背景:除臨床病史詢問之外，若經由尿化學、尿沉渣分析顯示陽性結果，臨床上常被疑似泌尿道感染。包括流式細胞儀等幾種自動尿沉渣分析儀，目前廣泛應用於單次尿液分析，這些超靈敏度尿液自動分析儀不可避免地有較高的陽性預測率，經由隨後尿液定量培養結果證實，高達85％的陽性篩檢結果，實則為陰性。這篇研究中我們企圖在流式細胞儀檢測參數中建立排除性的閾值，期能減少不必要的尿液培養，甚至避免經驗性抗生素處方的開立。
方法:由三軍總醫院門診及住院病人中，收集總共630個初步臆斷為泌尿道感染的尿液檢體。所有檢體均經由全自動化尿沉渣分析儀(Sysmex UF-1000i)分析，隨後即進行定量培養和細菌鑑定。流式細胞儀檢測參數包括白血球，細菌，前方散亂光強度(FSC)和螢光染色強度(FLH)，且經由建立接受者操作特徵曲線(ROC)及以(SPSS)統計分析軟體(第18版)的邏輯迴歸模組計算其敏感性和特異性後評估其預測正確性。為了預測泌尿道感染，我們尋求最佳參數及其最佳切點值。最佳參數定義為接受者操作特徵曲線下最大面積(AUC)。最佳切點值定義為接受者操作特徵曲線上最靠近點(0，1)的點，即偽陽性率為零，且敏感度度100％。
結果：研究結果發現，利用全自動尿沉渣分析儀篩檢泌尿道感染時，以儀器計數之bacteria/μL 較白血球(WBC/μL)為更佳之篩選項目。未使用抗生素患者之尿液檢體(N=458)中，當bacteria cut - off value＞190/μL 其敏感度＝95%、特異性＝64%、陰性預測率＝97%，相對可減少 42% 培養率；當WBC cut - off value＞20/μL 其敏感度＝96%、特異性＝33%、陰性預測率＝94%，相對可減少21%培養率。而使用抗生素患者檢體(N=172)中，當bacteria cut - off value＞12/μL其敏感度＝94%、特異性＝41%、陰性預測率＝94%，相對可減少 28% 培養率；當WBC cut - off value＞20/μL 其敏感度＝96%、特異性＝32%、陰性預測率＝95%，相對可減少22% 培養率。根據全自動尿液沉渣分析儀原理，發現在革蘭氏菌種判讀上細菌計數管路中FSC訊號較FLH 訊號具判讀性，但全自動尿沉渣分析儀主要是以篩檢為主，不作菌種判讀。結論: 對於未使用抗生素患者，將bacteria cut – off value設為190 /μL，大於此設定值認為是UTI而不作尿液定量培養，可降低尿液定量培養量42%。
對於使用抗生素患者，將bacteria cut – off value設為12 /μL，大於此設定值，則可認為UTI
尚未改善。有效降低不必要成本與工時支出，還可以作為泌尿道感染預後監控用。

Background: UTI is highly suspected when chemical and/or sediment analysis shown posi-
tive finding besides clinical history taking. Several automated urine sediment analyzers including
the flow cytometer have been applied for spot urine analysis. These ultra-sensitive urine
auto-analyzers inevitably have higher positive prediction rate. As high as to 85% of the positive
prediction turn out to be negative after confirmed by the subsequent urine quantitative culture. To
reduce unnecessary urine culture and provide negative prediction information for avoiding empiri-
cal antibiotics prescription, an attempt to set up exclusion thresholds for flow cytometry parameters
were performed in this study. Method: Total 630 urine specimens of UTI impression were collected
from both outpatient and inpatient departments in Tri-Service General Hospital. All of them were
analyzed with the automated urinary sediment analyzer (Sysmex UF-1000i). And each of them was
subsequently confirmed by the quantitative culture and bacterial identification. Flow cytometry pa-
rameters including white blood cell (WBC), bacterium, forward scattered light intensity (FSC) and
fluorescent light intensity-height (FLH) were evaluated for their predictive accuracy by constructing
receiver-operating characteristic (ROC) curves and calculating their sensitivities and specificities with the
logistic regression module of Statistical Package for Social Sciences (SPSS) Version 18. We sought the
best parameter and its optimal cut-off point for predicting UTI. The best parameter was defined as having the
largest area under curve (AUC) on the ROC curve. The optimum cut-off point was defined as the closest
point on the ROC curve to the point (0, 1) i.e., false positive rate of zero and sensitivity of 100%. Result:
Overall accuracy was expressed as area under the ROC curve (AUC) for comparing parameter
performance between urinary WBC and bacterial count. Our result demonstrated that bacterial
count (AUC: 0.881) is better than WBC count (AUC: 0.758) for predicting UTI in untreated patient
group, but similar AUC (bacterial count: 0.774, WBC: 0.793) was found in patients under antibiotics
treatment. In the ROC curve analysis, lower cut-offs are more likely to exclude patients from UTI, in
other words higher sensitivity and negative predictive value, giving a better “rule-out” test. To ob-
tain the highest negative prediction rate for ruling out UTI and reducing unnecessary urine cultures,
bacterial count cut-offs were set as 12 and 190/μl for the treated and non-treated specimens, re-
spectively. We can presumably reduce 42% urine culture of the non-treated UTI specimens and
28% of the non-treated UTI specimens. Another attempt to predict bacterial characteristics by FSC
and FLH did not show promising result. Conclusion: The sensitive automated urinary analyzer
provides predictive value to exclude UTI. Unnecessary urine culture and empirical antibiotic pre-
scription might be reduced by introducing the screening workflow in clinical practice.