Reproducibility of urine sediment examination using two different methods

Veterinary literature lacks of standardized procedures to perform the analysis of urinary sediment. This may decrease the reproducibility of the test. The aim of our study was to evaluate the interoperator and intraoperator reproducibility of two methods of sediment preparation (A and B). Forty-five canine urine samples (20 mLs) were collected during health screening, under owner informed consent, and analysed. Method A: Five mLs of urine were transferred in two conic glass tubes (0.1 mL of graduation). After centrifugation (1500rpm, 5 min), 4.75 mL of supernatant were removed by suction from the meniscus tubes (20 fold concentration). After resuspension, 50 µLs of sediment from each tube were pipetted on microscope slides. A 24x32mm coverslip was used to standardize the height of the fluid layer. Furthermore, from each glass tube, a count chamber was filled with 10 μLs of sediment. Method B: Five mLs of urine were transferred in two conic plastic tubes (graduated at 1, 2.5, 5 and 10 mLs). After centrifugation, the supernatant was decanted by a unique motion. A drop of resuspended sediment was put on a microscope slide by a Pasteur pipette. A 24x32mm coverslip was used. From each plastic tube a counting chamber was prepared. Three different operators counted Red Blood Cells (RBCs) and White Blood Cells (WBCs) in 10 fields per slide (40x magnification) and in 10 small grids of each chamber. The Friedman test revealed the absence of inter-operators and intra-operator variability. Furthermore, the Friedman test showed intra-operator variability between the cells counts assessed in slides but not in chambers in method B. The Wilcoxon test showed a significant difference between counts performed on slides and chambers although they were significantly correlated by Spearman's test (p<0.0001). WBC counts performed by chambers and slides were highly correlated in both methods (ρ>0.95), while RBC counts were poorly correlated (Method A: ρ>0.83; Method B: ρ=0.77). The Wilcoxon test showed that using method A, RBCs were counted differently in chambers (p<0.004) while WBCs in slides (p<0.009) using method B. The Cohen's kappa test showed strong and good agreement in cell counts on slides (RBC-k=0.86; WBC-k=0.70) and chambers (RBC-k=0.92; WBC-k=0.88). No agreement was found between slides and chamber within the same method for RBC (k = 0.30) while for WBC the agreement was good (k>0.60). Data suggest that Method A plus counting chambers provides a greater reproducibility of the sediment analysis, decreasing inter and intraoperator variability.