlikely to affect outcome; C, altered clinical action likely to affect clinical outcome; D, altered clinical action could have substantial medical risk; and E, altered clinical action could have dangerous consequences. Based on the International Organization for Standardization 15197:2013 criteria, ≥ 99% of the measured glucose results should fall within zones A and B.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.25.02.0055
Discussion
In this model of rapidly induced hypoglycemia in nondiabetic dogs, there was limited agreement between the second-generation FGMS and the clinical chemistry analyzer when assessed using ISO criteria. The FGMS was less accurate than the PBGM, not performing as well in measures of analytical and clinical performance. This is not surprising as the PBGM and clinical chemistry analyzer measure BG concentrations, whereas the FGMS measures interstitial glucose concentrations.
At the time of writing, this is the first study evaluating the FreeStyle Libre 2 in the dog. Berg et al10 assessed the accuracy of the FreeStyle Libre 2 when compared to the AlphaTrak PBGM in diabetic cats using an insulin clamp model. Their study10 concluded that the FGMS routinely underestimated glucose concentrations during euglycemia and equivocal hypoglycemia but overestimated the glucose concentrations during true hypoglycemia. Similarly, when the first iteration of the Freestyle Libre (FreeStyle Libre 14-day system) was evaluated in diabetic dogs, it was found to be less accurate (93% of readings within zones A and B on error grids) at a low BG concentration compared to a normal or high BG concentration (99% and 99% of readings within zones A and B, respectively).2 A recent study11 assessing the accuracy of an unspecified FreeStyle Libre model in sick juvenile dogs admitted to a veterinary teaching hospital ICU found that it had acceptable clinical accuracy as compared to a PBGM, although very few of the readings on either the FGMS or PBGM fell into the hypoglycemic range.
In this study, Passing-Bablok regression analysis comparing both the PBGM and FGMS to the reference method revealed a moderate positive correlation (r = 0.796 and r = 0.516, respectively). This correlation was stronger for the PBGM than the FGMS. Bland-Altman analysis for the PGMS and FGMS revealed a relatively modest positive bias (ie, they read lower than the clinical chemistry analyzer). However, the 95% limits of agreement were relatively wide, especially for the FGMS: −32.0 and 44.7 mg/dL for the PBGM and between −37.3 and 63.3 mg/dL for the FGMS (Figure 2). Proportional bias was not apparent for either. The combination of a relatively small absolute bias, wide 95% limits of agreement, and lack of proportional bias for the FGMS suggests that much of the variation in glucose concentration measurements between it and the clinical chemistry analyzer–determined BG is seemingly random. Our previous study5 evaluating the FreeStyle Libre 14-day system in healthy dogs undergoing rapidly induced hypoglycemia noted a similar pattern of variation.
The application of ISO 15197:2013 to BG monitors for self-monitoring in human patients with type I diabetes ensures accuracy and provides standards for emerging technologies to increase patient safety and optimize disease management. These standards require narrower differences at lower glucose concentrations given the relative risk of hypoglycemia as compared to hyperglycemia. In our study, both methods possibly performed better during times of relatively low BG (< 100 mg/dL), with 67.2% of PBGM readings and 50.9% of FGMS readings with acceptable limits as defined by ISO 15197:2013 (within ± 15 mg/dL of the reference BG value). In comparison, when BG concentrations were > 100 mg/dL, 36.7% of PBGM and 47.8% of FGMS values were within acceptable limits (± 15% of the BG value). It should be noted that the poorer performance of both devices when BG concentrations were > 100 mg/dL is likely overexaggerated by the relatively few readings within that range (n = 30 for the PBGM; n = 23 for the FGMS), likely due to the nature of our study design. A previous study5 from our group revealed a similarly low number of readings (n = 41 for the PBGM; n = 37 for the FGMS) and a narrowly better performance of both PBGM and FGMS in this higher BG range (44% and 54%, respectively). Neither device met the criteria as outlined by ISO 15197:2013, specifically that 95% of readings fall within ± 15 mg/dL of the reference BG readings for BG concentrations < 100 mg/dL and within ± 15% of the reference BG for BG concentrations ≥ 100 mg/dL. As discussed below, it is important to state that these criteria were not intended to assess the accuracy of FGMS.
The clinical accuracy of glucose measurements by both FGMS and PBGM is visually summarized in the Parkes consensus error grids (Figure 4). Error grids were initially devised in 1987 to determine the accuracy of PBGM in the self-monitoring of diabetes mellitus in a clinically meaningful way for both researcher and physician.12 Blood glucose values are plotted on a grid in which the x-axis represents the actual BG and the y-axis represents the measured BG. The zones, labeled zone A through E, correspond to the degree of risk posed by an incorrect glucose measurement, where zones A and B represent a relatively low risk of affecting clinical outcome. The error grids were updated in 2000 after a survey of 100 endocrinologists with the intention to apply advancements in the field while correcting nuances in the zone boundaries.7 In our study, 94.9% of PBGM measurements and 88.5% of FGMS measurements fell within zones A and B, reflecting a high proportion of clinically acceptable results. While neither device met the ISO 15197:2013 threshold of > 99% in these zones, the proportion of values in higher-risk zones D and E was minimal. Specifically, 0 PBGM values (0%) and 1 of 176 FGMS values (0.6%) fell within zone D, and neither device had values within zone E. These findings suggest that, despite not agreeing with the reference standard, both devices provide clinically useful data with a low likelihood of leading to harmful clinical actions.
The lag time between blood and interstitial glucose concentrations, as demonstrated by the 2-compartment model,13 is a clinically relevant consideration in the evaluation of the analytical accuracy of FGMS. Reported lag times in equilibration of glucose concentrations vary greatly regardless of species,14 and studies in dogs have found a range between 5 minutes and 25 minutes. The lag time is demonstrated in Figure 1, 10 minutes after insulin administration, when the FGMS glucose concentration remains within euglycemic range while the PBGM and the reference method display a notable decrease. Our model assessed healthy dogs, not diabetic dogs, and involved rapid shifts from euglycemia to hypoglycemia to hyperglycemia, possibly exacerbating and confounding the lag effect on IG measurement. This could have contributed to the relatively poor apparent performance of the FGMS.
The sensor error rate (number of times the FGMS reader did not display an IG concentration) in this study was 21.7% (50 of 230). This rate was higher than the rate we reported using a first-generation FGMS (7.1%) using a similar study design.5 Direct comparisons between the error rate reported in these studies cannot be made, and so it is not known if this is due to chance alone of if the second-generation FGMS is more prone to sensor errors. According to the manufacturer, Abbott Laboratories, error messages appear when the Libre 2 sensor is unable to provide a glucose reading (such as during a period of exercise).9 All the sensors in our study were placed by a veterinary technician (C. Knowles) with experience placing sensors in the conventional location on client-owned animals. It is possible that some of the sensors shifted during the duration of the laboratory. A recent study15 assessing the complications associated with flash glucose monitors in client-owned dogs reported dysfunctional sensors (the sensor is attached but displays an error message) in 4 of the 34 study dogs (12%). The most common complication was mild irritation at the site of application in 6 of 34 dogs (18%).15 In a parallel study16 performed in 20 cats, the most frequent complication was premature detachment of the sensor in 5 of the 33 study cats (15%), whereas only 1 sensor became dysfunctional while remaining in place. While sensor error or premature detachment is frustrating, in a patient care setting, it is important to reframe expectations for sensor life and overall data collection. Even a few days of continuous interstitial data provide more information than an 8-hour glucose curve, allowing a clinician to evaluate broader trends in the pet’s glycemic control. Therefore, the occurrence of sensor errors with FGMS may not preclude clinical utility.
As mentioned before, the failure of the studied second-generation FGMS to meet the ISO 15197:2013 criteria does not mean that it is clinically ineffective. These guidelines were instituted to evaluate BG monitors, and comparable standards do not exist for CGMSs. However, the parameters measured in these guidelines provide useful information about the measurements produced by an FGMS in comparison to a reference-standard measurement of BG. Additionally, the Parkes consensus grid was designed to evaluate the accuracy of BG monitors in human patients with type I diabetes. Canine diabetes mellitus resembles human type I disease, but no comparable canine or feline consensus for evaluation has ever been developed. Using these human standards can provide some insight, but failing to meet such standards should not invalidate the judicious use of these devices in practice. Indeed, studies evaluating the clinical performance of CGMSs in humans are met with a significant lack of consensus given the lack of CGMS-specific criteria, the diversity in study design, and the application of statistical analysis, precluding satisfactory comparison between studies17,18 even with the same devices. Emerging standards from the FDA for premarket approval of CGMSs provide additional recommendations, though these standards have not been applied in any veterinary studies to date.19 Studies investigating CGM performance in humans remain inconsistent in their use of ISO criteria or FDA standards.
There are several limitations of this study. First, we only assessed nondiabetic dogs and so were unable to assess the performance of the FGMS in the hyperglycemic range that might have been more clinically relevant. Second, as the model we used induced hypoglycemia rapidly, the lag time between BG and interstitial glucose concentration dropping may have been more pronounced than during slower changes in BG, which may be more clinically relevant. Third, diabetes, other endocrinopathies, and systemic sequalae can affect physiologic parameters, such as skin thickness, body condition, and hydration. The healthy dogs with consistent body condition scores and well-hydrated states represented in our study might not represent diabetic dogs. Finally, dogs with clinical or subclinical hypoglycemia received IV dextrose supplementation and food at the discretion of the students rather than at defined times.
In conclusion, in this model of rapidly induced hypoglycemia, there was incomplete agreement between the IG measured with the second-generation FGMS and reference-standard BG measurements. However, the FGMS seldom produced clinically harmful measurements according to Parkes error grid analysis. Clinicians should interpret FGMS measurements cautiously during rapid BG changes. Where sudden-onset hypoglycemia is suspected, the FGMS should not be the sole method by which the veterinarian makes clinical decisions.
Acknowledgments
The authors acknowledge Mandy Zachgo, Kaitlynn Wolf, and Crystal Knowles.
Disclosures
The authors have nothing to disclose. No AI-assisted technologies were used in the composition of this manuscript.
Funding
The authors have nothing to disclose.
ORCID
J. L. Herman https://orcid.org/0009-0005-3137-9388
C. A. Patterson https://orcid.org/0000-0002-1087-7880
References
- 1.↑
Baldo FD, Fracassi F. Continuous glucose monitoring in dogs and cats: application of new technology to an old problem. Vet Clin North Am Small Anim Pract. 2023;53(3):591–613. doi:10.1016/j.cvsm.2023.01.008
- 2.↑
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):983–988. doi:10.1111/jvim.14355
- 3.↑
Baldo FD, Canton C, Testa S, et al. Comparison between a flash glucose monitoring system and a portable blood glucose meter for monitoring dogs with diabetes mellitus. J Vet Intern Med. 2020;34(6):2296–2305. doi:10.1111/jvim.15930
- 4.↑
Malerba E, Cattani C, Baldo FD, et al. Accuracy of a flash glucose monitoring system in dogs with diabetic ketoacidosis. J Vet Intern Med. 2020;34(1):83–91. doi:10.1111/jvim.15657
- 5.↑
Howard LA, Lidbury JA, Jeffery N, Washburn SE, Patterson CA. Evaluation of a flash glucose monitoring system in nondiabetic dogs with rapidly changing blood glucose concentrations. J Vet Intern Med. 2021;35(6):2628–2635. doi:10.1111/jvim.16273
- 6.↑
Alva S, Bailey T, Brazg R, et al. Accuracy of a 14-day factory-calibrated continuous glucose monitoring system with advanced algorithm in pediatric and adult population with diabetes. J Diabetes Sci Technol. 2022;16(1):70–77. doi:10.1177/1932296820958754
- 7.↑
Parkes JL, Slatin SL, Pardo S, Ginsberg BH. A new consensus error grid to evaluate the clinical significance of inaccuracies in the measurement of blood glucose. Diabetes Care. 2000;23(8):1143–1148. doi:10.2337/diacare.23.8.1143
- 8.↑
Laflamme D. Development and validation of a body condition score system for dogs. Canine Pract. 1997;22(4):10–15.
- 9.↑
Freestyle Libre 2 system support. Abbott. Accessed March 25, 2024. https://www.freestyle.abbott/us-en/freestyle-libre-2-resources.html
- 10.↑
Berg AS, Crews CD, Adin C, et al. Assessment of the FreeStyle Libre 2 interstitial glucose monitor in hypo- and euglycemic cats. J Vet Intern Med. 2023;37(5):1703–1709. doi:10.1111/jvim.16820
- 11.↑
Vigh Z, Johnson PA, Weng HY, Thomovsky EJ, Brooks AC. Interstitial glucose monitoring has acceptable clinical accuracy in juvenile dogs. J Am Vet Med Assoc. 2023;261(10):1475–1408. doi:10.2460/javma.23.02.0103
- 12.↑
Clarke WL, Cox D, Gonder-Frederick LA, Carter W, Pohl SL. Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care. 1987;10(5):622–628. doi:10.2337/diacare.10.5.622
- 13.↑
Cengiz E, Tamborlane WV. A tale of two compartments: interstitial versus blood glucose monitoring. Diabetes Technol Ther. 2009;11(suppl 1):S11–S16. doi:10.1089/dia.2009.0002
- 14.↑
Rebrin K, Steil GM, Van Antwerp WP, Mastrototaro JJ. Subcutaneous glucose predicts plasma glucose independent of insulin: implications for continuous monitoring. Am J Physiol-Endocrinol Metab. 1999;277(3):E561–E571. doi:10.1152/ajpendo.1999.277.3.E561
- 15.↑
Campbell C, Shoelson A, Mahony O. Complications associated with a flash glucose monitoring system in diabetic dogs. Can J Vet Res. 2023;87(4):260–264.
- 16.↑
Shoelson AM, Mahony OM, Pavlick M. Complications associated with a flash glucose monitoring system in diabetic cats. J Feline Med Surg. 2021;23(6):557–562. doi:10.1177/1098612X20965012
- 17.↑
Freckmann G, Pleus S, Grady M, Setford S, Levy B. Measures of accuracy for continuous glucose monitoring and blood glucose monitoring devices. J Diabetes Sci Technol. 2019;13(3):575–583. doi:10.1177/1932296818812062
- 18.↑
Freckmann G, Eichenlaub M, Waldenmaier D, et al. Clinical performance evaluation of continuous glucose monitoring systems: a scoping review and recommendations for reporting. J Diabetes Sci Technol. 2023;17(6):1506–1526. doi:10.1177/19322968231190941
- 19.↑
§ 862.1355: integrated continuous glucose monitoring system. National Archives. Updated March 31, 2025. Accessed August 12, 2024. https://www.ecfr.gov/current/title-21/part-862/section-862.1355
