Evaluation of a Conventional Urine Glucose Test Strip Method for Detection of Glucosuria in Dogs and Cats
Measurement of urine glucose concentration can be important for disease detection as well as for monitoring disease control, e.g., diabetes mellitus. Point-of-care testing such as the Bayer MultistixTM is often used; however, the overall accuracy of the strips in canine and feline urine has not been evaluated. Therefore, the purpose of this study was to assess the sensitivity, specificity and positive and negative predictive values of the Bayer MultistixTM for detection of glucosuria in canine and feline urine.
Urine samples submitted to the Auburn University Clinical Pathology Laboratory for urinalysis were collected over a 3-month period; 258 and 98 canine and feline samples, respectively, were used. The Multistix reagent strips were used according to the manufacturer's instructions. Analyte reactions were determined by means of a Clinitek 50 Analyzer (Bayer). The glucose concentration in all samples was quantified with a Hitachi 911 Chemistry Analyzer (Boehringer Mannheim Corp.) and considered to be the true concentration (gold-standard). Reference intervals that correlate with the colors on the test strip were devised based on the package insert (negative = <75 mg/dL; trace=76-175 mg/dL; 1+ = 176-375 mg/dL; 2+ = 376-750 mg/dL; 3+ = >750 mg/dL). All samples with concentrations >76 mg/dL were deemed positive.
Sensitivity for the conventional test strip for glucosuria in canine and feline samples was 15% and 73%, respectively. Specificity for the conventional test strip for glucosuria in canine and feline samples was 99% and 97%, respectively. The positive predictive values for detection of glucose were 90% and 73% and the negative predictive values were 59% and 97% for dogs and cats, respectively. Overall, the accuracy of the test strip for classifying the concentration of a sample in the correct interval was 59% in dogs and 91% in cats. Of the misclassifications, 101/106 (95%) and 5/9 (56%) were underestimations in dogs and cats, respectively.
Therefore, the strips appear to be more accurate in cats than in dogs for detection and quantization of glucosuria, and inaccuracies tend to be underestimations. In dogs, the test strips have a high percentage of false negative results for detection of glucosuria.