Use of Automated Urine Microscopy Analysis in Clinical Diagnosis of Urinary Tract Infection: Defining an Optimal Diagnostic Score in an Academic Medical Center Population

Investigating urinary characteristics and optimal urine white blood cell threshold in paediatric urinary tract infection: A prospective observational study

Introduction

While the definitive diagnosis of urinary tract infection (UTI) requires a positive urine culture, the likelihood of UTI can be determined by urinalysis that includes white blood cell (WBC) count. We aimed to determine the optimal urine WBC threshold in urinalysis to predict UTIs in children presenting at the emergency department (ED).

Method

We performed a prospective observational study in the ED at KK Women's and Children's Hospital for children below 18 years old who underwent both urine microscopy and urine cultures, between 10 January and 7 November 2022. We assessed the various urine WBC thresholds associated with culture-proven UTIs using sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios, and area under receiver operating characteristic curve.

Results

We found a culture-proven UTI rate of 460/1188 (38.7%) among all patients analysed, and 278/998 (27.9%) among those with nitrite-negative urine samples. Among all patients, a urinalysis WBC threshold of 100/μL had a sensitivity of 82.2% (95% confidence interval [CI] 78.4-85.5) and negative predictive value of 86.2% (95% CI 83.6-88.4). Among those who were nitrite-negative, a WBC threshold of ≥100/μL resulted in a potential missed rate of 48/278 (17.3%). By lowering the WBC threshold to ≥10/μL, the potential missed cases reduced to 6/278 (2.2%), with an estimated increase in 419 urine cultures annually.

Conclusion

A urine microscopy WBC threshold of ≥100/μL results in a clinically significant number of missed UTIs. Implementation of various thresholds should consider both the potential missed UTI rate and the required resource utilisation.

Development and validation of artificial intelligence models to predict urinary tract infections and secondary bloodstream infections in adult patients

BACKGROUND

Traditional culture methods are time-consuming, making it difficult to utilize the results in the early stage of urinary tract infection (UTI) management, and automated urinalyses alone show insufficient performance for diagnosing UTIs. Several models have been proposed to predict urine culture positivity based on urinalysis. However, most of them have not been externally validated or consisted solely of urinalysis data obtained using one specific commercial analyzer.

METHODS

A total of 259,187 patients were enrolled to develop artificial intelligence (AI) models. AI models were developed and validated for the diagnosis of UTI and urinary tract related-bloodstream infection (UT-BSI). The predictive performance of conventional urinalysis and AI algorithms were assessed by the areas under the receiver operating characteristic curve (AUROC). We also visualized feature importance rankings as Shapley additive explanation bar plots.

RESULTS

In the two cohorts, the positive rates of urine culture tests were 25.2% and 30.4%, and the proportions of cases classified as UT-BSI were 1.8% and 1.6%. As a result of predicting UTI from the automated urinalysis, the AUROC were 0.745 (0.743-0.746) and 0.740 (0.737-0.743), and most AI algorithms presented excellent discriminant performance (AUROC > 0.9). In the external validation dataset, the XGBoost model achieved the best values in predicting both UTI (AUROC 0.967 [0.966-0.968]) and UT-BSI (AUROC 0.955 [0.951-0.959]). A reduced model using ten parameters was also derived.

CONCLUSIONS

We found that AI models can improve the early prediction of urine culture positivity and UT-BSI by combining automated urinalysis with other clinical information. Clinical utilization of the model can reduce the risk of delayed antimicrobial therapy in patients with nonspecific symptoms of UTI and classify patients with UT-BSI who require further treatment and close monitoring.

Current Pyuria Cutoffs Promote Inappropriate Urinary Tract Infection Diagnosis in Older Women

BACKGROUND

Pre-existing lower urinary tract symptoms (LUTS), cognitive impairment, and the high prevalence of asymptomatic bacteriuria (ASB) complicate the diagnosis of urinary tract infection (UTI) in older women. The presence of pyuria remains the cornerstone of UTI diagnosis. However, >90% of ASB patients have pyuria, prompting unnecessary treatment. We quantified pyuria by automated microscopy and flowcytometry to determine the diagnostic accuracy for UTI and to derive pyuria thresholds for UTI in older women.

METHODS

Women ≥65 years with ≥2 new-onset LUTS and 1 uropathogen ≥104 colony-forming units (CFU)/mL were included in the UTI group. Controls were asymptomatic and classified as ASB (1 uropathogen ≥105 CFU/mL), negative culture, or mixed flora. Patients with an indwelling catheter or antimicrobial pretreatment were excluded. Leukocyte medians were compared and sensitivity-specificity pairs were derived from a receiver operating characteristic curve.

RESULTS

We included 164 participants. UTI patients had higher median urinary leukocytes compared with control patients (microscopy: 900 vs 26 leukocytes/µL; flowcytometry: 1575 vs 23 leukocytes/µL; P < .001). Area under the curve was 0.93 for both methods. At a cutoff of 264 leukocytes/µL, sensitivity and specificity of microscopy were 88% (positive and negative likelihood ratio: 7.2 and 0.1, respectively). The commonly used cutoff of 10 leukocytes/µL had a poor specificity (36%) and a sensitivity of 100%.

CONCLUSIONS

The degree of pyuria can help to distinguish UTI in older women from ASB and asymptomatic controls with pyuria. Current pyuria cutoffs are too low and promote inappropriate UTI diagnosis in older women. Clinical Trials Registration. International Clinical Trials Registry Platform: NL9477 (https://trialsearch.who.int/Trial2.aspx?TrialID=NL9477).

A novel approach to screening and managing the urinary tract infections suspected sample in the general human population

Background

Automated urine technology providing standard urinalysis data can be used to support clinicians in screening and managing a UTI-suspected sample. Fully automated urinalysis systems have expanded in laboratory practice. Commonly used were devices based on digital imaging with automatic particle recognition, which expresses urinary sediment results on an ordinal scale. There were introduced fluorescent flow cytometry analyzers reporting all parameters quantitatively. There is a need to harmonize the result and support comparing bacteria and WBC qualitative versus semiquantitative results.

Methods

A total of 1,131 urine samples were analyzed on both automated urinalysis systems. The chemical components of urinalysis (leukocyte esterase and nitrate reductase) and the sediment results (leukocytes and bacteria) were investigated as potential UTI indicators. Additionally, 106 specimens were analyzed on UF-5000 and compared with culture plating to establish cut-offs that can be suitable for standard urinalysis requirements and help to guide on how to interpret urinalysis results in the context of cultivation reflex.

Results

The medians of bacteria counts varies from 16.2 (absence), 43.0 (trace), 443.5 (few), 5,389.2 (moderate), 19,356.6 (many) to 32,545.2 (massive) for particular digital microscopic bacteriuria thresholds. For pyuria thresholds, the medians of WBC counts varies from 0.8 (absence), 2.0 (0-1), 7.7 (2-3), 21.3 (4-6), 38.9 (7-10), 61.3 (11-15) to 242.2 (>30). Comparing the culture and FFC data (bacterial and/or WBC counts) was performed. Satisfactory sensitivity (100%), specificity (83.7%), negative predictive value (100%), and positive predictive value (75%) were obtained using indicators with the following cut-off values: leukocytes ≥40/µl or bacteria ≥300/µl.

Conclusions

Accurate urinalysis gives information about the count of bacteria and leukocytes as useful indicators in UTIs, in general practice it can be a future tool to cross-link clinical and microbiology laboratories. However, the cut-off adjustments require individual optimization.

Verification of Atellica 1500 and comparison with Iris urine analyser and urine culture

Introduction

The aims of study were to assess: 1) performance specifications of Atellica 1500, 2) comparability of Atellica 1500 and Iris, 3) the accuracy of both analysers in their ability to detect bacteria.

Materials and methods

Carryover, linearity, precision, reproducibility, and limit of blank (LoB) verification were evaluated for erythrocyte and leukocyte counts. ICSH 2014 protocol was used for estimation of carryover, CLSI EP15-A3 for precision, and CLSI EP17 for LoB verification. Comparison for quantitative parameters was evaluated by Bland-Altman plot and Passing-Bablok regression. Qualitative parameters were evaluated by Weighted kappa analysis. Sixty-five urine samples were randomly selected and sent for urine culture which was used as reference method to determine the accuracy of bacteria detection by analysers.

Results

Analytical specifications of Atellica 1500 were successfully verified. Total of 393 samples were used for qualitative comparison, while 269 for sediment urinalysis. Bland-Altman analysis showed statistically significant proportional bias for erythrocytes and leukocytes. Passing-Bablok analysis for leukocytes pointed to significant constant and minor proportional difference, while it was not performed for erythrocytes due to significant data deviation from linearity. Kappa analysis resulted in the strongest agreements for pH, ketones, glucose concentrations and leukocytes, while the poorest agreement for bacteria. The sensitivity and specificity of bacteria detection were: 91 (59-100)% and 76 (66-87)% for Atellica 1500 and 46 (17-77)% and 96 (87-100)% for Iris.

Conclusion

There are large differences between Atellica 1500 and Iris analysers, due to which they are not comparable and can not be used interchangeably. While there was no difference in specificity of bacteria detection, Iris analyser had greater sensitivity.

Evaluation of the Atellica® UAS 800: a new member of the automated urine sediment analyzer family

BACKGROUND

In 2017 the Atellica^® UAS 800 urine sediment analyzer was introduced by Siemens Healthineers. We investigated its applicability in the standardization and automation of the laboratory urinalysis workflow, including the prediction of urine culture outcome and glomerular pathology.

METHODS

We evaluated the performance characteristics of the Atellica^® UAS 800 and its correlation with the iQ200 (Beckman Coulter). In addition, we studied the agreement between Atellica^® UAS 800 and CLINITEK Novus^® and determined the predictive value of bacteria and leukocyte counts for urine culture outcome. Furthermore, we investigated the ability of Atellica^® UAS 800 to identify pathological casts and dysmorphic erythrocytes in comparison to manual microscopy.

RESULTS

Erythrocyte and leukocyte analyses indicated high intra- and inter-run precisions and good correlations with the iQ200. We found that the Atellica^® UAS 800 detects bacteria with higher sensitivity than the iQ200. The Atellica^® UAS 800 and CLINITEK Novus^® showed a high degree of conformity. We determined seven combinations of clinical cut-off values of bacteria and leukocytes for predicting urine culture outcome with sensitivity, specificity, and negative predictive values of 95%, 52%, and 93%, respectively. Using the Atellica^® UAS 800, hyaline casts, erythrocyte casts, leukocyte casts, and dysmorphic erythrocytes were correctly recognized in 76%, 22%, 2%, and 39% of the samples, respectively.

CONCLUSIONS

The Atellica^® UAS 800 is a robust, fast, and user-friendly analyzer, which accurately quantifies erythrocytes, leukocytes, bacteria and squamous epithelial cells, and may be utilized for predicting positive urine cultures. The detection of clinically important pathological casts and dysmorphic erythrocytes proved insufficient.

Deconstructing the urinalysis: A novel approach to diagnostic and antimicrobial stewardship

The extensive use of the urinalysis for screening and monitoring in diverse clinical settings usually identifies abnormal urinalysis parameters in patients with no suspicion of urinary tract infection, which in turn triggers urine cultures, inappropriate antimicrobial use, and associated harms like Clostridioides difficile infection. We highlight how urinalysis is misused, and suggest deconstructing it to better align with evolving patterns of clinical use and the differential diagnosis being targeted. Reclassifying the urinalysis components into infectious and non-infectious panels and interpreting urinalysis results in the context of individual patient’s pretest probability of disease is a novel approach to promote proper urine testing and antimicrobial stewardship, and achieve better outcomes.

BactoSpin: Novel Technology for Rapid Bacteria Detection and Antibiotic Susceptibility Testing

Inappropriate use of antibiotics is one of the leading causes of the increasing numbers of resistant bacteria strains, resulting in 700,000 deaths worldwide each year. Reducing unnecessary use of antibiotics and choosing the most effective antibiotics instead of broad-spectrum drugs will slow the arms race between germs and humans. Urinary tract infections (UTIs) are among the most common bacterial infections. Currently, accurate diagnosis of UTI requires approximately 48 h from the time of urine sample collection until antibiotic susceptibility test (AST) results. This work presents a rapid bacterial detection device that integrates a centrifuge, microscope, and incubator. Two disposable microfluidic chips were developed. The first chip was designed for bacteria concentration, detection, and medium exchange. A second multi-channel chip was developed for AST. This chip contains superhydrophobic and hydrophilic coatings to ensure liquid separation between the channels without the need for valves. The designed chips supported the detection of E. coli at a concentration as low as 5 × 10³ cells/mL within 5 min and AST in under 2 h. AST was also successfully performed with Klebsiella pneumonia isolated from a human urine sample. In addition, machine-learning-based image recognition was shown to reduce the required time for AST and to provide results within 1 h for E. coli cells. Thus, the BactoSpin device can serve as an efficient and rapid platform for UTI diagnostics and AST.

Establishment of the Risk Prediction Model for Significant Bacteriuria in Adult Patients with Automated Urine Analysis

INTRODUCTION

Urinary tract infections (UTIs) have been proven to be the most encountered bacterial infection in humans. We hope to establish a prediction model for significant bacteriuria by comprehensively analyzing the relevant parameters of age, gender, and urine automatic analysis data.

METHODS

A retrospective study was performed at Tai'an Central Hospital. All samples included in the study were tested for urine culture and urine automatic analysis. Data analysis was conducted with the SPSS.

RESULTS

The binary logistic regression module is used to establish the forecast formula, which gender, age, leukocyte count, bacterial count, leukocyte esterase, and nitrite were included. Receiver operating characteristic (ROC) curves showed that the area under ROC curve (AUC) of the prediction model was 0.878, bigger than the AUCs of the other 6 independent variables. The sensitivity and specificity of prediction model were 61.68 and 95.98%, respectively. The positive and the negative predictive values of the predictive model are 87.13 and 85.02%, respectively.

CONCLUSIONS

The prediction formula obtained in our study can achieve good prediction effect for significant bacteriuria, which can effectively avoid the treatment delay or antibiotic abuse caused by the subjective judgment of doctors.

Prediction of urine culture results by automated urinalysis with digital flow morphology analysis

To investigate the association between the results of urinalysis and those of concurrent urine cultures, and to construct a prediction model for the results of urine culture. A total of 42,713 patients were included in this study. Patients were divided into two independent groups including training and test datasets. A novel prediction algorithm, designated the UTOPIA value, was constructed with the training dataset, based on an association between the results of urinalysis and those of concurrent urine culture. The diagnostic performance of the UTOPIA value was validated with the test dataset. Six variables were selected for the equation of the UTOPIA value: age of higher UTI risk [odds ratio (OR), 2.069125], female (OR, 1.400648), nitrite (per 1 grade; OR, 3.765457), leukocyte esterase (per 1 grade; OR, 1.701586), the number of WBCs (per 1 × 10⁶/L; OR, 1.000121), and the number of bacteria (per 1 × 10⁶/L; OR, 1.004195). The UTOPIA value exhibited an area under the curve value of 0.837 when validated with the independent test dataset. The UTOPIA value displayed good diagnostic performance for predicting urine culture results, which would help to reduce unnecessary culture. Different cutoffs can be used according to the clinical indication.

UriSed 3 PRO automated microscope in screening bacteriuria at region-wide laboratory organization

BACKGROUND AND AIMS

We assessed the possibility to rule out negative urine cultures by counting with UriSed 3 PRO (77 Elektronika, Hungary) at Helsinki and Uusimaa Hospital District.

MATERIALS AND METHODS

Bacteria counting of the UriSed 3 PRO automated microscope was verified with reference phase contrast microscopy against growth in culture. After acceptance into routine, results of bacteria and leukocyte counting from 56 426 specimens with eight UriSed 3 PRO instruments were compared against results from parallel samples cultured on chromogenic agar. Laboratory data including preanalytical details were accessed through the regional database of the Helsinki and Uusimaa Hospital District.

RESULTS

A combined sensitivity of 87-92% and a negative predictive value of 90-96% with a specificity of 54-50% was reached, depending on criteria. Preanalytical data (incubation time in bladder) combined with the way of urine collection would improve these figures if reliable.

CONCLUSIONS

Complex patient populations, regional logistics and data interfases, and economics related to increased costs of additional particle counts against costs of screening cultures of all samples, did not support adaptation of a screening process of urine cultures. This conclusion was made locally, and may not be valid elsewhere.

Evaluation of the detection ability of uropathogen morphology and vaginal contamination by the Atellica UAS800 automated urine microscopy analyzer and its effectiveness

Background

To help combat the worldwide spread of multidrug‐resistant Enterobacterales, which are responsible for many causes of urinary tract infection (UTI), we evaluated the ability of the Atellica UAS800 automated microscopy system, the only one offering the capability of bacterial morphological differentiation, to determine its effectiveness.

Methods

We examined 118 outpatient spot urine samples in which pyuria and bacteriuria were observed using flow cytometry (training set: 81; cross‐validation set: 37). The ability of the Atellica UAS800 to differentiate between bacilli and cocci was verified. To improve its ability, multiple logistic regression analysis was used to construct a prediction formula.

Results

This instrument's detection sensitivity was 106 CFU/ml, and reproducibility in that range was good, but data reliability for the number of cocci was low. Multiple logistic regression analysis with each explanatory variable (14 items from the Atellica UAS800, age and sex) showed the best prediction formula for discrimination of uropathogen morphology was a model with 5 explanatory variables: number of bacilli (p < 0.001), squamous epithelial cells (p = 0.004), age (p = 0.039), number of cocci (p = 0.107), and erythrocytes (p = 0.111). For a predicted cutoff value of 0.449, sensitivity was 0.879 and specificity was 0.854. In the cross‐validation set, sensitivity was 0.813 and specificity was 0.857.

Conclusions

The Atellica UAS800 could detect squamous epithelial cells, an indicator of vaginal contamination, with high sensitivity, which further improved performance. Simultaneous use of this probability prediction formula with urinalysis results may facilitate real‐time prediction of uropathogens and vaginal contamination, thus providing helpful information for empiric therapy.

Automated urinalysis combining physicochemical analysis, on-board centrifugation, and digital imaging in one system: A multicenter performance evaluation of the cobas 6500 urine work area

BACKGROUND

We evaluated the analytical performance of the fully automated cobas® 6500 urine work area and its automated components-cobas u 601 and cobas u 701.

DESIGN AND METHODS

The study was conducted at three European centers using un-centrifuged surplus routine urine samples; all measurements were performed within 2 h of sample collection. Precision, sample carry-over, and method comparisons were evaluated per Clinical and Laboratory Standards Institute guidelines. Method comparisons: cobas u 601 versus Urisys 2400 and cobas u 411 urine test strips; and cobas u 701 versus KOVA® visual microscopy and iQ200 analyzer. Operability and functionality were assessed using questionnaires.

RESULTS

Precision of the entire cobas 6500 system was within predefined acceptance limits and no significant carry-over was observed. Erythrocytes, leukocytes, nitrites, and protein were in good agreement (≥93%) with cobas u 411 reflectometry. High correlation was shown between the cobas u 701 analyzer and KOVA visual microscopy for red blood cells (RBC; slope, 0.89; Pearson's r, 0.95) and white blood cells (WBC; slope, 0.96; Pearson's r, 0.96), demonstrating equivalence of test results. The 97.5% percentile reference values on the cobas u 701 analyzer were 5.3 cells/μL (RBC) and 6.2 cells/μL (WBC). The cobas 6500 system showed good sensitivity for small bacteria (>1 μm) and pathological casts, and the user interface, maintenance wizards, and system design were highly rated by operators.

CONCLUSIONS

The fully automated workflow, high precision, and high throughput of the cobas 6500 system have the potential to facilitate standardization of urine screening.

Using artificial intelligence to reduce diagnostic workload without compromising detection of urinary tract infections

BACKGROUND

A substantial proportion of microbiological screening in diagnostic laboratories is due to suspected urinary tract infections (UTIs), yet approximately two thirds of urine samples typically yield negative culture results. By reducing the number of query samples to be cultured and enabling diagnostic services to concentrate on those in which there are true microbial infections, a significant improvement in efficiency of the service is possible.

METHODOLOGY

Screening process for urine samples prior to culture was modelled in a single clinical microbiology laboratory covering three hospitals and community services across Bristol and Bath, UK. Retrospective analysis of all urine microscopy, culture, and sensitivity reports over one year was used to compare two methods of classification: a heuristic model using a combination of white blood cell count and bacterial count, and a machine learning approach testing three algorithms (Random Forest, Neural Network, Extreme Gradient Boosting) whilst factoring in independent variables including demographics, historical urine culture results, and clinical details provided with the specimen.

RESULTS

A total of 212,554 urine reports were analysed. Initial findings demonstrated the potential for using machine learning algorithms, which outperformed the heuristic model in terms of relative workload reduction achieved at a classification sensitivity > 95%. Upon further analysis of classification sensitivity of subpopulations, we concluded that samples from pregnant patients and children (age 11 or younger) require independent evaluation. First the removal of pregnant patients and children from the classification process was investigated but this diminished the workload reduction achieved. The optimal solution was found to be three Extreme Gradient Boosting algorithms, trained independently for the classification of pregnant patients, children, and then all other patients. When combined, this system granted a relative workload reduction of 41% and a sensitivity of 95% for each of the stratified patient groups.

CONCLUSION

Based on the considerable time and cost savings achieved, without compromising the diagnostic performance, the heuristic model was successfully implemented in routine clinical practice in the diagnostic laboratory at Severn Pathology, Bristol. Our work shows the potential application of supervised machine learning models in improving service efficiency at a time when demand often surpasses resources of public healthcare providers.