Abstract
Performance comparison of automated urine sediment analyzers is difficult due to cutoff-related trade-offs. In this study, we evaluated and compared UF-5000, UAS800, and Cobas u701 using receiver operating characteristic (ROC) analysis. A total of 774 urine samples were analyzed using manual microscopic examination and microbial culture as the reference test. The area under the ROC curve (ROC-AUC) was compared using DeLong's test. UF-5000, a flow cytometry analyzer, demonstrated significantly better performance than both the automated imaging analyzers UAS800 and Cobas u701 for detecting significant bacteriuria or mucous thread. However, its performance was significantly worse for detecting crystals. UAS800 and Cobas u701 showed generally similar performance, but with significant differences for detecting red blood cells, sperm, and significant bacteriuria. Our results highlight the operating mechanism as a key determinant of performance, and also show that analyzers with similar operating mechanisms can still have significant differences in their performance.