Poster Presentation Australian Society for Microbiology Annual Scientific Meeting 2019

Automated urine particle analysis: A performance evaluation of three analysers in the investigation of urinary tract infection (#131)

Claire Gregory 1 , Marcel Leroi 1
  1. Austin Health, Heidelberg, VIC, Australia


Whilst considered the gold standard, manual microscopy has become increasingly replaced with automation in the investigation of urinary tract infection. We evaluated the performance of cobas u 701 (Roche Diagnostics) and Atellica UAS 800 (Siemens Healthineers) against our incumbent IRIS iQ200 (Beckton Dickinson). Detection of particles was compared to the reference standard of microscopy, and a correlation of analyser organism detection was made to semiquantitative culture counts and significant bacterial growth.


608 urine samples were run on all analysers and compared to manual microscopy. Data was recorded quantitatively for white blood (WBC) and red blood cells (RBC), semi-quantitatively for bacteria and squamous epithelial cells (SEC), and qualitatively for yeast. WBC and RBC data were divided into categories of 0-10, 11-100, 101-500, and >500 cells/uL. Assessment of bacteria detection was also performed by comparison with total bacterial counts on culture, and with the presence of significant growth of bacteria requiring release of susceptibility results.



Within critical ranges of WBC counts of 0-10 and >500 cells/uL, all platforms had concordance of ≥85% with microscopy. In the non-critical cell types of RBC and SEC, concordance was lower with no clearly superior performance. Concordance for detection of yeast ranged from 72.9-81.7%. The NPV of nil organisms detected by each platform ranged from 74.9-77.9% across all analysers compared to microscopy. When compared to growth of a predominant organism, and to total bacterial counts on culture, the NPV remained below 80%.



Both the u 701 and UAS 800 were considered acceptable alternatives to the iQ200. All performed similarly in correctly designating WBC, RBC, and SEC categories. However, performance for detection of RBC and SEC was poor. Additionally, the NPV for detection of bacteria using any method as the gold standard, was considered too low to permit screening out samples not requiring culture.