Accuracy of the Freestyle Libre 3 Continuous Glucose Monitoring System in Hypo- and Euglycemic Cats

Abstract

Background: The FreeStyle Libre 3 (FSL3) has several improvements compared to previous FreeStyle Libre systems, but its accuracy has not yet been determined in cats. In diabetic people, FSL3 offers increased accuracy, and its smaller size could be advantageous for use in veterinary patients. Objectives: Assess the accuracy of FSL3 in cats with experimentally induced hypoglycemia. Animals: Seven healthy, purpose-bred cats. Methods: Hyperinsulinemic-hypoglycemic clamps were performed. Interstitial glucose concentration (IG), measured by FSL3, was compared to blood glucose concentration (BG) measured by AlphaTrak2. Data were analyzed for all paired measurements (n = 474) and during stable BG (≤ 1 mg/dL/min change over 10 min). Pearson's r test, Bland–Altman test, and Parkes Error Grid analysis (EGA) respectively used to determine correlation, bias, and clinical accuracy. Results: Blood glucose concentration and IG correlated strongly (r = 0.86, p < 0.0001) in stable glycemia and moderately at all rates of change (r = 0.73, p < 0.0001). Analytical accuracy was not achieved, whereas clinical accuracy was demonstrated with 99%–100% of the results in zones A + B of the Parkes EGA. Interstitial glucose concentration underestimated BG in euglycemia and mild hypoglycemia (mean −11.7 ± 11.2, −5.5 ± 9.1, −1.5 ± 6.0 mg/dL in the ranges 91–120, 66–90, and 56–65 mg/dL, respectively), but overestimated BG in marked hypoglycemia (mean 6.3 ± 5.7, 15.7 ± 5.6 mg/dL in the ranges 46–55 and < 45 mg/dL, respectively). Conclusions: The FSL3 underestimates BG across most of the hypo-euglycemic range but overestimates BG in marked hypoglycemia (< 55 mg/dL). Recognizing the proportional, glycemic-dependent bias of FSL3 improves the safety of its clinical application in cats.