Using the Sysmex UF‐4000 urine flow cytometer for rapid diagnosis of urinary tract infection in the clinical microbiological laboratory

Abstract

Background

Urinary tract infections are responsible for a significant worldwide disease burden. Performing urine culture is time consuming and labor intensive. Urine flow cytometry might provide a quick and reliable method to screen for urinary tract infection.

Methods

We analyzed routinely collected urine samples received between 2020 and 2022 from both inpatients and outpatients. The UF‐4000 urine flow cytometer was implemented with an optimal threshold for positivity of ≥100 bacteria/μL. We thereafter validated the prognostic value to detect the presence of urinary tract infection (UTI) based on bacterial (BACT), leukocyte (WBC), and yeast‐like cell (YLC) counts combined with the bacterial morphology (UF gram‐flag).

Results

In the first phase, in 2019, the UF‐4000 was implemented using 970 urine samples. In the second phase, between 2020 and 2022, the validation was performed in 42,958 midstream urine samples. The UF‐4000 screen resulted in a 37% (n = 15,895) decrease in performed urine cultures. Uropathogens were identified in 18,673 (69%) positively flagged urine samples. BACT > 10.000/μL combined with a gram‐negative flag had a >90% positive predictive value for the presence of gram‐negative uropathogens. The absence of gram‐positive flag or YLC had high negative predictive values (99% and >99%, respectively) and are, therefore, best used to rule out the presence of gram‐positive bacteria or yeast. WBC counts did not add to the prediction of uropathogens.

Conclusion

Implementation of the UF‐4000 in routine practice decreased the number of cultured urine samples by 37%. Bacterial cell counts were highly predictive for the presence of UTI, especially when combined with the presence of a gram‐negative flag.

Urinary tract infections are common and responsible for a significant worldwide disease burden. Diagnosing uropathogens by urine culture is time consuming and labor intensive. Urine flow cytometry might provide a quick and reliable method to screen for urinary tract infection. Implementation of the UF‐4000 in routine practice decreased the number of cultured urine samples by 37%. Bacterial cell counts were highly predictive for the presence of UTI, especially when combined with the presence of a gram‐negative flag.

1. INTRODUCTION

Urinary tract infections (UTI) are among the most common bacterial infections with estimates showing a worldwide incidence of 404.6 million UTI in 2019. ¹ These recent estimates indicate a rise in UTI incidence compared to 1990 which might be related to the increasingly aging world population. ¹ The frequent occurrence of UTI is responsible for a vast amount of antibiotic prescriptions and is a major driver for the occurrence of antimicrobial resistance (AMR). ² The gold standard for the diagnosis of UTI is urine culture. ³ , ⁴ However, diagnosis via urine culture may take up to 24–48 h, and is labor intensive and costly. Rapid diagnosis in patients with suspected UTI can benefit early guided therapy and may avoid more serious complications. Additionally, rapid identification of culture‐negative samples prior to culture may reduce workload and response time and can possibly prevent the unnecessary start of antibiotics.

Automated urine flow cytometry (UFC) is increasingly being implemented in microbiology laboratories to screen for UTI as part of the routine diagnostic analysis of urine samples. UFC analyzes the presence of various human cells and bacteria which can aid to determine the probability of the presence of a UTI. Parameters for the evaluation of UTI in these machines include bacterial count (BACT), leukocyte count (WBC), yeast‐like cell count (YLC), and bacterial morphology (proxy for classical Gram‐stain, from hereon called UF Gram‐flag). Flow cytometers of the UF series (Sysmex, Kobe, Japan) have been evaluated for their ability to detect parameters associated with UTI as compared to urine culture and/or manual Gram stain. ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ The UF‐4000 analyzes various particle types including human cells and bacteria using forward scatter light (FSC) for particle size, side scatter light (SSC) for nucleic acid content, side fluorescent light (SFL) for complexity of particles, and depolarized side scattered light (DSS) to gain information on anisotropy of the particle. ¹⁴ UF‐Fluorocell dyes are used to dye different components of the cells including the cell membrane which indicates dyeability and is used as a proxy for Gram staining. ¹⁴ The UF‐4000 can process up to 80 urine samples per hour. ¹⁴ Previously, various cutoffs for bacterial count (15–350/μL) and leukocyte count (10–300/μL) have been evaluated to predict bacterial growth showing an area under the receiver operating characteristic curve (AUROC) of 0.80‐0.99 and 0.74‐0.88, respectively. ⁵ , ⁶ , ⁷ , ⁸ , ¹⁰ , ¹¹ , ¹² , ¹³ However, while most studies described characteristics of these parameters for their prognostic value, the clinical implementation of these systems and the interpretation of results in clinical practice are not well described. ⁵ , ⁶ , ⁷ , ⁸ , ¹⁰ , ¹¹ , ¹² , ¹³

In this study, we implemented the Sysmex UF‐4000 flow cytometer in our clinical microbiology laboratory and subsequently evaluated whether flow cytometry could improve UTI diagnosis by rapid identification of culture‐positive and culture‐negative samples prior to urine culture. We aimed to provide clinical guidance on how to interpret the results of flow cytometry in the clinical microbiology laboratory.

2. METHODS

2.1. Study design and setting

This mono‐center cohort study included two phases. In the first phase, the UF‐4000 was implemented and thresholds for positivity were determined based on parallel UF‐4000 analysis and urine culture. This threshold for positivity was chosen aiming to reduce the number of performed urine cultures while maintaining high sensitivity for urine samples with pathogenic bacteria. This pre‐culture screening process was thereafter evaluated in the second phase of the study. We also evaluated the prognostic value of the different components of the UF‐4000 to predict pathogenic growth in subsequent urine cultures. These components included bacterial (BACT), leukocyte (WBC), and yeast‐like cell (YLC) counts, and the UF Gram‐flag.

2.2. Sample collection

Urine samples used in this study were routinely analyzed at the microbiology laboratory of the Jeroen Bosch Hospital in the Netherlands. For implementation of the UF‐4000, we used urine samples analyzed in October 2019 while urine samples analyzed between January 2020 and March 2022 were used for validation. Urine samples were obtained from both inpatients and outpatients. Urine samples were collected in sterile cups (Deltalab containers REF 409531 with Vacutest tubes REF 14936) and analyzed ideally within 24 h after collection. Urine samples in this study were collected from two regional hospitals (Jeroen Bosch Hospital and Bernhoven Hospital), 144 general practitioners' offices and 32 elderly care facilities. The specimen transport network that routinely runs between all affiliated facilities and the microbiology laboratory results in a maximal transportation time of 24 h after the sample is brought to the healthcare provider.

A selected group of ‘high risk’ urine samples was excluded from urine flow cytometry analysis by protocol. These samples originated from the oncology or intensive care departments, were collected from children <1 year, women aged between 15 and 30, and pregnant women, or were collected by means of nephrostomy catheter, suprapubic aspiration/catheter, or single‐use catheter. These ‘high risk’ samples were always directly inoculated for culture and were, therefore, excluded from analysis. Additionally, urine samples that were not collected as mid‐stream (e.g., first‐stream, catheter urine, collection bag) were also excluded from analysis since they pose a higher risk of contamination or colonization.

2.3. Diagnostic work‐up

Following implementation, the UF‐4000 was included as part of the routine diagnostic work‐up for urine samples in the microbiology laboratory. In this diagnostic work‐up, urine samples were first screened with the UF‐4000 for the presence of bacteria prior to performing urine culture. Only samples that were flagged positive (≥100 bacteria/μL) by the UF‐4000 were inoculated and streaked on blood agar (trypticase soy agar + 5% sheep blood) and MacConkey agar plates (BioMérieux, Marcy‐l'Étoile, France). The agar plates were incubated at 37°C in 5% CO₂ for 16 h. Growth of bacterial colonies was assessed, and the amount of bacterial growth was classified as either no growth (<10³ colony forming units; CFU/mL), 10³–10⁴, 10⁴–10⁵, or >10⁵ CFU/mL. Only colonies that showed ≥10⁴ CFU/mL growth were identified using the MALDI‐TOF (Bruker Daltonics, Bremen, Germany) and susceptibility was determined using the VITEK2 (BioMérieux, Marcy‐l'Étoile, France) according to EUCAST breakpoints. Identification and susceptibility testing of samples with minor growth (<10⁴ CFU/mL) was only performed in the previously discussed ‘high risk’ samples in line with literature. ¹⁵ Since ‘high risk’ samples were excluded from analysis, minor growth in our study was judged to be non‐pathogenic contamination and defined as either skin or fecal flora. When ≥3 different colony types were present, the urine culture was classified as mixed and no identification or susceptibility testing was performed unless one colony type clearly dominated over the others.

2.4. Definitions

Pathogens that were classified as uropathogens in this study are Enterobacterales and other Gram‐negative rods (including non‐fermenters), Staphylococcus saprophyticus, Enterococcus spp., Aerococcus urinae, Aerococcus sanguinicola, Corynebacterium urealyticum, Gardnerella vaginalis, Actinotignum spp., Alloscardovia omnicolens, Staphylococcus aureus complex, and ß‐hemolytic streptococci. Urinary tract infection (UTI) was defined as growth of ≥10⁴ CFU/mL of at least one uropathogen in urine culture. The UF‐4000 Gram‐flag included: Gram‐positive, Gram‐negative, mixed (both Gram positive and negative), and no bacteria. We did not perform manual Gram stains directly from urine samples. Instead, we compared the UF‐4000 Gram‐flag to actual growth in urine culture. Skin flora in culture growth was classified as being Gram positive, while fecal flora was considered both Gram positive and Gram negative (mixed). We compared the UF‐4000 Gram‐flag to all growth observed in culture and to pathogenic growth in culture to see whether the UF Gram‐flag would provide information about the specific nature of the causative uropathogen.

2.5. Statistical analysis

First, we validated the UF‐4000 total bacterial count (BACT), white blood cell count (WBC), yeast‐like cell count (YLC), and UF‐4000 Gram‐flag for their predictive value of growth of uropathogens in urine culture. We compared median cell counts for samples that were UF negative and UF positive. Culture results of UF‐positive samples were furthermore stratified in pathogenic growth, non‐pathogenic growth, mixed growth, and no growth in urine culture. Statistical testing for continuous variables was performed using non‐parametric Kruskal–Wallis test since BACT, WBC, and YLC counts were not normally distributed. Significance was defined as a p‐value < 0.05. Using arbitrary cutoffs for both bacteria and WBC counts, we calculated a matrix of positive predictive values (PPV) for the presence of a UTI. This matrix was further stratified based on the UF Gram‐flag. Secondly, we evaluated the sensitivity, specificity, PPV, and negative predictive values (NPV) of the UF Gram‐flag compared to actual culture growth. Cohen's kappa (κ) was calculated to assess the agreement corrected for change. The interpretation of κ was performed according to the literature (κ: 0.01–0.20 = slight; 0.21–0.40 = fair; 0.41–0.60 = moderate; 0.61–0.80 = substantial; 0.80–0.99 = almost perfect agreement). ¹⁶ Data were analyzed in R version 4.0.3. ¹⁷

3. RESULTS

3.1. Implementation

The UF‐4000 was implemented in October 2019 using 970 routinely collected clinical urine samples that were tested in parallel on the UF‐4000 and by urine culture. Various cutoffs for bacterial count were evaluated after which an optimum cutoff for positivity (presence of pathogenic growth) was defined at a bacterial count of ≥100 bacteria/μL. This threshold yielded a sensitivity of 96.2%, specificity of 61.9%, positive predictive value (PPV) of 72.7%, and negative predictive value (NPV) of 93.9% (Supporting Information). At this threshold, 19/498 potential uropathogens were not detected, all of which showed growth <10⁵ CFU/mL in urine culture (Table S1). No cutoff for WBC count was set.

3.2. Evaluation after implementation

After the UF‐4000 was implemented, validation was performed with samples collected between January 2020 and March 2022. A total of 66,019 urine samples were received in the microbiology laboratory (Figure 1). Of these, 18,385 ‘high risk’ samples were excluded from analysis since they were not tested on the UF‐4000. Another 4348 samples were excluded since they were not collected as mid‐stream urines. Additionally, 328 urine samples were excluded because these were reported as “not interpretable” by the UF‐4000 due to technical error. Altogether, 42,958 urine samples were included for analysis; 19,760 (46%) samples originated from inpatients (including samples from the outpatient department) and 23,288 (54%) from outpatients (general practitioners and elderly care facilities). Characteristics of included urine samples are shown in Table 1.

FIGURE 1.

Flowchart of included urine samples in validation.

TABLE 1.

Baseline characteristics of urine samples stratified by outcome of UF‐4000 screening.

Characteristics	UF‐4000 positive (n = 27,063)	UF‐4000 negative (n = 15,895)
Female	18,911 (69.9%)	7074 (44.5%)
Median age [IQR]	71 [55–81]	65 [50–76]
Inpatient a	10,543 (39.0%)	9217 (58.0%)
BACT (median/μL [IQR])	8207 [537–48,202]	19 [7–45]
WBC (median/μL [IQR])	243 [42–1275]	8 [2–27]
YLC (median/μL [IQR])	0.7 [0.2–2.2]	0.1 [0–0.6]
Gram‐flag
Gram negative	13,239 (49%)	0 (0%)
Gram positive	7647 (28.3%)	0 (0%)
Mixed gram	3474 (12.9%)	0 (0%)
No bacteria	0 (0%) b	15,895 (100%)

Abbreviations: BACT, total bacterial count; WBC, White blood cell count; YLC, yeast‐like cell count.

^a Includes hospitalized patients and patients from the outpatient department.

^b In 101 (0.4%) UF‐positive samples, the gram‐flag was unidentified, while in 2534 (9.4%) samples, the gram‐flag was not available due to an erroneous technical setting.

3.3. UF‐4000 screen and urine culture

Urine samples that were flagged positive by the UF‐4000 (n = 27,063, 63.0%) were inoculated for urine culture. Growth >10³ CFU/mL after incubation was observed in 25,545 (94.4%) cultured samples while 21,683 (80.0%) showed growth ≥10⁴ CFU/mL. In cultures with pathogenic growth, the majority (99.5%) showed growth of ≥10⁴ CFU/mL. In cultures without uropathogens, the majority (71%) showed only minor growth (10³–10⁴ CFU/mL) (Figure S1). Uropathogens were found in 18,673 (73.1%) samples with growth in urine culture. The most prevalent identified uropathogens were Escherichia coli (n = 11,351), Klebsiella pneumoniae (n = 1949), and Enterococcus faecalis (n = 1199) (Figure 2). A single uropathogen was found in 16,975 (90.9%) urine cultures with pathogenic growth while two uropathogens were identified in 1698 (9.1%) urine cultures. A total of 1529 (6.0%) samples showed growth of ≥3 different colony types and were, therefore, classified as mixed growth. In 5343 samples (20.9%), only growth of non‐pathogenic bacteria was identified. Non‐pathogenic skin flora (n = 4230) or fecal flora (n = 4325) was also observed either solitary or in addition to more abundant pathogenic growth.

FIGURE 2.

Uropathogens identified by urine culture.

3.4. UF‐4000 predictors

The distribution of BACT, WBC, and UF Gram‐flag for each category of the culture outcome is shown in Table 2. The highest numbers of BACT and WBC were observed in urine cultures where uropathogens were identified followed by cultures with mixed growth (Figures S2 and S3). These differences were statistically significant between all groups (p < 0.001). The UF Gram‐flag was more often Gram negative when uropathogens were identified (Table 2).

TABLE 2.

UF‐4000 predictors stratified by outcome of urine culture.

UF‐4000 outcome	Negative	Positive
Culture outcome	–	Uropathogens N = 18,673	Mixed N = 1529	No uropathogens N = 5343	Negative N = 1528
BACT	19	27,556	2780	383	295
(median/μL [IQR])	(7–45)	(3309–64,485)	(419–24,195)	(180–1209)	(161–765)
WBC	7	404	110	68	54
(median/μL [IQR])	(2–27)	(79–1716)	(26–617)	(16–378)	(14–283)
Gram‐flag
Gram negative	–	11.977 (68.8%)	344 (26.1%)	682 (15.5%)	236 (19.2%)
Gram positive		2773 (15.9%)	548 (41.6%)	3408 (77.3%)	918 (74.6%)
Mixed gram		2651 (15.2%)	426 (32.3%)	320 (7.3%)	77 (6.3%)

Abbreviations: BACT, total bacterial count; CFU, colony‐forming units; IQR, interquartile range; WBC, White blood cell count.

Using various cutoffs, we created a matrix of positive predictive values for the presence of a uropathogen in urine culture (Figure 3A). This showed that primarily an increase in bacterial count enhanced the positive predictive value for detecting uropathogens. These predictive values were furthermore stratified for Gram‐negative (GN) (Figure 3B) and Gram‐positive (GP) flagged samples (Figure 3C). At lower bacterial counts, patients with a GN flag were more likely to have a uropathogen as compared to those with a GP flag. The combination of >10.000 bacteria/μL and a Gram‐negative flag had a >90% positive predictive value for the presence of Gram‐negative uropathogens.

FIGURE 3.

(A) Percentage of patients with a UTI based on bacterial and leukocyte counts. (B) Percentage of patients with a UTI based on bacterial count and leukocyte counts in urine samples with a Gram‐negative flag from the UF‐4000. (C) Percentage of patients with a UTI based on bacterial count and leukocyte counts in urine samples with a Gram‐positive flag from the UF‐4000. Grey boxes indicate that none of the samples fulfilled these criteria.

The accuracy of the UF Gram‐flag was compared to the true bacterial morphology of all bacteria found in urine culture and to the identified uropathogens specifically. The GP flag, including samples with a mixed flag (GP and GN), had a sensitivity of 74% and positive predictive value of 83% for the presence of GP bacteria in urine culture. The GP flag identified 95% of GP uropathogens, but the positive predictive value for GP uropathogens was only 20%. The lack of a GP (or mixed) flag had a 99% negative predictive value for the absence of GP uropathogens (Table 3). The GN flag (again including a mixed flag) showed a higher sensitivity (80%) and positive predictive value (97%) for the presence of GN urine cultures. Sensitivity for GN uropathogens was 95% and, in contrast to the GP flag, the positive predictive value of a GN flag was much higher at 80%. The clinical value of a mixed flag was limited. Cohen's kappa was highest for the UF GN‐flag for the presence of GN uropathogens (0.63, substantial agreement).

TABLE 3.

Agreement in gram morphology between the UF‐4000 and urine culture growth.

UF‐4000	Culture (all growth) (n = 23,188)			Culture (pathogenic growth) (n = 14,461/23,188)
	GP a (n = 11,271)	GN a (n = 19,743)	Mixed (n = 7826)	GP a (n = 2164)	GN a (n = 13,856)	Mixed (n = 1559)
True positive	8343	15,881	1550	2057	13,152	563
False positive	1759	511	1845	8045	3240	2832
True negative	10,158	2934	13,517	12,979	6092	18,797
False negative	2928	3862	6276	107	704	996
Sensitivity (95% CI)	74% (73–75)	80% (80–81)	20% (19–21)	95% (94–96)	95% (95–95)	36% (34–39)
Specificity (95% CI)	85% (85–86)	85% (84–86)	88% (87–88)	62% (61–62)	65% (64–66)	87% (86–87)
PPV (95% CI)	83% (82–83)	97% (97–97)	46% (44–47)	20% (20, 21)	80% (80–81)	17% (15–18)
NPV (95% CI)	78% (77–78)	43% (42–44)	68% (68–69)	99% (99–99)	90% (89–90)	95% (95–95)
Cohen's Kappa	0.59	0.47	0.09	0.21	0.63	0.15

Note: All growth includes both pathogenic as well as non‐pathogenic growth while pathogenic growth only includes cultures in which a uropathogen was found.

Abbreviations: NPV, negative predictive value; PPV, positive predictive value.

^a Gram‐positive (GP) and gram‐negative (GN) morphology both include cases that are also included in the mixed stain (GP + GN) category.

The presence of YLC was significantly higher in those with a yeast in urine culture (n = 75, median YLC = 208/μL) as compared to those without yeast (median 0.7/μL, p < 0.001). Nevertheless, the positive predictive value was low because of the limited amount of yeast UTI cases (n = 75) and the high amount of false positives (Table S2). The absence of YLC in the urine had a high (>99%) negative predictive value.

4. DISCUSSION

In this study, we implemented and subsequently evaluated the clinical value of the Sysmex UF‐4000 urine flow cytometer in a microbiology laboratory. First, we investigated the use of the UF‐4000 as a diagnostic screening tool to select urine samples for culture. This decreased the need for urine cultures in our laboratory by 37%. Additionally, we determined the clinical value of multiple parameters provided by the UF‐4000 for the presence of uropathogens in UF‐4000 positive urine samples. We showed that bacterial cell counts (BACT) >10.000/μL were highly predictive for the presence of uropathogens, especially when combined with the presence of a GN flag. The absence of GP flag or yeast‐like cells (YLC) had high negative predictive values and are, therefore, best used to rule out the presence of GP bacteria or yeasts.

Our results showed a decrease in the need for urine cultures by 37% which is in line with the 30‐39% found in previous studies. ⁵ , ⁸ , ¹⁰ The cost per sample screened by the UF‐4000 is approximately 1/3 of that of a negative urine culture in our laboratory. With a 37% reduction in performed urine cultures, but with the extra costs for diagnostics when the UF screen is positive and subsequent urine culture is performed, costs were comparable to only performing urine culture. The clinical benefit of this diagnostic strategy, however, is that 37% will receive a (negative) outcome within the same day which can prevent unnecessary antibiotic prescriptions. Interestingly, the predictive value of the WBC count for the presence of uropathogens was limited in our study. We observed significant WBC counts in samples without growth of uropathogens and even in samples that were screened negative by the UF‐4000 (Figure S3). The finding of pyuria in samples without significant uropathogens is in line with previous studies. ⁵ , ⁷ Sterile pyuria (without bacteriuria) was found in 9% of patients with lower urinary tract symptoms and suspected UTI presenting at the general practitioners' office and might be caused by non‐infectious etiologies. ¹⁸ Another important cause of pyuria without growth on standard cultivation media is anaerobes, especially in patients with underlying illness such as diabetes or urinary tract abnormalities. ¹⁹ , ²⁰ Pyuria without culture growth in our study might also be caused by previous antibiotic treatment or by the presence of slow‐growing or fastidious bacteria that did not grow using our diagnostic routine. Inversely, 19% (3548/18536) of urine cultures with uropathogens had a WBC count <50/μL. This might be associated with asymptomatic bacteriuria in which pyuria is not always present. ²¹ It has also been shown that extended time between sample collection and analysis is associated with decreased leukocyte counts. ²² Since 96% of urine samples included in our study were collected and analyzed within 1 day, we do not believe this has affected WBC counts in our study.

The main strength of our study is the large number of included urine samples. We provide the largest study to date in which the clinical use of the UF‐4000 is validated in routine practice. Our study also describes the clinical benefit for predicting the presence of uropathogens which could affect clinical decision‐making within 1 h after the sample is brought to the laboratory.

There are also limitations in our study. First, we were unaware of the clinical symptoms of the patients and could, therefore, not estimate the added value of the UF‐4000 prediction over the a priori likelihood of UTI. We can, however, assume that samples were sent in for suspected UWI. Since Gram‐negative bacteria are the most common cause of UTI, it is not surprising that UF‐4000 markers indicating the presence of Gram‐negative bacteria had the most predictive power. Second, we excluded specific patient groups (pregnant women, young children, ICU, and oncology patients) and alternative collection methods (not midstream) from analysis since these samples were always directly cultured in our laboratory. The results of our study may not be applicable to these excluded groups.

Third, the screening threshold of 100 bacteria/μL for positivity inevitably resulted in false‐negative cases. The sensitivity of this chosen threshold was 96.2% which is in line with previous studies. ⁷ , ¹⁰ All false‐negative samples identified during implementation showed growth <10⁵ CFU/mL in urine culture and included several microorganisms for which the clinical relevance for causing UTI is doubtful. Lower thresholds between 15 and 58 bacteria/μL, such as proposed in previous studies ¹⁰ , ¹² , ¹³ might furthermore decrease false negatives but can also increase the risk of sample‐to‐sample carryover. ⁸ Carryover in the study by Haugum and colleagues was only observed for the threshold of 100 bacteria/μL when a sample was directly tested after a previous sample with high bacterial counts (>10⁵ CFU/mL) and rinsing between samples was limited (setting 0‐0‐1‐2‐3, reflecting the number of rinse cycles after a sample containing 10‐10²‐10³‐10⁴‐10⁵ bacteria/μL, respectively). The rinse mode in our laboratory is 0‐0‐1‐1‐1 (one rinse cycle after samples with ≥10³ bacteria/μL) which could, therefore, theoretically pose a risk of sample‐to‐sample carryover following samples with high bacterial counts.

Fourth, by using this threshold, we made a pre‐selection for UF‐positive urine samples on which we validated the predictive value of WBC and YLC counts. Therefore, we could only indicate the clinical value of these parameters in urine samples with a bacterial count of ≥100/μL since those with lower bacterial counts were not cultured. Our results might be less applicable when alternative thresholds are used.

Fifth, we used bacterial morphology based on urine culture growth instead of making true Gram stains from the original urine samples. A previous study compared the concordance between the UF Gram‐flag, Gram stain from urine, and subsequent urine culture. ⁶ In 344 samples, they observed a Cohen's kappa agreement which was substantial (κ = 0.78) for GN and fair (κ = 0.40) for GP between the UF Gram‐flag and urine culture. This is in line with the classification in our study. Nevertheless, their reported Cohen's kappa was higher when the Gram stain was compared to urine culture (κ = 0.89 for GN and κ = 0.65 for GP) which would suggest that a classical Gram stain is to be preferred over the UF‐flag. ⁶ In our large dataset, we show how to interpret the UF Gram‐flag and benefit from its strengths and weaknesses.

Last, due to an erroneous technical setting, the UF Gram‐flag was unavailable for analysis in 2534 urine samples which had ≥100 bacteria/μL but <10 WBC/μL. This might have reduced power in the bottom row of Figure 3B,C although results are consistent with the rest of the provided matrix.

Of interest, 1528 (5.6%) UF‐4000 positive samples showed no growth in urine culture. This might be caused by the presence of dead bacteria when patients were already receiving antibiotics or be caused by slow‐growing, fastidious, or anaerobic bacteria.

5. CONCLUSION

The UF‐4000 urine flow cytometer reduced the number of routinely performed urine cultures in our microbiology laboratory by 37%. We provide insight into the performance of the UF‐4000 to detect urine samples with a higher probability of including uropathogens. We showed the clinical value of the bacterial cell count and Gram‐flag for predicting the presence of uropathogens and provided insight into how to interpret these parameters. These data can help with the implementation of the UF‐4000 system in microbiology laboratories and help clinicians to interpret the results and aid in their clinical decision‐making.

AUTHOR CONTRIBUTIONS

K.K. designed the study, interpreted and analyzed data, and wrote the article. A.G. collected data and revised the intellectual content. A.L. and P.W. revised the intellectual content. E.K. designed the study, provided critical writing, and revised the intellectual content.

FUNDING INFORMATION

No funding was received for conducting this research.

CONFLICT OF INTEREST STATEMENT

None of the authors disclose any conflict of interest.

Supporting information

Data S1.

Korsten K, de Gier A, Leenders A, Wever PC, Kolwijck E. Using the Sysmex UF‐4000 urine flow cytometer for rapid diagnosis of urinary tract infection in the clinical microbiological laboratory. J Clin Lab Anal. 2024;38:e25004. doi: 10.1002/jcla.25004

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.