Performance of Three Portable Blood Glucose Meters in Inland Bearded Dragons (Pogona vitticeps)
Blood glucose concentration measurement is essential for the diagnosis and management of many bearded dragon (Pogona vitticeps) diseases. Portable blood glucose monitors (PBGMs) are inexpensive alternatives to traditional benchtop analyzers and require whole blood volumes as small as 0.3μL. However, PBGMs should be assessed for analytical and clinical agreement with a reference analyzer prior to use in a new species. The potential effects of variables such as packed cell volume (PCV) should also be evaluated. Using blood samples from 48 bearded dragons, three PBGMs were assessed, including a veterinary PBGM (VPBGM) using the canine and feline settings, a human PBGM (HPBGM), and a human point-of-care analyzer (LDX). Statistical analysis was performed using difference plots and Passing-Bablok regression analysis. Analytical agreement was determined using the bearded dragon specific inherent imprecision of each analyzer, and clinical agreement was based on mammalian total allowable error (TEa) guidelines. A multiple linear regression model was used t o investigate the potential effects of PCV, glucose, total solids (TS), lipemia, and hemolysis. The VPBGM overestimated blood glucose on both settings, while the HPBGM and LDX underestimated blood glucose. These respective discrepancies became more pronounced at higher blood glucose concentrations due to proportional biases. No analyzers had analytical agreement with the reference analyzer, and only the LDX was within acceptable clinical decision limits. However, if correction formulas were applied, all analyzers were in clinical agreement. A higher PCV was overall associated with an increasingly negative constant bias. There was no effect of TS concentration or lipemia. While the VPBGM and HPBGM are inexpensive analyzers compared to the LDX and reference analyzer, additional steps, such as the application of corrective formulas, are necessary to ensure acceptable diagnostic results. Alternatively, as precision was good for all analyzers and correlation to the reference analyzer was strong, method-specific reference intervals could be generated.Abstract
Contributor Notes