Escherichia coli bacteremia : Clinical features , risk factors and clinical implication of antimicrobial resistance

Escherichia coli is an important cause of both community acquired (CA) and hospital acquired (HA) bacteremia. A prospective study was conducted at a tertiary care University Hospital from January, 2012 to July 2014, to compare the clinical features, risk factors, outcomes and antimicrobial resistance between E. coli bacteremia acquired from the community (CA) versus E. coli bacteremia acquired from the hospital (HA). Clinical and laboratory data of 171 adult patients with at least one positive blood culture of E. coli were analyzed. Data were collected from patients with significant blood stream infection, using medical and laboratory record files and information from treating medical staff. The overall incidence of extended spectrum beta lactamase (ESBL) infection was high, 67/171 (77.4%). Thirty-eight (40.9%) of the CA isolates were found to produce ESBL, while 28 (35.9%) of the HA isolates were ESBL producers. Patients with CA bacteremia tend to be older than those with HA bacteremia (0.003). Neoplastic diseases (hematological malignancy (<0.001), solid tumors (<0.001)), renal transplantation end stage renal disease (ESRD) (<0.006), and wound infection (<0.001) were the most commonly associated conditions in patients with HA bacteremia. Patients from the community are more likely to present with UTI (<0.001), fever and pyelonephritis (0.001). Both CA and HA E. coli isolates showed the highest sensitivity to imipenem, meropenem and amikacin followed by gentamicin and tazocin. The CA isolates are more susceptible to amikacin, tazocin and ciprofloxacin than the HA isolates. No significant difference in the mortality rate between patients with CA bacteremia and patients who acquire the bacteremia in a hospital setting (0.836) was observed. Clinicians need to be aware of the risk factors and changing pattern of antimicrobial resistance of this pathogen and should consider adequate empirical therapy with coverage of these pathogens for patients with risk factors


INTRODUCTION
Escherichia coli are part of the normal gastrointestinal flora and a leading cause of Gram negative bacteremia (Tenaillon et al., 2010).Sepsis and septic shock caused by E. coli and other Gram-negative bacteria is due to the inflammatory response activated by endotoxin (lipopolysaccharide) present in the Gram-negative cell wall (Johnson et al., 2006).Blood stream infection (BS I) in developing countries is a serious issue t hat is rarely addressed in the scientific literature (Aiken et al., 2011).Bloodstream infection (BSI) due to extended-spectrum βlactamase (ESBL) Enterobacteriaceae has emerged as a major cause of in-hospital mortality (Hyle et al., 2005;Pitout and Lauplan, 2008).The spread of communityacquired and hospital -acquired (nosocomial) bacteremia cause by E. coli imposes a major health burden.However, only few regional information is available on the differenc es between hospital-acquired and community acquired E. coli bacteremia (Hoenigl et al., 2014).Community and hospital spread of E. coli producing extended-spectrum beta-lactamases has increasingly been reported, most notably E. coli producing CTX-M strains (Woodford et al., 2004).This poses significant challenges to clinicians caring for patients presenting to hospital with suspected sepsis as empiric antibiotic treatment is often targeting pres umed, antibioticsusceptible community organisms (Rodriguez-Bano et al., 2006;Tumbarello et al., 2008).Accordingly, this study was conducted to assess any demographic variation in the incidence, the clinical characteristics, risk factors and antimicrobial-resistance trends of community-associated (CA) and hospital associated (HA) E. coli-bacteremia, presenting to t he hospital.To the best of the authors' knowledge, there are no other studies comparing the epidemiology and risk factors between the communityacquired and healt h-care associated E. coli bacteremia from the Gulf region.

Patients
This study w as conducted at King Khalid University Hospital, a 2500 bed major teaching hospital in Riyadh that provides both primary and tertiary medical care.From January 1, 2012 to July 30, 2014, adult patients (>14 years old) w ith at least one positive blood culture of ESBL-producing E. coli and non-ESBL-producing E. coli were review ed.Only the first episode of bacteremia in each patient w as included in the analysis.

Definitions and data collection
Data w ere prospectively collected from patients w ith significant blood stream infections using daily review of blood culture r esults, patients' medical record files, infor mation from treating medical staff and by a computerized method using the blood culture register numbers in the microbiology laboratory of each positive case.Standardized data forms w ere used to record demographic details including underlying diseases, hospital unit, and exposure to the healthcare system in the previous year, site of infection , ESBL production in organis ms isolated from culture samples, clinical progress and mortality.Patients w ere divided into tw o groups based Alotaibi and Bukhari 39 on the onset of bacteremia.Bacteremia w ith E. coli detected w ithin the first 48 h of hospitalization w as classi fied as "community-onset" according to the US Centers for Disease Co ntrol and Prevention definition and hospital acquired E coli infection w as defined as an infection that occurred > 48 h after admission to the hospital, or an infection that occurred < 48 h after admission of patients that had been transferred from another hospital or nursing home ( National Committee for Clinical Laboratory Standards, 1999) and w ere further classified into community-acquired or health care associated infections ( modified from the study of Siegman-Igra et al., 2002).
The for mer definition repr esents truly community-acquired infection, while the latter consists of infections in patients recently discharged (≤ 6 months), infections associated w ith invas ive procedures performed ear lier, or at the time of admission and infections in patients admitted from nursing homes.E. coli bacteremia w as defined as the isolation of E. coli from ≥ 1 set of aseptically inoculated blood culture bottles.In patients w ith clinical features compatible w ith systemic inflammatory response syndrome.Patients w ere classi fied as immunosuppressed if neutropenia (defined as < 1,000 poly morphonuclear neutrophils cells/mm 3 ), hematologic malignancy, corticosteroid therapy (equivalent to > 20 mg prednisolone/day) for at least 2 w eeks, and/or cancer chemotherapy or radiation therapy w ere documented w ithin 30 days of the onset of bacteremia.Patients w ith serum creatinine level > 3 mg/dL, or under dialysis, before the onset of bacteremia w ere considered to have chronic renal insufficiency.

Identification and antim icrobial susceptibility testing
Isolates of E. coli w ere identified by standard microbiologic methods in the microbiology laboratory using an automated identification system (Vitek System; bioMe´rieux).Susceptibilities to antimicrobial agents (ampicillin, amox icillin/clavulnate, cefradine, cefuroxime, ceftriaxone, cefotaxime, ceftazidime, cefipem, ciprofloxac in, imipenem, mer openem, gentamicin, amikac in piperacillin/ tazobactam, trimethoprim/sulfmethoxazole) w ere deter mined by use of an automated susceptibility testing system (Vitek 2 System; bioMe´rieux).ESBL production w as detected and interpreted using CLSI criter ia for broth dilution in accordance w ith the Clinical and Laboratory Standards Institute standards (Wayne, 2005).

Statistical analysis
All statistical analyses w ere performed using the SA S softw are package (version 9.1; SAS Institute Inc., Cary, NC, USA).For univariate analysis, categorical variables w ere compared using χ 2 or Fisher's exact test and continuous variables w ere analyzed w ith Student's t test or Mann-Whitney U test.A p value < 0.05 w as considered to be statistically significant, and all probabilities w ere tw o-tailed.
Author(s) agree that this article remains permanently open access under the terms of the Creativ e Commons Attribution License 4.0 International License  2. Patients with community acquired E coli bacteremia tend to be older than those with hospit alacquired infection (0.003); they were more than 55 years old and were mostly female.Hematological malignancy (<0.001), solid tumors (<0.001), renal transplantation, end stage renal disease (ESRD) (<0.0 06), and wound infection including diabetic foot infection (< 0.001) were associated with hospitalization and development of E. coli bacteremia.
Among patients with malignancy, hematological malignancy was found to be a significant risk factor for acquisition of E coli bacteremia in hospitalized patients (14.1%) (<0.001).Patients from t he community are more likely to present with urinary tract infection (< 0.001), fever and pyelonephritis (0.001) or vomiting and diarrhea (0.039).Among the 78 hospit alized patients, oncology (30.8%), medicine (28.2% ), and critical care (23.1% ), were the commonest specialists at the onset of bacteremia (Table 3).The overall incidence of ESBL infection was high, 67/171 (77.4% ).Thirty-eight (40.9% ) of the community acquired isolat es were found to produce ESBL, while 28 (35.9%) of the hospital acquired isolates were ESBL producers.There was no significant differenc e in acquiring infection with ESBL E. coli between patients from the community and hospit alized patients.Both community-acquired and hospital -acquired E. coli isolates showed the highest sensitivity to imipenem, meropenem and amikacin followed by gentamicin and piperacillin/tazobactam (Figure 1).The sensitivity pattern of ESBL producing E. coli of the community-acquired and hospital -acquired isolates is shown in Figure 2. Meropenem and imipenem are t he most sensitive antimicrobial agents followed by the amikacin and piperacillin/tazobactam.The communityacquired isolates are more susceptible t o amikacin, piperacillin/tazobactam and ciprofloxacin than t he hospital-acquired isolates.No significant difference was observed in the mortality rat e bet ween patient who acquire the bacteremia from the community or those who acquire the bacteremia in a hospital setting (0.836)  (Table 4).

DISCUSSION
E. coli-blood stream infection is a major cause of morbidity and mortality with a relatively high associated population burden (Pitout et al., 2004;Uslan et al., 2007;Williamson et al., 2013).Little data exists on the demographic variation and potential risk factors between CA and HA E. coli blood stream infection (Pitout et al., 2004;Rodríguez-B año et al., 2010) such populationbased demographic information is important in implementing strategies for treatment and prevention of these serious infections.There were many studies from the Saudi Arabia region that determine the prevalence of bacterial pathogens isolated from all specimen types including blood and assessed t he multi-drug resistant rates of ESBLs among Enterobacteriaceae.The prevalence between 4.8 and 15.8% have been reported from Saudi Arabia with the finding of the lowest frequency rates of ESBL produc ers in t he eastern region and t he highest frequency was observed in the central region (El-Khizzi and Bakheshwain, 2006;Kader and Kumar, 2004;Masoud et al., 2011;Rodríguez-B ano et al., 2009).In t wo studies (El-Khizzi and Bakheshwain, 2006;Khanfar et al, 2009) from the Arabian Gulf region, ESBL detection in Enterobacteriaceae was described.In the first study (El-Khizzi and Bakheshwain, 2006), different patient populations with nosoc omial and community-acquired infections were assessed, the majority (83%) of the ESBL-producing isolates were E. coli.ESBL producers were significantly higher among isolates from in-patients, 15.4% as compared to those from out-patients, 4.5%.Urine was the most common specimen for the isolation of ESBL pathogens among in-patients and out -patients.In the second study from Qatar, Khan et al., 2010 report ed the occurrence of resistant Gram-negative organisms in 63.1% of bacteremia patients with the following prevalence: ESBL-producing p0k i9 (34%), followed by Klebsiella spp.(13.7%) and finally Pseudomonas aeruginosa (7.4%).A recent study on the characteristics of hospital-acquired and community-onset blood stream infections from Austria (Hoenigl et al., 2014), E. coli followed by Staphylococcus aureus were the most frequently isolated pathogens.This study has shown that, ESBL producing E. coli is an important cause of bloodstream infection presenting from bot h, the community and the hospit al settings (40.9 and 35.9%, respectively).The overall incidence of E. coli ESBL bacteremia in this study is high, higher than the rate reported by Memom et al., 2009 andKang et al., 2013, from the eastern region of Saudi Arabia (31%), and from Korea (33% ), respectively.In another retrospective study from Taiwan [6], of 404 episodes of community-onset E. coli bacteremia, the frequency of ESBL produc ers was 4.7%.This rate is considerably lower than the rate found in our study.The differences in risk factors between CA and HA bacteremia was als o identified.Patients with community acquired E coli bacteremia tend to be older than those with hospital-acquired infection and are mostly females.This finding is in agreement with a populationbased incidence and comparative study (Williams on et al., 20113) of community-associated and healt hcareassociated E. coli bloodstream infection from New Zealand, which revealed that, the incidence of E. coli bacteremia was highest in the under one year and over 56 year-old age groups.Previous population -bas ed studies have documented the association of all bloodstream infections with old age (Hyle et al., 2005;Johnson et al., 2006).Uslan et al., 2007 identified an increased risk of E. coli bacteremia in females across all age ranges which contrasts the finding of an increase risk in only those above 55-year-old of age.In contrast, Kang et al., 2013 found that elderly males were at highest risk.
The study showed that, solid tumors (19.2%), hematological malignancy (14.1% ) and end-stage renal disease/post renal transplant (12.8%), are t he most common underlying diseases and were identified as significant risk factors for health-care associated E. coli bacteremia.
Comparably, Kang et al., 2013 has found that, solid tumors, diabetes mellitus and liver diseases were the most common underlying diseases and predisposing factors for community onset bacteremia caused by ESBL producing E. coli.In a case controlled study from Spain of 96 patients with nosocomial blood stream infections (BS I) due to ESBL producing E. coli, the risk factors were found t o be organ t rans plant, previous use of oxyimino-βlactams, unknown BSI source and duration of hospital stay (Rodrguez-Bano et al., 2008).In addition, a population-based s urveillance involving a total of 2368 episodes of E. coli bacteremia conducted in the Calgary Healt h Region has found that, the very young and the elderly were at highest risk for E. coli bacteremia.Additionally, dialysis, solid organ transplantation and neoplastic disease were identified to be the most important risk factors for acquiring E. coli bacteremia (Laupland et al., 2008).Among the 422 patients with neoplastic disease, 270 (64%) had malignant tumors, 96 (23% ) had hematological malignancies, one patient had both a tumor and a hematological malignancy, and 55 (13%) patients had neoplastic disease in remission.
In anot her study by Chen et al., 2010, on t he epidemiology of bloodstream infections in patients with haematological malignancies with and without neutronpenia, the authors found that E. coli (12%) predominat ed the Gram-negative isolates causing BSI in neutropenic patients (Chen et al., 2010).Over the past two decades, treatment of E. coli bacteremia has become increasingly complicated by the emergence of antimicrobial -resistant E. coli strains, particularly those strains possessing acquired resistance genes encoding extended-spectrum beta-lactamases (ESBLs) and carbapenemases.Bloodstream infections with these resistant organisms have been associated with adverse clinical cons equences and significant therapeutic challenge to t reating physicians.The initiation of an antimicrobial agent is usually empirical, requiring knowledge of the likely pathogen and usual antimicrobial susceptibility patterns.This work has highlighted concerning trends towards greater antimicrobial resistance in E. coli causing bacteremia.However, in this study, both communityacquired and hospital -acquired E. coli isolates showed the highest sensitivity to carbepenem and amikacin followed by gentamicin and taz ocin.The communityacquired isolates are more susceptible t o amikacin, tazocin and ciprofloxacin than the hospital -acquired isolates.Similar to this study, Khanfar et al., 2009 found in his study, none of the strains isolated were resistant to carbapenems.In addition, recent studies showed that previous use of oxyimino-β-lactams or fluoroquinolones is a risk factor for ESB L-producing isolates in patients with bacteremia caused by E. coli (Quirante et al., 2011;Rodrguez-B ano et al., 2010).A retrospective cohort analysis (Rodriguez-B ano et al., 2006) has shown that, when compared with β-lactam/β-lactamase-inhibitor and carbapenem -bas ed regimens, empirical therapy of ESB Lproducing E. coli bacteremia with cephalosporins or fluoroquinolones were associated with a higher mortality rate.Resistance to drugs other than penicillins and cephalosporins was associated with increased mortality (Rodrguez-Bano et al., 2010).The mortality rat e in this study (25%) is higher than previously (11.4 %) reported [6].In a recent prospective cohort studies, carried out in hospitals from 31 countries that participated in the European Antimicrobial Resistance Surveillanc e System (EARSS), excess mortality associated with BSIs caused by MRSA and third-generation cephalosporin-resistant E. coli (G3CREC) is significant, and the prolongation of hospital stay imposes a considerable burden on health care systems.
These studies are essential to assist with the challenges of empiric antibiotic prescribed for those presenting to hospitals with suspected sepsis.As both communityacquired and hospital -acquired E coli isolates showed the highest sensitivity to imipenem, meropenem, in this study, it is believed in view of their excellent in vit ro activity, carbapenems along with amikacin should be the initial empiric choice for serious life threatening infections caused by ESBL producing Enterobacteriaceae, with prompt de-escalation when culture and susceptibility results become available.In this study, there was no significant difference in the mortality rate bet ween community and nosocomial bacteremia.Identification of risk factors for MDR organisms in patients presenting from the community with sepsis is necessary to help optimize patient outcomes and minimize the use of broad-spectrum antibiotics.To the authors' knowledge, this is the first report presenting data differentiating between nosocomial and community acquired ESBL E. coli bacteremia in Saudi Arabia.Continued surveillance, appropriate use of antibiotics and implementation of strict infection control measures are recommended to reduce ESBL frequency.

Figure 2 .
Figure 2. Percentage sensitivity of community acquired ( CA) and hospital acquired ( HA) ESBL produc ing E. coli.
IMP GM AK AMC AMP CRD FEP CIP CTX FOX CAZ CRO CXM

Table 1 .
Classification of 171 patients w ith E. coli bacteremia.

Table 2 .
Clinical characteristics of 171 patients w ith E. coli bacteremia.

Table 3 .
Admission character istic of 78 hospitalized patient w ith E. coli bacteremia.

Table 4 .
Mortality among ESBL and non-ESBL patients.
Figure 1.Percentage sensitivity of community acquired ( CA) and hospital acquir ed ( HA) E. coli to antimicrobial agents.