Perioperative Validation of Two Handheld Glucometers in Dogs Under General Anesthesia: Analytical Robustness and Clinical Risk Assessment

Accurate perioperative glucose monitoring is essential in dogs undergoing general anesthesia, yet most validation studies of handheld glucometers have been performed under stable outpatient conditions. This prospective clinical validation study evaluated the analytical agreement, diagnostic performance, and ISO 15197 compliance of a human-calibrated (Accu-Chek) and a veterinary-specific (Centrivet GK) handheld glucometer compared with a laboratory spectrophotometric reference method in 34 anesthetized dogs (99 paired measurements per device). Linear mixed-effects modeling demonstrated significant method effects (<i>p</i> < 0.001), with the veterinary-specific device overestimating glucose concentrations relative to the reference method (β = 20.79 mg/dL; 95% CI: 8.08-33.50; <i>p</i> = 0.001), whereas the human-calibrated device did not differ significantly (β = 7.18 mg/dL; 95% CI: -5.53-19.89; <i>p</i> = 0.267). Bland-Altman analysis showed mean bias of 4.44 mg/dL (95% CI: 0.73-8.16) for the human-calibrated device and 22.72 mg/dL (95% CI: 18.22-27.21) for the veterinary-specific device. Passing-Bablok regression identified proportional bias only for the veterinary-specific device (slope 1.19; 95% CI: 1.01-1.34). ISO compliance was 69.7% and 39.4%, respectively. For hyperglycemia detection, AUC values were 0.9566 (95% CI: 0.8955-1.0000) and 0.9757 (95% CI: 0.9479-1.0000); for hypoglycemia, 0.8567 (95% CI: 0.7557-0.9578) and 0.7376 (95% CI: 0.6056-0.8697). In anesthetized dogs, the human-calibrated device demonstrated superior analytical robustness, whereas the veterinary-specific device showed greater bias and variability.