Accuracy of a flash glucose monitoring system in dogs with DKA | VETgirl Veterinary Continuing Education Podcasts
In today’s VETgirl online veterinary CE podcast, we discuss the use of flash glucose monitoring systems (FGMS) in the management of dogs with diabetic ketoacidosis (DKA). Emerging evidence has promoted the use of FGMS devices for measuring interstitial glucose (IG) levels in diabetics, and using this information as a patient management tool. In fact, you may have even noticed commercials for such devices, which are becoming increasingly popular for humans with diabetes mellitus. Essentially the FGMS consists of a small, disc-shaped sensor that is placed on a shaved patch of skin, typically near the base of the neck in dogs. Interstitial glucose concentrations are read, and the data is stored and made available through a reader device, such as a smartphone. The ability to obtain more frequent glucose assessments without repeated blood draws is clearly appealing! While this device has been investigated in dogs with diabetes mellitus, there is little information about its performance specifically in the setting of DKA. Therefore, in a recent study by Malerba et al entitled Accuracy of a flash glucose monitoring system in dogs with diabetic ketoacidosis, the authors sought to assess the performance and accuracy of FGMS in dogs with DKA.
This study took place at the University Veterinary Teaching Hospital of Bologna over an approximately two-year period. Client-owned dogs that were diagnosed with DKA were enrolled in the study. After diagnosis, they were hospitalized and treated with a continuous rate infusion (CRI) of regular insulin and other therapies. The diagnosis of DKA relied on a few parameters, including a blood glucose (BG) concentration > 250 mg/dL, a venous pH < 7.3 or bicarbonate < 15 mEq/L, and a blood β-hydroxybutyrate (BHB) level > 3.8 mmol/L. Additionally, the dog must have demonstrated at least 2 of the following clinical signs – polyuria, polydipsia, anorexia, severe lethargy, vomiting, or dehydration. Once their BHB was ≤1.0 mmol/L, and once their venous pH was ≥ 7.3 or bicarbonate was ≥ 15 mEq/L, the DKA was deemed resolved. At presentation, the investigators collected and recorded a variety of clinical information, such as body condition score (BCS), venous pH, bicarbonate, lactate, and degree of ketosis, and these parameters continued to be monitored throughout their stay.
The FGMS used in this study was the FreeStyle Libre, which has a glucose detection limit between 20 and 500 mg/dL. Dogs enrolled in the study had the fur on their dorsal neck clipped, and the FGMS was applied and secured with additional tape and a bandage (Vetrap) around the neck. The FGMS was applied either immediately following diagnosis if during working hours, or the following morning if the DKA diagnosis was made after hours. The IG concentrations read from the FGMS were compared to BG concentrations that were obtained within 10 seconds. BG values were measured using a portable BG monitor every 1-2 hours until DKA resolution in all dogs, but then the frequency varied on a case-by-case basis. Importantly, the authors note that the BG monitor used in this study has been previously validated for use in dogs. The sensors were then removed at discharge, and the sites were evaluated to assess for any visible adverse changes to the skin.
The authors used a variety of statistical techniques in this study, including methods to evaluate both the analytical and clinical accuracy of the FGMS. While we will not discuss all of the details of these methods, it is important to note that one measure of analytical accuracy that they used was based on ISO criteria, which requires that 95% of the readings fall within 15 mg/dL of the BG level when the BG is < 100 mg/dL. However if the BG is ≥ 100 mg/dL, the value should fall within 15%. This will be important to know when discussing results.
So, what did the investigators find out, and how can I apply this to my veterinary patients?
A total of 14 dogs were enrolled in this study, and they all had the FGMS applied within 13 hours of admission (median 3 hours). Data was collected up to 14 days in some patients, and for at least 3 days in all patients (median 5.5 days). All dogs tolerated both the application and the presence of the device well. Aside from 1 dog with mild, transient erythema at the application site, no adverse events were noted. This is great news! Dogs’ BCSs ranged from 3 to 9, but the majority of dogs had a BCS of 5 or less. The investigators collected hundreds of paired glucose readings to evaluate, which were gathered both during DKA and after resolution. Median BG during DKA was 240 mg/dL and after DKA resolution was 274.5 mg/dL, but as expected, the range of values was quite large. The authors also evaluated pH, bicarbonate, lactate, and BHB levels several times throughout the pups’ stay, but importantly, none of these variables were found to impact the relationship between IG and BG. In fact, good correlations between IG and BG were observed both during DKA and after resolution, but significant inter-patient variability was observed using the FGMS, which indicates that the device was more accurate in certain patients compared to others.
When looking specifically at the ISO criteria that we discussed, the percentages of results that met these criteria ranged from 0% (during DKA with a BG <100 mg/dL) to 53.6% (after DKA resolution with a BG ≥ 100 mg/dL). In other words, analytical accuracy was not achieved using these parameters. That said, IG and BG still had good agreement both before and after DKA resolution (r = 0.88, r = .93 respectively), and when evaluating if or how the values would affect clinical responses, the authors found good clinical accuracy of the devices. The authors discuss that the ISO standards are not designed for comparing glucose levels from different anatomic compartments, such as IG versus BG, but these criteria were adapted for use here. However, given the degree of agreement overall and the assessment of how these values reflect clinical decision-making, the authors conclude that the FGMS should be considered acceptable for clinical use.
This study had several limitations, which the authors discussed. For instance, they explain that by using a human portable BG monitor for BG measurements, as opposed to the classical reference method, the differences between IG and BG concentrations could be affected. This study also had few hypoglycemic data points, and the use of a single sensor per dog did not allow the precision of the FGMS or the impact of sensor location to be evaluated. The investigators did not record thickness of the skin or subcutaneous adipose tissue in this study, which has the potential to affect FGMS accuracy. To the author’s cred, they were really thorough in their explanations of the study limitations, which were quite informative.
With all of this information in hand, what can we take away from this VETgirl podcast? Well this was the first study that evaluated the FGMS performance specifically in dogs with DKA, and despite not meeting the analytical ISO 2013 accuracy requirements, the device did provide clinically accurate BG estimates. In other words, the authors found that the FGMS performed well enough to help guide clinical decision-making, even though the IG values did not perfectly reflect BG values. In this study, pH, lactate, BCS, BHB, or duration of sensor placement did not impact the sensor’s accuracy, which is helpful given the unpredictability of these variables in DKA. This VETgirl believes that flash glucose monitoring system could be quite useful in monitoring hospitalized DKA patients, with the understanding that we still have more to learn about the precision and accuracy of these devices. Saves our patients’ some pokes!
1. Malerba E, Cattani C, Del Baldo F, et al. Accuracy of a flash glucose monitoring system in dogs with diabetic ketoacidosis. J Vet Intern Med 2020;34:83-91.
BCS – Body condition score
BG – Blood glucose
BHB – β-hydroxybutyrate
CRI – Continuous rate infusion
DKA – Diabetic ketoacidosis
FGMS – Flash glucose monitoring system
IG – Interstitial glucose
In today’s VETgirl podcast is sponsored by Chesterfield Animal Hospital, in Chesterfield Indiana, part of the NVA community of hospitals. Chesterfield is looking for a Managing Veterinarian to Lead a passionate team, help design the practice, and shape the future of veterinary medicine! Now offering $100,000 in bonuses! Go to careers.nva.com.