Degree of variation and reproducibility of different methods for the diagnosis of subclinical endometritis.

Endometrial cytology as a reliable diagnostic technique has been established for the diagnosis of subclinical endometritis (SE) in cows. Several counting techniques have been used to determine polymorphonuclear neutrophils (PMN) in endometrial samples. Information on the agreement between different techniques, however, is limited. Therefore, the objective of this study was to analyze the degree of variation in the percentage of endometrial cells and PMN determined by six different counting techniques. A second objective was to evaluate the interobserver reproducibility of the cell counting by two different examiners. One hundred samples were examined by the different counting techniques. The applied methods counted a total of 100, 300, or 500 cells (C100, C300, C500), respectively. In addition, method HPF100 and HPF300 counted 100 and 300 cells in 10 high-power fields per slide. Finally, one method estimated (EST) the percentage of PMN by screening the slide under the microscope. The interobserver reproducibility between two examiners was analyzed for method C300. The comparison between the six different methods showed a strong compliance (r = 0.77-0.90) with greatest correlation coefficient between C100 and C300. The results of Kappa statistics revealed agreement between methods varying from ĸ = 0.30-0.85, with the greatest agreement between HPF300 and EST. Furthermore, the impact of the different methods on the resulting prevalence of SE was calculated, with the greatest prevalence determined by C100 (33.0%) and the least by HPF300 (10.0%). The results of the interobserver reproducibility showed good correlation and agreement (r = 0.86, ĸ = 0.79). In conclusion, all examined methods were suitable for the cytological evaluation of PMN, with method C100 showing lowest agreement with the other methods. This confirms the hypothesis that a suitable threshold for PMN is not only influenced by, for example, time of sampling postpartum, but also by the diagnostic method. A threshold of 5% PMN seems to be useful when C300 and HPF100 are used, whereas counting 100 cells or estimating the percentage of PMN seems to overestimate or underestimate the prevalence of SE, respectively. In conclusion, method C300 and HPF100 can be recommended as methods of choice for evaluating the percentage of PMN in endometrial samples to diagnose SE.