帶有廣效乙內醯胺酶大腸桿菌菌血症之 感染危險因子及死亡危險因子分析：著重於藥物治療

Abstract

目的： 此病例對照分析研究主要是為了了解感染廣效乙內醯胺酶（extended-spectrum β-lactamase；ESBL）大腸桿菌（Escherichia coli；E. coli）菌血症的危險因子，另外針對經驗性抗生素恰當與否及不同抗生素治療對於臨床治療效果作進一步的評估，最後找出影響病人死亡的危險因子以提供醫療人員在臨床上選擇抗生素的參考。 設計： 自2005年1月1日至2007年6月30日為期30個月的單醫學中心回溯性病例對照分析研究，以病歷回顧方式進行。 地點： 國立台灣大學醫學院附設醫院—位於台灣北部的一間擁有2200張病床的教學醫院。 對象： 病例組及對照組的病人都是經由台大醫院細菌室報告至少有一套血液培養長出大腸桿菌且伴隨有發燒及/或其它感染徵兆之16歲以上患者。將感染ESBL E. coli菌血症的病人定為病例組，根據檢體採檢日期最接近的日期以1:1比例配對選取出感染non-ESBL E. coli菌血症的病人定為對照組。每一位病人只納入其第一次的感染。 方法： 利用查閱病人紙本病歷及電子病歷的方法記錄病人的基本資料、合併症、和這次住院相關的資訊、過去住院或感染大腸桿菌的病史、在菌血症發作前抗生素的使用情況、感染菌血症時的臨床表徵、感染時所使用的的抗生素治療、病人的預後等等。本研究的觀察終點為菌血症發作第30天的死亡率。 在危險因子和預後的分析主要利用單變項分析及多變項羅吉斯迴歸分析。在存活方面利用Kaplan-Meier法作存活分析圖及Log-rank法去比較。 結果： 研究期間共有97位感染ESBL E. coli菌血症的病人，有五位病人因為追蹤不完全，另外一位病人因為無法借閱病歷而被排除。同時配對91位感染non-ESBL E. coli菌血症病人。在病例組的91位病人中有71位病人屬於單一菌種感染（78%），20位病人是多重菌種感染（22%）。兩組的第30天死亡率分別為病例組28.6%、對照組9.9% (p = 0.001)。在多變項分析中發現感染當時有留置尿管的病人（勝算比為6.21，p = 0.003）、發生菌血症前30天內曾使用過第三代或第四代cephalosporins（勝算比為5.16，p = 0.046）、以及發生菌血症前30天內使用過抗生素（勝算比為2.93，p = 0.021）為感染ESBL E. coli菌血症的獨立危險因子；反之，社區型感染（勝算比為0.22，p = 0.002）為感染ESBL E. coli菌血症的保護因子。在治療方面，病例組病人中給予經驗性抗生素治療恰當與否對第30天的死亡率並沒有顯著差異（p = 0.382），即使是分析單一菌種感染的病人也得到相同的結果（p = 0.233）。若分析確切治療中carbapenem組及non-carbapenem組兩組的第30天死亡率也沒有顯著的差異（p = 0.703）。最後，對於死亡危險因子的分析可發現在全體病例病人中，若病人是在加護病房得到感染（勝算比為7.47，p = 0.006）、本身是免疫抑制的病人（勝算比為3.82，p = 0.018）、原發性菌血症感染的病人（勝算比為3.15，p = 0.045）都是影響死亡的危險因子。分析單一菌種感染的病人發現病人若感染時出現敗血性休克的情況（勝算比為7.01，p = 0.003）是一危險因子，而在加護病房得到感染（勝算比為6.92，p = 0.017）依然是一獨立危險因子。 結論： 當病人過去使用過抗生素，尤其是廣效的cephalosporin（第三代及第四代）及感染當時有留置尿管的病人，都要小心ESBL E. coli菌血症發生的可能性；反之，如果是社區型的感染則發生ESBL E. coli菌血症的機會會比較低。在本研究中對於感染ESBL E. coli菌血症的病人，一開始給予恰不恰當的經驗性抗生素並不會明顯影響預後，在嚴重的病人中還是以carbapenem為首選藥物。如果病人本身免疫功能不全、原發性菌血症感染的病人、感染當時病人有出現敗血性休克的情況及在加護病房感染的病人往往都會是死亡率較高的危險族群。 關鍵詞： 廣效性乙內醯胺酶、大腸桿菌、菌血症、危險因子、抗藥性、病例對照研究、治療效果、死亡率

Objective： A case-control study was conducted in order to identify the risk factors associated with bloodstream infection caused by extended-spectrum-β- lactamase (ESBL) producing Escherichia coli (E. coli). Secondly, to evaluate the impact of different antibiotics and adequate empirical treatment on clinical outcomes. And also to find out the risk factors for mortality of ESBL-producing E. coli bacteremia in order to provide a concept for clinicians to choose appropriate antibiotic therapy. Design： A retrospective case-control chart-review study was conducted during a 30-month period between Jaunary 2005 and June 2007. Setting： National Taiwan University Hospital---a 2200-bed teaching hospital in northern Taiwan Patients： Both case and control patients were identified by systemically reviewing the results from the microbiology laboratory. “Case patients” were those who had ESBL-positive E. coli bactermia. One case patient was matched with one control patient according to the closest date to isolation of ESBL-positive E. coli. “Control patients” were those who had ESBL-negitive E. coli bacteremia. We only included the first episode of each patient. Methods： Data were collected from medical records and hospital computerized databases. The records for each patient including patients’ profiles, underlying diseases and comorbidities, the informations about this admission, previous hospitalization histories, previous E. coli infection or colonization histories, antibiotics exposure before bacteremia onset, clinical presentations when bacteremia onset, antibiotic regimens during treatment period, clinical outcomes after bacteremia onset. The primary endpoint was 30-day mortality. Risk factors and clinical outcomes were examed using univariate analysis and muitiple logistic regression analysis. Survival curves shown by Kaplan-Meier method were compared with Log-rank test. Results： During the study period, there were 97 patients of ESBL E. coli bacteremia, 5 patients were excluded because of incomplete follow up and 1 patient with missing chart. Then 91 control patients were matched. In case group, there were 71 (78.0%) monomicrobial infections and 20 polymicrobial infections. The 30 day mortality was 28.6% (26/91) in the case group, compared to 9.9% (9/91) in the control group (p = 0.001).By multivariate analysis, urinary catheterization [odds ratio(OR)6.21, p = 0.003], previous treatment with 3rd、4th-cephalosporins [odds ratio(OR)5.16, p = 0.046], prior exposure to antibiotics [odds ratio(OR)2.93, p = 0.021] were independent predictors for ESBL production. On the contrary, community-acquired infection [odds ratio(OR)0.22, p = 0.002] was a protective factor for ESBL production. Analyses of treatment outcomes showed that there was no significant difference between patients received adequate or inadequate empirical therapy for case patients (p = 0.382) and monomicrobial infection patients (p = 0.233). Comparison between carbapenem and non-carbapenem treatment groups also showed no significant influence on the 30-day mortality (p = 0.703). Multivariate logistic regression analysis for all case patients showed that ICU-acquired [odds ratio(OR)7.47, p = 0.006], immunocompromised patients [odds ratio(OR)3.82, p = 0.018], primary bacteremia [odds ratio(OR)3.15, p = 0.045] were independent risk factors for 30-day mortality. The results from monomicrobial infection patients showed that septic shock [odds ratio(OR)7.01, p = 0.003], ICU-acquired [odds ratio(OR)6.92, p = 0.017] were risk factor for 30-day mortality. Conclusions： Our results showed that prior exposure antibiotics, especially 3rd or 4th- cephalosporins, and urinary catheterization were independent predictors for ESBL production. On the contrary, community-acquired infection was a predictor for non-ESBL E. coli bacteremia. Appropriateness of empirical therapy and regimens of definitive therapy did not significantly influence mortality on day 30 in our study. Carbapenem is still the drug of choice for severe patients. Immunocompromised patients, primary bacteremia, septic shock and ICU-acquired infection had a significant impact on day 30 mortality. Keywords： extended-spectrum-β- lactamase(ESBL), Escherichia coli(E. coli), bacteremia, risk factor, resistant, case-control study, treatment outcome, mortality