Urinary tract infections are characterized by the presence of microbial pathogens within the urinary tract. They represent one of the most common infections in hospitalized and clinic patients.
To model the parameters of the Sysmex UF-1000i to the gold standard, urine culture, and to compare the detection of dipstick leukocyte esterase and nitrates to urine cultures and UF-1000i results.
Data were compared from urine samples collected in sterile containers for bacterial culture and microscopic analysis. One sample was used to inoculate a 5% sheep blood agar and MacConkey agar plate using a 0.001-mL calibrated loop. The second sample was analyzed by urinalysis-associated microscopy. The media plates were investigated for growth after 18 to 24 hours of aerobic incubation at 37°C. The second sample was analyzed for bacteria and leukocytes with the Sysmex UF-1000i according to the manufacturer's guidelines. Three definitions for culture results, sensitivity, and specificity at different cutoff values were calculated for the UF-1000i.
The negative predictive value for any positive culture in the adult population included in the study was 95.5%, and the negative predictive value for positive cultures containing growth of 100 000 or more colony-forming units was 99.3% using the Sysmex UF-1000i.
Sysmex UF-1000i showed 98% sensitivity and 93.7% specificity with a 95.5% negative predictive value. Thus, a negative screen with the UF-1000i using defined thresholds for white blood cell counts and bacteria was likely to be a true negative, decreasing the need for presumptive antibiotics.
Urinary tract infections (UTIs) are characterized by the presence of microbial pathogens within the urinary tract, and they represent one of the most common infections identified in both hospitalized and clinic patients. Most UTIs are easily treated without difficulty; however, in certain patient populations, such as children, pregnant women, and the elderly, significant complications may arise.1 In addition, UTIs can be challenging to diagnose because, although many are obviously symptomatic, others may be seemingly asymptomatic—particularly in patients who are obtunded or comatose.2 Furthermore, the symptoms observed with UTIs are diverse and can range from mild irritation with voiding to bacteremia or even sepsis. When sepsis is present, it can lead to death if it is not diagnosed and treated early.3 To further complicate the issue, some UTIs present with atypical signs and symptoms that are similar to other disease processes. For these reasons, an early, accurate diagnosis of UTI is important.
Because UTIs are so common in clinical medicine, examination of urine specimens accounts for a large part of the workload in many microbiology laboratories. Traditional urine testing may consist of a macroscopic examination, chemistry (dipstick or manual), microscopic examination, urine cytology (in certain situations), and urine culture. Culture of a urine specimen is the method currently used to definitively document a bacterial urine infection and is considered the gold standard, confirmatory test in scientific studies.4 However, there are several issues surrounding the conventional urine culture as currently used. Preparing the culture is laborious and costly. It also requires at least 24 hours of incubation before a result can be reported. This 24-hour delay in laboratory diagnosis can have a significant effect on patient outcomes.5 Additionally, many urine samples sent for culture will not yield any bacteria at all. In fact, percentages for negative urine cultures range from 40% to 70% in some laboratories.6 Thus, a fast, reliable screening method that can accurately identify negative urine samples to exclude them from culture procedures would decrease turnaround time of analysis, reduce workloads and costs, and expedite treatment.
Current screening methods, such as dipstick testing for nitrite and leukocyte esterase in urine and microscopic sediment analysis for bacteria and white blood cells, are rapid, but lack sensitivity.5,7 Furthermore, scoring variation between laboratory technologists and labor intensity complicate microscopic sediment analysis. This was the driving force toward the development of automated methods for microscopic sediment analysis. An example of one such method is the Sysmex UF-1000i (Sysmex, Kobe, Japan). The UF-1000i is a fluorescent flow cytometer that stains cell components with a fluorescent dye and rapidly identifies and measures cells in the urine, including leukocytes, bacteria, and erythrocytes.8 The UF-1000i identifies the different cell types in urine specimens using forward and side light scatter to measure size and density. Next, to measure stainability, fluorescence is added. Bacteria are further incubated in a separate bacterial chamber using a specific stain that prevents interference from other particles to increase the specificity.
In this study, we modeled the parameters of the Sysmex UF-1000i to the gold standard, urine culture. In addition, we compared the detection of dipstick leukocyte esterase and nitrates to urine culture and UF-1000i results. This comparison was done to determine parameters that could be used in the design and implementation of an automated method for urine screening with reflex urine culture.
MATERIALS AND METHODS
Data were compared from urine samples collected in sterile containers for bacterial culture and microscopic analysis during a 3-month period in a laboratory at a 572-bed, acute-care hospital. Upon receipt, one sample was used to inoculate a 5% sheep blood agar and a MacConkey agar plate with a 0.001-mL calibrated loop. The second sample was analyzed by urinalysis-associated microscopy for the presence of epithelial cells, leukocytes, yeast, and bacteria.
The media plates were investigated for growth after 18 to 24 hours of aerobic incubation at 37°C. After incubation, the number of colonies of each morphological type seen on each plate was multiplied by 1000 to give the number of colony-forming units (CFUs) per milliliter of urine. The number was rounded up or down to the nearest power of 10 (log), for example, 12 colonies = 10 000 CFUs/mL and 15 colonies = 20 000 CFUs/mL. Based on preset, validated threshold values, the amount of growth was scored as follows: (1) a negative Gram stain and no growth on the culture plates (no growth), (2) growth of bacteria at less than 104 CFUs/mL, (3) growth of bacteria more than 105 CFUs/mL.
The second sample was analyzed for bacteria and leukocytes with the Sysmex UF-1000i according to the manufacturer's guidelines. The Sysmex UF-1000i is a fully automated, fluorescent flow cytometer that categorizes red blood cells, white blood cells, epithelial cells, small round cells, sperm cells, crystals, and pathologic and hyaline casts, yeast, and bacteria.9–13 An uncentrifuged urine specimen was diluted in 2 different reaction chambers: 1 for bacteria and 1 for remaining urine particles. The staining was performed with fluorescent dyes that bind to the cell nucleus, cytoplasm, and membrane or to the nucleic acid inside a bacterial cell. After staining, the urine particles were transported to a flow cell and passed through a laser beam (wavelength, 635 nm). Side scatter, forward scatter, and intensity of the fluorescence were used to determine characteristics of the individual urine particles, such as size and structure. These features, along with adaptive cluster analysis, were used to identify and quantify particles, and that information is presented in histograms and scatter gram charts.
Using the 3 definitions, culture results, sensitivity, and specificity at different cutoff values were calculated for the UF-1000i. Receiver operating characteristic curves were made using MedCalc software (MedCalc, Ostend, Belgium).
Cost reduction as the result of eliminating cultures for urine samples that were negative according to the UF-1000i was calculated with data for technician hands-on time and material costs. Those costs were compared with hands-on time and costs of the used materials for the UF-1000i procedure.
For the 3-month observation period, 4406 unique urine specimens were included in this study; 4033 (92%) were from adult patients and the remainder (8%) were from children. The 4033 results were obtained from outpatient and inpatient adults. Urine cultures showed no growth in 2034 samples (50%) and growth of less than 104 CFUs/mL in 1812 samples (45%). Of the 4033 results, 754 samples (19%) showed bacterial growth of 105 CFUs/mL or more, and 190 samples (5%) were considered contaminated specimens because they contained multiple species. However, if the growth of bacteria exceeded the definition of a negative result, they were considered to be positive, even though the likelihood of a UTI was low in those patients. Of the culture-positive samples, the most commonly identified microorganisms were Escherichia coli and Klebsiella spp, which is consistent with known epidemiologic data.
Of the 4033 unique urine samples obtained from adult patients, 2736 (67.8%) specimens had positive white blood cell counts and bacteria and 1215 (30.1%) specimens had negative white blood cell counts and bacteria as reported by the UF-1000i. Of the 1215 samples that were negative for both parameters, 55 (4.5%) corresponded to positive urine cultures by any of the previously mentioned definitions. Of those 55 samples, 8 (14.5%) had 100 000 or more CFUs in culture. The negative predictive value for any positive culture in the adult population included in the study was 95.5% and, the negative predictive value for positive cultures containing growth of 100 000 CFUs or more was 99.3%. The estimated potential reduction in urine cultures performed based on those results was 30%. That number is the projected implementation rate and not the absolute value.
There were 373 unique urine specimens collected in pediatric patients, defined as those specimens from subjects younger than 18 years. Using the UF-1000i method, 269 specimens (72%) had positive white blood cell counts and bacteria, and 104 (27.9%) tested negative for an elevation in white blood cell counts and bacteria. Of the 104 specimens negative for both parameters, 5 (4.8%) were positive for any amount of bacterial growth in culture, and there were no cases containing growth of 100 000 CFUs or more. The negative predictive value for any positive culture in a pediatric specimen was 95.2%, and the negative predictive value for positive cultures containing bacterial CFUs of 100 000 or more was 100%. The estimated potential reduction in urine cultures performed in this population was 28%.
The receiver operating characteristic curves for the UF-1000i method among adult and pediatric patient populations are shown in Figures 1 and 2, respectively. The summary of data and calculations are listed in Tables 1 and 2.
In a system with approximately 33 000 urine cultures with microscopic examination performed every year and a 72% negative urine culture rate, the proposed UF-1000i urine screen with reflex culture is predicted to decrease the number of urine cultures preformed every year by 28% to 30% in both adult and pediatric populations. In addition, there will be additional positive effects associated with that reduction in the number of urine cultures performed annually, including a decrease in full-time equivalent technologists by 0.8 to 1.2, a decrease in turnaround time of 3 to 5 minutes, and a decrease in the variability of the assessment of urine cultures among technologists. The results show a high sensitivity and specificity with a high negative predictive value. Thus, clinicians and patients can be assured that a negative screen on the UF-1000i using defined thresholds for white blood cell counts and bacteria is likely to be a true negative. This may also reduce empirical antibiotic treatment, which has profound implications in this time of increased antibiotic resistance. Perhaps of greater importance is the ability to start appropriate antibiotics with an hour of positive testing with this method, rather than waiting 24 to 48 hours for a culture to incubate. That time savings could prove crucial in reducing morbidity from a given resistant bacterial infection. In current practice, when a clinician has a suspicion of a UTI, antibiotic therapy is initiated and a wait of 24 to 48 hours is required before a urine culture test confirms a positive diagnosis. If a patient turns out to have a negative culture, they have been exposed to an unnecessary treatment. That unnecessary exposure to antibiotics is considered acceptable, given the deleterious effects of not treating a UTI. However, as shown in our study, the use of the Sysmex UF-1000i can reduce that time on presumptive treatment practice with antibiotics. Decreased antibiotic use will not only reduce direct toxicity to the individual patient but also reduce the risk of Clostridium difficile infection (which has profound morbidity and cost issues), and it will also reduce the risk of antimicrobial resistance. In addition, if the absence or presence of a UTI is a major determinant of whether a patient is discharged from a hospital or sent to a lesser-acuity unit, the decreased turnaround time may result in decreased hospital days for the patient and decreased costs for the hospital. Further research should be performed to validate our study, as well as a clinical study to examine the feasibility of an automated method, such as this one with the Sysmex UF-1000i, to replace traditional urinalysis for infection.
We do acknowledge that if a clinical laboratory moves to this method alone, with no urine culture tests for negative results, there will be a small percentage (roughly 2% to 3%) of cases that are missed, although the specific parameters may be changed to modify that number. In addition, this method cannot currently be used for analysis of other analytes in urine, such as protein or glucose. Finally, additional studies should be done to further elucidate whether the specific scattergrams can be used to identify specific bacterial species or categories of species, such as gram negative versus gram positive results. In examining the value equation in laboratory test use, we believe this method will improve quality (increased sensitivity and specificity, decreased turnaround times) and decrease costs (less testing, less technologist full-time equivalent hours), thus improving overall value. Implementing similar changes in workflow management will have a positive effect on other laboratories and health care institutions.
The authors have no relevant financial interest in the products or companies described in this article.