ABSTRACT
BACKGROUND: The IDEXX SediVue Dx (SediVue) is an automated, in-clinic urine sediment analyzer for veterinary patients. The bias between the results from manual microscopy and the SediVue is currently unknown. OBJECTIVES: To assess the diagnostic accuracy of the SediVue, we aimed to determine the bias between the SediVue (index test) and manual microscopy (reference standard) for the quantification of RBCs and WBCs in urine. METHODS: Urine remnant samples were collected from cats and dogs that contained RBCs (n = 462) and WBCs (n = 510). Retrospective analysis of results from urine sediment examinations using both manual microscopy (using a KOVA and DeciSlide system) and the SediVue (1.0.1.3) was performed. Bias was determined with Bland-Altman plots. SediVue-captured images from high-bias samples were reviewed, and biases were compared. RESULTS: The median bias for semi-quantitative RBC and WBC counts was determined for RBC and WBC counts. The cutoffs were RBC ≤ 5/HPF, 0.3; RBC 5.1-10/HPF, 10.1; RBC 10.1-20/HPF, 10.6; and RBC > 20/HPF, 28.93; WBC ≤ 5/HPF, 0.1; WBC 5.1-10/HPF, 2.2; WBC 10.1-20/HPF, 9.4; and WBC > 20/HPF, 26.6. High bias between the methods was identified in 98 samples (21.0%) with RBCs and 77 samples (15.7%) with WBCs. Reviewer-based enumeration of the SediVue-captured images decreased the percentage of samples with high bias to 17.3% for RBCs and to 11.4% for WBCs. CONCLUSIONS: Bias in the RBC and WBC counts between manual microscopy and the SediVue was unlikely to impact clinical interpretations in a majority of cases. Although reviewer enumeration of SediVue-captured images reduced observed bias, inherent differences between methodologies appeared to have a larger impact on the bias.
