The aim of this study was to determine the intra-operator, inter-operator, and inter-method reproducibility and concordance of semiquantitative evaluation of urinary leukocytes examined in bright field, phase contrast, and stained sediment.

Forty urine samples were prospectively collected for diagnostic purposes from client-owned dogs irrespective of underlying disease. Urines (10 mL) were centrifuged at 450 G for 5 minutes and sediments were resuspended in 0,5 mL. From each sample, sediment examination was performed combining bright-field microscopy (BFM), phase-contrast microscopy (PCM), or stained sediment (SS) with microscope slide (SL) or counting chamber (CC). SS was prepared adding 8 µL of stain (modified Sternheimer-Malbin stain) to 80 µL of resuspended sediment. SL was prepared placing 50 µL of sediment or stained sediment on a slide covered by 24 x 32 mm coverslip. CC was prepared placing 10 µL of sediment (or stained sediment) in a counting chamber. Three operators with different experience (expert technician, PhD student, and undergraduate student) quantified with each method the number of leukocytes in 10 fields at 400 x magnification in a blind fashion and the mean was used for statistical analysis. Results of SS + SL and SS + CC were normalized for the stain dilution before analysis. For each sample, coefficient of variation (CV) of the 10 fields interpreted by the expert operator with all the methods were calculated. Friedman and Cohen's k tests were used to compare the difference and the concordance, respectively, between methods for each operator (intra-operator) and between operators for the six methods (inter-operator). The difference between methods was also calculated with Friedman test.

Among the 40 samples, mean ± SD of leukocytes in the six methods were 6.4±8.2, 7.0±9.0, 7.7±12.5, 9.9±14.1, 10.0±13.6, and 9.9±14.1. Mean CV obtained by the expert operator were 90.1%, 99.3%, 92.2%, 77%, 78.8%, and 83.8%. No significant difference was found between methods for each operator. Significant difference between operators was found with BFM+SL (p=0.008) and PCM+SL (p=0.009). No significant difference was found between methods. The lowest intra-operator concordance was found with SS+SL (k coefficient between 0.70 and 0.79) whereas for all the operators lower concordances were found between PCM+SL and SS+SL.

SS yielded lower variability among different operators but it tended to have lower concordance. All methods were reliable for expert operators. However, BFM yielded higher concordance and lower CV. BFM is preferred according to results of this study and given the inexpensiveness and easiness of this method.

Disclosures

No disclosures to report.