Evaluation of one portable blood glucose meter and one portable glucose-ketones meter in dogs

Numerous portable blood glucose meters (PBGMs) have been developed during the last decade, the majority of which is designed for use in humans. Recently one glucometer (Gluco Calea, WellionVet; GC) and one glucose‐ketones meter (Belua, WellionVet; BE) have been developed for use in veterinary medicine. The aims of this study were to assess the accuracy and precision of these devices in feline venous and capillary blood samples based on ISO 15197:2013 and to evaluate packed cell volume (PCV) interferences. Samples were obtained from 29 non anemic cats (PCV 30–47%) and 18 anemic cats (PCV<30%) divided into three glycemic ranges: high (>140 mg/dL), medium (90–139 mg/dL), and low (<90 mg/dL). Paired measurements of glucose and 3‐á°ž‐hydroxybutyrate (3‐HB) from capillary and venous blood samples were determined using the two devices and compared with the results of reference methods (enzymatic hexokinase and 3‐HB‐dehydrogenase, respectively) obtained by an automated chemistry analyzer (Beckman‐Coulter AU480). Linear regression, Bland Altman plots and the Parkes error grid analysis (EG) were used to assess the accuracy. PCV interferences for glucose measurement were assessed comparing the differences between PBGMs readings and reference method values in anemic and non‐anemic cats. To assess within‐run precision, glucose concentrations obtained from 14 samples, belonging to the three glycemic ranges, were measured 10 times within 10 minutes. Between‐day precision was assessed by testing each manufacturer's glucose control solution over 10 consecutive days. P < 0.05 was considered significant. Mean differences (mg/dL) between measurements of each PBGM on capillary and venous blood and values measured by the reference method were: GC 30.7 ± 35.4, 35.6.2 ± 40.5, BE 15.5 ± 35.5 and 15.0 ± 24.1 respectively. A positive significant correlation between all paired samples was found for both devices (r > 0.89). However neither PBGMs totally fulfilled ISO requirements, but 100% of glucose values measured on venous blood using BE fell in zone A+B of EG. Within‐run and between‐day precision were adequate. The effect of PCV was significant (higher results with lower PCV) only for BE. The correlations between capillary and venous 3‐HB and reference 3‐HB were r = 0.66 and r = 0.82 respectively. Mean differences between capillary and venous 3‐HB and reference method were −0.07 (±1.15) and −0.30 (±1.48) respectively; within‐run precision was adequate. Our results show that GC is not sufficiently accurate while the superior performances of BE supports its clinical use in cats.