Urinary tract infection (UTI) refers to the presence of microbial pathogens within the urinary tract. UTIs represent the most frequently occurring infectious diseases in hospitals and community populations [1]. Catheter-associated UTI is the most common nosocomial infection [2], and genitourinary infections are the second most common form of infection among noninstitutionalized elderly [3]. The gold standard diagnostic method for UTI is urine culture, but it takes time and manpower. Rapid screening methods such as flow cytometry have been introduced to reduce the number of urine samples requiring cultures. The Sysmex UF-100 and UF-1000i units (Sysmex, Kobe, Japan) are automated urine flow cytometers that can detect particles, including leukocytes and bacteria, in urine, and some have compared urine bacterial and leukocyte counts with urine culture results with respect to the diagnosis of UTI [4567]. Recently, the UF-1000i was updated to the Sysmex UF-5000 (Sysmex), which has better accuracy for bacterial quantification and stainability [8]. We considered this advantage might provide better performance for the screening out of negative urine samples. Accordingly, we investigated whether WBC, bacteria, and yeast-like cell (YLC) counts and Gram stainability flag results of the Sysmex UF-5000 could be used to screen out unnecessary urine cultures.

Among the 126 urine samples, 46 isolates of microorganisms were grown from 41 samples (32.5%); 12 Escherichia coli, 7 Enterococcus faecalis, 5 Enterococcus faecium, and 22 other bacteria or yeasts (Table 1), which means that 85 (67.5%) samples were negative by culture.

When using the three parameters, leukocyte esterase, nitrite and flag, to select unnecessary culture requests, 69 of the 85 (81.2%) culture negative samples were negative or trace for leukocyte esterase and negative for nitrite by urine reagent strip analysis, and ‘no’ or ‘unclassified flag’ by UF-5000 analysis (Fig. 1). However, using the same criteria, a significant portion of culture positive urines (4/41, 9.8%) were classified in the same way.

Distributions of bacteria, YLC, WBC counts and culture results are shown in Fig. 2. Mean bacteria, YLC and WBC counts were significantly lower in the culture negative group than in the culture positive group (P<0.0001). Receiver operating characteristic (ROC) curves for UF-5000 WBC, bacteria and YLC counts are shown in Fig. 3. Areas under the curves (AUCs) for the WBC, bacteria and YLC counts were 0.74 (95% CI=0.68–0.81), 0.90 (95% CI=0.86–0.94), and 0.92 (95% CI=0.82–1.00), respectively. As shown in Fig. 3, the cut-off values of bacteria and YLC were shifted to the right. This was because our aim was not to find optimal cutoffs based on considerations of sensitivity and specificity, but to find cutoffs that could be used to screen out culture negative samples. When a bacteria count of 50/µL was used the cut-off, 49 negative cultures (including one positive yeast culture) were below this level and these samples did not contain any culture positive samples. When a YLC count of 100/µL was used as the cut-off value, 114 negative urine yeast cultures (including 33 positive bacterial cultures) were below this level, and no positive yeast culture sample was included. For bacteria and YLC counts combined, 48 of the 85 (56.5%) negative urine cultures (38.1% of the 126 samples analyzed; 10.6% of the 453 requested cultures) were found to be below each cut-off value, and no culture positive urine sample was included. WBC count did not well differentiate culture positive and negative samples. More specifically, when a WBC count of 3/µL was used as the cut-off value, only 22 negative cultures had a count below the cutoff.

Urinalysis is performed as a primary routine screening test for almost all in- and outpatient evaluations, and is one of the most frequently requested tests in hospital laboratories. Traditionally, urinalysis is performed using the urine reagent strip test and a manual urine sediment examination. However, recent advances made it possible to perform sediment analysis automatically. These analyzers use two analytical principles for urine sediment analysis. One type is based on image analysis, such as, the IRIS IQ200 analyzer (Iris Diagnostics, Chatsworth, CA, USA) [9], whereas the other is based on flow cytometry, such as, the Sysmex UF Series. These analyzers count many types of particles accurately in a very short time, and classify particles by shape, size and other characteristics, and thus, can predict urine culture results and possibly reduce unnecessary urine cultures by reducing unnecessary urine cultures without sacrificing positive urine cultures. The Korean healthcare insurance system is considering the adoption of diagnosis-related groups (DRGs) for reimbursement, instead of cost-based reimbursement, which means that reducing unnecessary laboratory tests is equivalent to cost savings. Actually, cost calculations on this topic have been previously described. Kim et al. [6] reported a cut-off value of 1,500/µL for bacterial count using the Sysmex UF-100 analyzer could reduce urine cultures by ~40% without any false negative results. Broeren et al. [4] using a negative culture definition of <10⁴ CFU/mL, determined a reduction in unnecessary urine cultures of 28% with 5% false negative results at a bacterial cutoff of 39/µL using the UF-1000i, and De Rosa et al. [5] reported 1.2% false negative results with a culture reduction of 57.1% at a bacterial cutoff of 170/µL and a WBC cutoff of 150 WBCs/µL for the same unit.

The definition used for a positive urine culture result obviously affects screening performance. Some authors consider a urine culture positive if it is >10³ CFU/mL [6], whereas others consider a urine culture positive if it is >10⁵ CFU/mL [10], and thus, sensitivity and specificity are dependent on the definition used even though at the same cut-off values. For example, when the definition of a negative culture value is changed from <10⁴ to <10⁵ CFU/mL, sensitivity increases from 82% to 96%, while specificity decreases from 83% to 78%, respectively, at the same UF-1000i cutoff value for bacteria (200/µL) [4]. In general, ≥10⁵ CFU/mL is considered significant bacteriuria as most symptomatic patients usually have ≥10⁵ bacteria/mL in bladder [11]. However, colony counts used to define significant bacteriuria depend on clinical symptoms, age, gender, urine collection method, isolated species and other factors.

The present study was conducted to determine whether analyzer based testing could predict which urine samples would produce a negative culture result, and not to detect specimens with bacteriuria, which meant the identification of culture negative urines with no or minimal false negative results. This is why we selected <10⁴ CFU/mL as screening criteria for culture negatives. Actually, <10³ CFU/mL would have been more conservative, but only one of the 126 samples had a count between 10³ and 10⁴ CFU/mL (data not shown). A medical record review showed this male patient was treated for respiratory disease and his urine culture result was not clinically significant. So, we tentatively defined urine culture positivity as 10⁴ CFU/mL in this study.

We also evaluated whether urine reagent strip analysis and UF-5000 results predicted urine culture results. When the two reagent strip parameters, leukocyte esterase and nitrite, and one UF-5000 parameter, Gram stainability flag were used, the number of negative cultures required was reduced by 81.2%, but ~10% of culture positive samples were excluded. However, using a bacteria count of <50/µL and a YLCs count of 100/µL as determined by the UF-5000 analyzer, 56.5% of negative cultures requested from departments other than Urology and Nephrology, were classified as unnecessary, and these samples did not contain any positive urine samples. The reason why we excluded samples from Urology/Nephrology Departments is that we assumed specimens from those departments were more likely to be specific disease-oriented and inappropriate to screen for unnecessary cultures. Different criteria should be evaluated for such samples. Ratio of urine cultures requested by Urology/Nephrology departments and others are probably dependent on patient characteristics and hospital size. PNUYH is a 1200-bed tertiary-care general hospital, and 126 of 453 urine specimens were submitted for culture by departments other than Urology and Nephrology during the 11 day evaluation period. Based on the findings of the present study, we estimate that the described screening method would obviate the need to test 10.6% of total samples submitted for urine culture.

In conclusion, the bacteria and YLC count results obtained using the UF-5000 analyzer could be used to predict negative cultures, and reduce urine culture volumes by ~10% above without sacrificing positive urine cultures.