Trang Vo Anh Vinh, Truyen Thien Tan Tri Tai, Nguyen Minh Thuan, Mai Huu Phong, Phan Tri Cuong, Phan Son Hoang, Le Nguyen Han My, Nguyen Huong-Dung Thi, Le Nguyen Hai Dang, Tu Man Nhi, Huynh Vo Thanh Vi, Nguyen Hoang Tram Anh, Ho Dac Bao Han, Tran Ngoc Thuy Uyen, Tran Nguyen Ha Uyen, Le Bich-Nhat Thi, Doan Duc Tuan, Pham Huu Doan, Phan Truong Bao, Pham Phu Phat, Nguyen Tuan Vinh, Nguyen Phuc Cam Hoang
Binh Dan Hospital, Ho Chi Minh, Vietnam.
Faculty of Medicine, Pham Ngoc Thach University of Medicine, Ho Chi Minh, Vietnam.
PLoS One. 2025 May 14;20(5):e0323664. doi: 10.1371/journal.pone.0323664. eCollection 2025.
Urinary tract infections (UTIs) are common globally, and are developing increased antibiotic resistance. Despite being the diagnostic "gold standard," urine culture is limited by slow results and a high rate of false negative findings, leading to treatment delays, higher costs, and overuse of empirical antibiotics. Our study aims to develop a rapid and reliable model to predict clinical outcomes.
From January 1st to October 31st, 2023, we enrolled patients with symptoms suggesting UTI from the Outpatient Department of our hospital. Inclusion criteria were patients aged ≥18, initially diagnosed with UTI, available urinalysis, flow cytometry, and urinary culture. Exclusion criteria included failed sample collection and cultures, and pregnant women. A case-control study was conducted, with UTI cases defined as ≥ 10^5 CFU/ µ L and controls as < 10^5 CFU/ µ L, matched for age and sex in a 1:1 ratio. For validation, retrospective cases from July to December 2022 were selected with matching controls. Using urine culture as the gold standard, the predictive model was developed with backward stepwise logistic regression. Model discrimination was assessed using area under the curve (AUC).
In our discovery cohort, we included 1,335 UTI cases and 1,282 non-UTI controls, with mean ages of 52.9 ± 17.1 years and 51.9 ± 16.4 years, and females of 76.9% and 77.7%. Using 100 cells/uL as a threshold, bacterial counts demonstrated a sensitivity of 91.0% and specificity of 45.7%. Our novel UTIRisk score, developed from urinalysis and flow cytometry parameters, showed strong discrimination for UTI, with a AUC of 0.82 (95% CI: 0.81-0.84). In the validation cohort, the AUC was 0.77 (95% CI: 0.74-0.80). The UTIRisk score exhibited excellent specificity (96.5%) and high positive predictive value (92.6%). The score performed strongly across subgroups, particularly in males and patients aged ≥65.
Our UTIRisk score can improve diagnosis, reduce unnecessary urine cultures, optimize antibiotic use, and help control antibiotic resistance in LMICs. Multicenter, and intervention-based studies are warranted before clinical implementation.
尿路感染（UTIs）在全球范围内都很常见，并且抗生素耐药性正在增加。尽管尿培养是诊断的“金标准”，但其结果出得慢且假阴性率高，导致治疗延迟、成本增加以及经验性抗生素的过度使用。我们的研究旨在开发一种快速且可靠的模型来预测临床结果。
2023年1月1日至10月31日，我们从我院门诊部招募了有尿路感染症状的患者。纳入标准为年龄≥18岁、最初诊断为UTI、可进行尿液分析、流式细胞术和尿培养的患者。排除标准包括样本采集和培养失败以及孕妇。进行了一项病例对照研究，UTI病例定义为≥10^5 CFU/µL，对照定义为<10^5 CFU/µL，年龄和性别按1:1比例匹配。为进行验证，选取了2022年7月至12月的回顾性病例并匹配对照。以尿培养作为金标准，采用向后逐步逻辑回归建立预测模型。使用曲线下面积（AUC）评估模型的辨别力。
在我们的发现队列中，我们纳入了1335例UTI病例和1282例非UTI对照，平均年龄分别为52.9±17.1岁和51.9±16.4岁，女性分别占76.9%和77.7%。以100个细胞/微升作为阈值，细菌计数的敏感性为91.0%，特异性为45.7%。我们从尿液分析和流式细胞术参数开发的新型UTIRisk评分对UTI显示出很强的辨别力，AUC为0.82（95%CI：0.81 - 0.84）。在验证队列中，AUC为0.77（95%CI：0.74 - 0.80）。UTIRisk评分表现出优异的特异性（96.5%）和高阳性预测值（92.6%）。该评分在各亚组中表现良好，尤其是在男性和≥65岁的患者中。
我们的UTIRisk评分可以改善诊断、减少不必要的尿培养、优化抗生素使用，并有助于在低收入和中等收入国家控制抗生素耐药性。在临床实施之前，有必要进行多中心和基于干预措施的研究。
