The aim of this study was to optimise the diagnostic screening algorithm for UTIs, enabling rapid, highly sensitive detection within minutes, while minimising the need for confirmatory cultures. High sensitivity is a fundamental requirement for an effective screening test [7, 17]. However, there is no universally accepted threshold for diagnostic sensitivity, as it depends on factors such as screening algorithm (unselected vs. selected populations), test availability, disease prevalence and the clinical consequences of underdiagnosis. In practice, a sensitivity of 90% is generally considered to be acceptable for diagnostic tests in symptomatic infections. Screening tests are typically followed by confirmatory assays with high specificity.

In the present study, we compared conventional DS with modern FC for rapid UTI screening, using culture as the diagnostic reference standard [10]. In patients with severe UTI, the impact of culture results on treatment initiation is often limited due to the prolonged turnaround time (48–72 h). Although experimental approaches such as direct disc diffusion testing [18] or rapid liquid culture [19] may shorten turnaround time, they still require incubation and conventional culture confirmation. Furthermore, contamination due to improper sampling is typically identified only after a delay of 2–3 days, limiting the clinical utility of culture alone.

Rapid diagnostic methods are therefore needed to distinguish between negative and potentially positive urine samples early. The results of initial screening will directly influence clinical decision-making. Sensitive screening strategies allow confirmatory cultures to be restricted to suspected samples and may guide decisions regarding early empirical antibiotic therapy. Both underdiagnosis (low sensitivity) and overdiagnosis (low specificity) should be minimised to reduce the risks of undertreatment and overtreatment [20].

For many years, urine DS testing based on LE and nitrite has been widely used despite its well-known limitations in sensitivity and specificity [17]. In contrast, newer methods enabling quantitative bacterial detection have become available and offer improved diagnostic performance [21].

In this study, we confirmed that surrogate markers of infection detected by DS (LE and nitrite) have limited diagnostic value due to suboptimal sensitivity and specificity, which was consistent with previous reports [16, 17]. A sensitivity of 90% was achieved only when LE and nitrite were combined. No single DS parameter met the threshold criteria for sensitivity expected for a reliable screening test. Moreover, reliance on leucocyturia is inherently problematic for the detection of UTI, as it depends on host immune response, which may be impaired in patients with leucocytopenia or immunosuppression.

Current international guidelines support the use of direct quantitative bacterial detection as a first-line approach. Indirect markers such as leucocytes are considered a secondary choice [7]. FC enables rapid quantitative assessment of bacteria and other urinary particles (LC, EC, SQC, crystals) within minutes and has recently been complemented by more advanced automated imaging systems [22]. In previous studies, FC appeared to be superior to imaging methods for quantitative bacterial detection with lower false-positive rates, whereas imaging may be advantageous for crystal characterisation and more detailed discrimination of morphological changes [23].

Our findings demonstrate the practical benefits of an optimised screening approach for both early rule-out and rule-in of UTIs. Within the constraints of the study design, a BC < 100/µL (FC) was associated with a very low probability of UTI and may allow a substantial proportion of samples (approximately 55%) to be excluded at an early stage from further testing. Compared with DS, this approach reduced the number of unnecessary cultures while maintaining higher sensitivity. Previous studies have proposed similar cut-off values of a BC ≥ 100/µL [13, 14, 19], which is consistent with our results. The optimised FC-based algorithm reduced unnecessary cultures by approximately 18% compared with combined DS criteria (surrogate markers).

Quantitative FC parameters were strongly associated with the probability of UTI. Higher BC values correlated with an increased likelihood of infection, which may facilitate early clinical decision-making, particularly with regard to empirical antibiotic therapy. We therefore suggest that quantitative BC values, together with an estimated probability of UTI, should be reported to clinicians to facilitate individualised disease management.

The combination of low BC (100–1,000/µL) with high SQC (≥ 10/µL) was a robust predictor of contamination (74%). As these parameters are available within minutes, early identification of samples with presumed contamination enables prompt repeat sampling of properly collected midstream urine samples, thus reducing diagnostic delay.

This study also demonstrated distinct quantitative profiles (FC) for Gram-negative and Gram-positive infections. These differences may reflect variations in bacterial growth (BC) and host cell response (LC). Growth kinetics and inflammation are influenced by pathogen-specific virulence factors and host characteristics [24]. Notably, infections involving two uropathogens showed similar FC profiles to monomicrobial Gram-negative infections, suggesting that such cases may represent true infections rather than contamination.

We also identified species-specific differences in FC parameters, including lower BC in infections caused by Proteus spp. and P. aeruginosa, and higher BC in Aerococcus spp. infections. Increased inflammatory response (LC) was particularly evident in infections with Enterobacter and Serratia spp. The underlying mechanisms for these differences remain unclear and warrant further investigation.

Interestingly, elevated SQC values were observed in samples with group B streptococci (S. agalactiae), closely resembling contamination patterns. This suggests that many such findings may reflect contamination rather than true infection. The clinical relevance of enterococci and GBS in UTIs is also discussed in other studies [25].

Significant sex-related differences were observed in infection rates, contamination rates and pathogen distribution. Women had higher rates of both UTIs and contamination, with E. coli predominating. In men, infections were more heterogeneous and included a higher proportion of more antimicrobial-resistant organisms, such as P. aeruginosa, KES group species and Gram-positive bacteria.

Unexpectedly, contamination rates were highest in younger and middle-aged women rather than in older individuals. This finding contrasts with expectations based on frailty and impaired sample collection in older patients. One possible explanation is a higher mucosal bacterial load in premenopausal women, although this hypothesis requires further confirmation. These findings highlight the importance of proper patient instruction for midstream urine collection, particularly in younger women.

Contamination leads to diagnostic uncertainty, delays and the need for repeat testing. Based on elevated SQC values, repeat sampling should be performed in cases with suspected contamination to improve diagnostic accuracy [26].

The exploratory UTI score demonstrated excellent discrimination but did not improve sensitivity compared with BC alone. In this study, direct bacterial detection with a low cut-off value remains the best unselected primary screening marker to rule out UTI. Multiparametric approaches are more complex and may contribute to secondary risk stratification. Recent studies have demonstrated the benefits of structured diagnostic algorithms to improve antimicrobial stewardship and clinical decision-making [21, 27]. Our results support the implementation of a stepwise diagnostic approach: first, exclusion of UTI in samples with BC < 100/µL; second, culture confirmation for samples above this threshold; and third, risk stratification of suspected samples based on quantitative FC values for bacteria and different host cells.

An excellent prediction of UTI by quantitative BC values was achieved for samples without squamous cell contamination (SQC < 10/µL) (Table 3). Samples with SQC ≥ 10/µL and with low, moderate and high BC (BC up to 3,000/µL) are at increased risk of contamination. Only very high BC values (> 3,000/µL) are highly predictive of UTI despite increased SQC (≥ 10/µL). A similar stratification of suspected samples can also be achieved using the UTI score, a multiparametric model based on quantitative FC parameters including BC and SQC (Supplementary File 1 and 2).

This risk-based approach may guide clinical decision-making, including the initiation of empirical antibiotic therapy in patients with a high probability of UTI, while encouraging a more cautious approach in low- and intermediate-risk cases, particularly when contamination is suspected.

This study has several important limitations. First, urine cultures were performed only in samples with BC ≥ 100/µL according to the routine diagnostic algorithm. Consequently, the study is subject to partial verification bias, as samples below this threshold were not microbiologically verified. This design may have resulted in an overestimation of the diagnostic performance of FC, particularly the apparent sensitivity at the predefined BC screening threshold. Therefore, the reported sensitivity should be interpreted within the context of the study design rather than as an absolute estimate. Prospective studies performing urine culture irrespective of FC findings are needed to determine the true diagnostic accuracy of this approach. Second, this was a single-centre observational study, which may limit generalisability to outpatient populations or other healthcare settings. Third, detailed clinical information – including symptoms, prior antibiotic therapy, catheterisation status, and other patient characteristics – was not systematically available. Finally, although the proposed UTI score showed promising discriminatory performance, it remains exploratory and requires independent external validation before routine clinical implementation.

The proposed diagnostic algorithm was designed for a hospital setting with a focus on high sensitivity to avoid underdiagnosis of infections. In outpatient settings predominantly characterised by E. coli infections in otherwise healthy individuals, a higher cut-off value (e.g. BC ≥ 400/µL) may be acceptable for ruling out uncomplicated UTIs. However, in hospitalised patients with complicated infections, immunosuppression and rare uropathogens, a higher threshold may compromise sensitivity, which may be particularly detrimental in Gram-positive infections with lower BC.

Further validation of this risk stratification approach is required in prospective, multicentre studies involving well-defined patient populations.