Aims Results of urinalysis are available earlier than urine culture results. If urinalysis can predict results of culture, early decision can be made on treatment and whether urine samples should be cultured. This study sought to compare the performance of urinalysis tests by automated test strip analyser (nitrite and leucocyte esterase) with flow cytometry (bacteria and white cell count) in different subpopulations and types of samples.

Methods Consecutive urine samples (n=2351) from a population with a median age of 45 years, 37.2% men, were tested. Sensitivity, specificity, positive predictive value and negative predictive value (NPV) of the tests were calculated using contingency tables. The gold standard was positive urine culture with cut-off >10⁵ CFU/mL.

Results 14% of the cultures were positive (95.6% monomicrobial, 74.7% Enterobacteriaceae). Overall, nitrite test was the most specific (98.7%) but the least sensitive (43.2%). Bacteria count was the most sensitive (91.7%) and highly specific (87.5%). In infants <24 months, the sensitivity of bacteria count was reduced (86.1%), but specificity was high (95.9%). The specificity of nitrite was reduced in urine from the in-and-out procedure (81.9%). The sensitivity of bacteria count was reduced in bag specimens urine (83.3%) and in urine from indwelling catheter (84.7%). All tests showed a high NPV. The NPV of the combined flow cytometry tests was higher than those of automated test strip analyser (99.1% vs 97.4%).

Conclusions Overall, the performance of urinalysis is excellent. Flow cytometry tests performed better than automated test strip analyser in ruling out urine to be cultured.