Safety-focused triage of suspected bacterial urinary tract infection using urine flow cytometry outputs: prediction of culture positivity.

AIM: To evaluate the urine flow cytometry flagging feature (UF-5000) for predicting urine culture positivity and assess the agreement between BACT-info Gram outputs and culture Gram classification, with subgroup analyses in catheterised, immunosuppressed, and age-stratified patients. METHODS: This single-centre cross-sectional diagnostic accuracy study included 885 routine urine samples with paired UF-5000 analysis and culture results (440 culture-positive, 445 culture-negative). The diagnostic performance of the urinary tract infection (UTI)-flag (UTI vs. non-UTI) was calculated with Wilson 95% confidence intervals. Agreement between two BACT-info definitions and culture Gram category was also evaluated. RESULTS: UTI flagging demonstrated a sensitivity of 0.670, specificity of 0.921, positive predictive value (PPV) of 0.894, negative predictive value (NPV) of 0.739, and an accuracy of 0.797. The negative predictive value declined markedly in high-risk subgroups: catheterised patients (NPV, 0.612) and immunosuppressed patients (NPV, 0.500). Sensitivity was substantially lower in males (0.412) than in females (0.833). Indicator-derived Gram classification showed excellent agreement with culture Gram category (κ = 0.826), whereas categorical BACT-info label mapping yielded fair agreement (κ = 0.294). Bacterial count (BACT) discriminated culture positivity with an area under the curve (AUC) of 0.812 (optimal cut-off 487/µL, sensitivity 0.795, specificity 0.807), whereas white blood cell count (WBC) showed an AUC of 0.749 (cut-off 76.9/µL, sensitivity 0.735, specificity 0.773). CONCLUSION: UF-5000 UTI flagging provides strong rule-in utility for bacterial UTI triage but should not be used as a universal rule-out test, particularly in catheterised and immunosuppressed patients. The BACT count improves discrimination and may support safety-oriented overriding rules.