Enterococci , Clostridium difficile and ESBL-producing bacteria : epidemiology , clinical impact and prevention in ICU patients

Most hospital-acquired infections arise from colonising bacteria. Intensive care patients and immunocompromised individuals are at highest risk for microbial invasion and subsequent infection due to multiple invasive procedures in addition to frequent application of chemotherapeutics and presence of poor microperfusion leading to mucosal disruption. In this narrative review, we summarise the literature on bacterial colonisation in intensive care patients, in particular the epidemiology, the clinical impact and respective infection control strategies of three pathogens, i.e., Enterococcus spp., extended-spectrum ß-lactamase producing gram-negative bacteria and Clostridium difficile, which have evolved from commensals to a public health concern today.


Introduction
Infections are the leading cause of death in intensive care units (ICUs) worldwide and mortality in infected ICU patients is more than twice as high compared to non-infected patients [1,2].Despite significant advances in intensive care therapy and infection prevention, incidence of nosocomial infections in ICU patients has remained high [1,3].The bacteria causing most hospital-acquired infections are staphylococci including methicillin-resistant S. aureus (MRSA), enterococci including vancomycin-resistant en- terococci (VRE), Candida spp., Clostridium difficile and different often multidrug-resistant gram-negative bacteria [1].In healthy individuals, an ecological community of commensals, symbiotes and pathogens -the microbiome -is in equilibrium with the host.If anatomical barriers or host defenses are disrupted, invasion of colonising bacteria and subsequent infection can arise [4].In ICU patients, multiple invasive procedures (e.g., central venous catheters) and the presence of poor microperfusion lead to integrity loss of skin and mucosae with risk of invasive infection [5].Furthermore, ICU patients are per se immunocompromised due to the severity of the disease [6].Selected by pressure of antibiotic treatments, colonising multidrug-resistant bacteria can outgrow commensals from the microbiome [7] and become invasive.In view of the global rise of infections with multidrug-resistant bacteria and a concomitant lean development pipeline for antimicrobial agents, the "Infectious Diseases Society of America" stated that we should consequently maximise infection control strategies [8].In this narrative review, we summarise the literature on bacterial colonisation in ICU patients, in particular the epidemiology, the clinical impact and respective infection control strategies focusing on three intestinal bacteria, i.e.Enterococcus spp., extended-spectrum ß-lactamase producing gram-negative bacteria (ESBL-GNB) and Clostridium difficile.

Methods
We searched PubMed/MEDLINE in November 2013, without restrictions, using the following search strategy with Boolean operators: "("enterococcus" OR "enterococci" OR "VRE" OR "ARE" OR "ESBL" OR "Clostridium difficile" OR "C.difficile") AND ("intensive care unit" OR "ICU")".In addition, we searched the references of cited articles in this review for other appropriate studies.Only articles focusing on adult populations (≥18 years) written in English, German, French or Italian language were included.Several other search terms were applied to identify appropriate studies regarding specific questions considered in this narrative review (e.g., to describe the global epidemiology of enterococci).

Enterococci
Background E. faecalis and E. faecium -the species most frequently encountered in clinical isolates [9] -have evolved from intestinal commensals to the third highest ranking cause of nosocomial infections in the United States [10].Enterococci are characterised by a remarkable genomic flexibility [11] with the ability to incorporate foreign mobile genetic elements carrying e.g., resistance genes to multiple antibiotics in addition to chromosomal resistance genes [12,13].The increasing rate of antimicrobial resistance in enterococci, e.g. to ampicillin or vancomycin, are of major clinical importance [14,15].Since the first description of VRE in a clinical isolate in Europe in 1988 [16], VRE are increasing in prevalence worldwide, capable of spreading vancomycin resistance genes (mainly vanA and vanB) via transposons to vancomycin-susceptible enterococci and rarely to other bacteria (e.g., MRSA) [17,18].VanA is widely prevalent in the United States and Korea, whereas vanB has been introduced as main genotype in VRE epidemics in Australia and Singapore [19].Chromosomal vancomycin resistance genes are less transmissible (e.g., vanC) and are related to the use of the animal growth promoter avoparcin in Europe until 1997 and rarely cause infection [20,21].Whereas in the Unites States VRE nowadays dominate the epidemiology of nosocomial enterococcal infections, the situation in Europe is more diverse: Germany, Greece, England, Ireland and Portugal have high VRE rates of >10%, whereas in most European countries an increase in ampicillin-resistant enterococci (ARE) is observed since the year 2000 [22,23].According to the European Center for Disease Prevention and Control (ECDC) report 2012, overall ampicillin-susceptibility rates of E. faecalis isolates were >75% in all European countries (mostly >95%) compared to <50% in E. faecium (range 0.8-33.3%)[22].

Background
One of the most important resistance mechanisms of gramnegative bacteria are ß-lactamases conferring resistance to ß-lactam antibiotics by hydrolisation of their ß-lactam-ring [68].Of the many different ß-lactamases, ESBLs comprise the largest group of enzymes [68], causing resistance to newer ß-lactam antibiotics, including the third-generation cephalosporins and monobactams, but not the cephamycins and carbapenems [69,70].
ESBLs were initially recognised in clinical bacterial isolates in the 1980's and are a rapidly increasing public health threat today [68,71].[77].In a recent Swiss hospital-wide surveillance study of patients with any clinical ESBL-GNB isolate and no current ESBL-GNB specific antimicrobial therapy, urine samples were positive in 110 of 133 patients (82.7%), rectal swabs in 69.2%, skin swabs of the groin in 35.3%, and throat swabs in 12.8% [78].

Clinical impact of infection and colonisation with ESBL-GNB in ICU patients
Global surveillance data from a 1-day point prevalence study on 1,265 ICUs in 2007 showed an overall ESBL rate of 3.0% among clinical isolates of gram-negative bacteria (North America 0.4%, Western Europe 3.0%, Asia 4.5%) [1].The published ESBL-GNB colonisation rates of ICU patients (table 2) range from 2.2% [79] to 49.0% [80] with important geographical differences.The highest ESBL-GNB colonisation rates on ICU admission have been found in Korean [81] (42.5%),Indian [80] (49.0%), and Spanish [82] (38.3%)ICUs, whereas especially ICUs from the Unites States [79,83,84] and Belgium [85] exhibited low colonisation rates (2.2-6.2%).The three main risk factors for colonisation/infection with ESBL-GNB in ICU patients are length of hospital stay [80,86], high ESBL-GNB colonisation pressure [86,87], and broad-spectrum antibiotics [79,80,82,[86][87][88] (table 2).ESBL-GNB infection rates in colonised ICU patients range from 4.9% [87] to 68.8% [85].The largest of these studies was a prospective 3.5-year single-centre study from the Unites States.Out of 5,209 ICU patients, 2.2% were rectally colonised with ESBL-producing E. coli or Klebsiella spp. on admission, and in 24.8% the same ESBL-GNB was found in a clinical sample thereafter [79].In contrast, among the 5,092 patients not colonised with ESBL-GNB, only 0.6% had a subsequent positive clinical culture [79].One of the few prospective studies on outcome of rectal ESBL-GNB colonisation was performed in 513 hospitalised haematological and oncological patients in Germany.Colonised patients had a risk ratio of 4.5 to develop a subsequent ESBL-GNB bloodstream infection (95% CI 2.9-7.0)[89].A 10-year prospective French study including 710 liver transplant patients showed an even higher infection rate in patients with pre-transplant fecal ESBL-GNB colonisation (44 1 solid organ transplant unit, and 1 haematology/oncology unit) in the Unites States found an ESBL-GNB bloodstream infection rate of 8.5% (35/413) in colonised patients [90].On the other hand, one study in patients with acute leukaemia or haematopoietic stem cell transplantation could not confirm an association between colonisation and infection with ESBL-producing E. coli or an increased inhospital mortality (bloodstream infections rate with ESBLproducing E. coli in 1.5% of colonised vs 1% of non-colonised patients; p = 0.7) [91].Several, mostly retrospective studies showed significantly longer length of hospital stay and higher mortality rates in patients with bloodstream infections due to ESBL-producing versus non-producing GNB [90,[92][93][94][95][96][97][98].The increased mortality in bloodstream infections with ESBL-GNB is mainly caused by inadequate initial therapy and is likely not a consequence of higher bacterial virulence [99].

Clostridium difficile
Background C. difficile is a gram-positive, anaerobic and spore-forming rod causing mainly antibiotic-associated diarrhea.Symptoms range from uncomplicated diarrhea to severe pseudomembranous colitis and toxic megacolon [100].C. difficile is a public health concern worldwide representing the leading cause of hospital-associated infectious diarrhea [101].Increases in incidence, morbidity, and recurrence rate have been reported in the United States, Canada, and Europe [102].In contrast, little is known on the epidemiology of C. difficile infection in Asian countries [103].The increased virulence of C. difficile has been attributed to the spread of fluoroquinolone-resistant ribotype 027 (RT027, BI/NAP01), which produces, in addition to toxins A and B, a binary toxin of unspecified significance [102,[104][105][106]. C. difficile RT027 is the cause of multiple healthcare-associated outbreaks in the United States, Canada, and Europe [107][108][109].Furthermore, communityacquired C. difficile infection is increasing, with another hypervirulent ribotype (078), as the main culprit.Compared to North America and Europe, ribotype 017 and 018 have been shown to be the most prevalent types in Asian hospitals [103].C. difficile is thought to be mainly transmitted via hands of healthcare workers and by the contaminated environment [110].Hand hygiene with alcoholic solutions is not associated with a higher risk of transmission despite the fact that alcohol does not have any antimicrobial effect against C. difficile [111].Healthcare workers could be asymptomatic intestinal C. difficile carriers acting as a reservoir for crosstransmission in the hospital.However, in a non-outbreak setting, intestinal colonisation of healthcare workers occurs at similar frequency as among healthy adults [112][113][114][115].The These observations suggest that genetically diverse sources play a major role in C. difficile transmission [116].

Clinical impact of infection and colonisation with
Clostridium difficile in ICU patients C. difficile is found as a part of the normal intestinal flora in 1.0% [117] to 12.9% [114] of healthy individuals.Most studies analysing C. difficile colonisation in hospitalised patients have been performed on geriatric wards [118][119][120][121].One study performed in ICU patients documented a colonisation rate of 34.6% [122].
Whether C. difficile colonisation is a risk for infection [123] or has a protective effect [124] is not clear yet.The published prevalence of C. difficile infection and corresponding mortality rate among ICU patients range from 0.5% [125] to 7.3% [126] and from 19.7% [127] to 36.7% [128], respectively (table 3).For ICU patients, C. difficile infection has not been associated with an increase in mortality [126,128,129].The recurrence rates of C. difficile infection in ICU patients are highly variable as follow-up periods differ in most studies.Following treatment for C. difficile, recurrence rates in ICU patients can be as high as 12.7% [129].Several risk factors for C. difficile infection and related mortality in ICU patients have been described (table 3).Antibiotic treatment is strongly associated with C. difficile infection (OR 6.67; 95% CI 1.76-25.31)[130] probably due to the fact that antibiotics interfere with intestinal colonisation resistance leading to overgrowth of C. difficile and toxin production, eventually causing infection [131].Most classes of antibiotics have been associated with C.
difficile infection in the hospital and community setting [132] with highest risks described for quinolones, cephalosporins, and clindamycin [133].Other risk factors are older age, use of proton pump inhibitor and the presence of hypervirulent strains [130].

Background
Generally, infection control on ICUs includes a bundle of prevention strategies.Wenzel and Edmond suggested to Infection control strategies in intensive care units: A continuum.Infection control in intensive care units can involve vertical and horizontal strategies [134,135].While vertical interventions aim to reduce colonisation and infection with a certain pathogen, horizontal interventions try to minimise the spread of pathogens in general by using a universal approach [134,135].For prevention of bacterial infection in ICU patients, vaccines may be of interest in the future.stratify such bundles in vertical and horizontal strategies (fig.1), although considerable overlap between the two approaches exists [134,135].Vertical strategies include all pathogen-specific modules to reduce colonisation and/or infection (e.g., selective decolonisation in MRSA carriers).In contrast, horizontal interventions focus on minimising the spread of all pathogens between patients by using universal approaches (e.g., hand hygiene, chlorhexidine body washing) [134,136].
Hand hygiene as an essential part of infection control is highly effective in reducing all sorts of hospital-acquired infections [137] and therefore is recommended for the prevention of infections caused by VRE, ESBL-GNB and C. difficile [138][139][140][141].A limitation of infection control studies is the fact that results from a study performed in a specific epidemiological context may not apply to other epidemiological settings.Furthermore, steadily evolving and changing antimicrobial resistance patterns make it difficult to draw long-term conclusions.

Colonisation screening and contact precautions (vertical strategy)
Screening for nosocomial pathogens in asymptomatic carriers aims to early identify colonised patients and to timely apply appropriate isolation precautions to prevent spread in the hospital.Interventions in patients with a positive screening result are e.g., contact isolation and decolonisation [142].A challenge is selection of patients at riskranging from targeted screening of high-risk patients (e.g., haematopoietic stem cell recipients, ICU patients) to universal screening performed on every admitted patient [143].
The effect of screening on ICU admission on acquisition and infection rates is mainly documented for VRE/MRSA [144][145][146][147][148][149][150][151].Few studies focused on ESBL-GNB or C. difficile [152][153][154][155]: Due to the increasing prevalence of VRE in the Unites States, the Centers for Disease Control and Prevention (CDC) [156], 1995, and the Society for Healthcare Epidemiology of America (SHEA) [139], 2003, recommended VRE screening on hospital admission with subsequent isolation of colonised patients [52,[157][158][159][160][161][162].A recent randomised trial did not find an additional benefit of universal screening for MRSA/VRE on ICU admission and strict contact precautions compared to pre-existing practice (standard hand hygiene and use of gloves for contact with patient's mucous membranes, wounds, and body fluids) [144].The limitation of this trial was poor compliance with hand hygiene and wearing of gloves and gowns in intervention ICUs, as well as application of barrier precautions in only 35.0-50.7% of all ICU patient-days due to late reporting [163].The findings of this study were confirmed by another recent cluster-randomised cross-over trial including 20 medical and surgical ICUs in the Unites States [145].Universal care with gloves and gowns did not reduce the acquisition of MRSA or VRE compared to contact isolation of colonised patients (difference in acquisition density for MRSA or VRE -1.71 acquisitions per 1,000 person-days; 95% CI -6.15-2.73;p = 0.57) [145].
A European study on 13 ICUs analysed the effect of different vertical and horizontal infection control strategies on acquisition density of VRE, MRSA, and ESBL-GNB [146].After a 6 month baseline surveillance period (phase 1) starting in May 2008, hand hygiene improvement programmes and chlorhexidine body washing were implemented at all ICUs (phase 2), followed by a cluster-randomised trial (phase 3, until April 2011) analysing the additional effect of admission colonisation screening for VRE, MRSA, and ESBL-GNB with subsequent contact isolation of carriers on acquisition incidence.Interventions in phase 2 significantly reduced MRSA acquisition, but had no impact on VRE and ESBL-GNB.Additional screening with subsequent contact isolation of carriers (phase 3) -whether performed by rapid testing (PCR) or conventional testing with chromogenic media -did not further reduce the acquisition incidence of antimicrobial-resistant bacteria (MRSA, VRE, and ESBL-GNB).However, it has to be taken into account that these results are only generalisable to settings with sustained high level of compliance to hand hygiene and chlorhexidine body washing.In a study from our centre we could demonstrate that with the use of strict contact precautions (i.e., hand hygiene, gloves, gowns, single room) for every VRE-colonised patient (no active surveillance), the incidence of VRE at our university hospital decreased to zero, after multiple cases in the mid 90's [21].
Altogether, these studies show that screening with contact isolation of carriers might not be equally effective for different bacteria.The failure to reduce VRE and ESBL-GNB compared to MRSA may be partly explained by differences in colonisation characteristics [146].In contrast to MRSA, VRE and ESBL-GNB mainly colonise the intestinal tract, which is not affected by chlorhexidine body washing [146].Furthermore, the colonisation of the environment probably plays a much more important role in nosocomial enterococcal transmission than previously thought [30].In addition, around 5% of healthcare workers are colonised with MRSA [164] and may also spread the pathogen adding to the difficulties to identify key factors for transmission.

Selective digestive tract decontamination (mainly vertical strategy)
Over the last years, the effect of selective digestive tract decontamination regimens on colonisation and infection rates of ICU patients has been studied using different non-absorbable antibiotics reducing intestinal carriage of mainly gram-negative bacteria (e.g., ESBL-GNB) but also S. aureus and yeasts, sparing the anaerobic flora [165].In a large cluster-randomised trial from the Netherlands, selective digestive tract decontamination and selective oropharyngeal decontamination both significantly reduced incidence of ICU-acquired bacteremia and overall mortality [166].A review article of 65 randomised-controlled trials and 11 meta-analyses showed a reduction in lower airway infections of 72%, bloodstream infection of 37% and overall mortality of 29% with the use of selective digestive tract decontamination regimens [165].

Vaccination (vertical strategy)
Developing vaccines against nosocomial pathogens such as S. aureus and enterococci has been complicated, as the mechanisms leading to protective immunity are only partly understood [183].A temporary effect has been shown for a S. aureus conjugate vaccine in dialysis patients [184].Other vaccines are currently being investigated; the most recent S. aureus vaccine (V710) failed to prevent surgical site infections after cardiothoracic surgery [185].To date, enterococcal vaccines have been solely evaluated in animal studies [186][187][188] and its clinical use needs to be determined.The importance of humoral immune response to C. difficile toxins A and B [189] lead to the development of vaccines as a promising strategy against C. difficile infection.Different vaccines, containing toxoid A and/or B, have been proven safe, immunogenic, and possibly effective in the prevention of C. difficile infection and recurrence [190][191][192][193][194].

Chlorhexidine body washing (horizontal strategy)
The effect of chlorhexidine body washing on bloodstream infection rates and on cross-transmission of multidrug-resistant bacteria has been demonstrated in a cluster-randomised trial in 9 ICUs in the United States showing a significant reduction in hospital-acquired bloodstream infections of 28% with daily chlorhexidine body washing [195].Of note, the reduction was significant only for coagulase-negative staphylococci and not MRSA or VRE.In contrast, a recent meta-analysis showed significantly lowered MRSA/ VRE colonisation and infection densities in patients treated with daily chlorhexidine body washing compared to patients without (incidence rate ratio 0.51; 95% CI 0.36-0.73and 0.57; 95% CI 0.33-0.97;for VRE colonisation and VRE infection, respectively) [196].So far, only a few studies have addressed the effect of chlorhexidine body washing on ESBL-GNB [146,149] and C. difficile [197,198] acquisition, not allowing the drawing of definite conclusions for these pathogens.

Antimicrobial stewardship
Antimicrobial stewardship programs encompass interventions promoting a responsible use of antimicrobial agents in order to improve patient outcome, enhance patient safety, reduce antimicrobial resistance and cut health-care costs [199].A recently published Cochrane systematic review showed that hospital-wide antimicrobial stewardship is safe, reduces antimicrobial resistance and hospital-acquired infection incidence [200].In a current meta-analysis, implementation of antimicrobial stewardship programmes on ICUs, reduced antibiotic use up to 55.4% and direct antibiotic costs by 4.6-72.3US$ per patient-day [201].More importantly, antimicrobial stewardship was associated with reductions in antimicrobial resistance and adverse events, without compromise of short-term clinical outcome [201].
The value of antibiotic cycling or mixing on prevention of multidrug-resistance is unknown [202].A recent Spanish interventional study in ICU patients with ventilator-associated pneumonia indicated that mixing might prevent the emergence of antimicrobial resistance [203].Nevertheless, currently, no definitive conclusions can be drawn on the value of antibiotic cycling/mixing [202][203][204][205][206][207].

Gaps in knowledge
The degree and full extent of health consequences following changes in the human microbiome have only recently been studied and are still little understood.Few studies could show a change in nosocomial infection rate after interventions targeting colonisation with VRE, ESBL-GNB and C. difficile.The rise of multidrug-resistant bacteria in the colonising flora of hospitalised patients but also healthy persons is one of the major challenges of future medicine.Internationally standardised, evidence-based and mandatory policies to control the emergence of multidrug-resistance are urgently needed and the ideal "bundle" of infection control strategies in ICU patients has yet to be defined.

Conclusion
The shift from a normal intestinal microbiome to a 'selected' gut flora dominated by antibiotic-resistant enterococci, ESBL-GNB and C. difficile in critically ill patients is a major risk factor for subsequent infection.The global rise of antimicrobial resistance, the increasing spread of bacteria and antimicrobial-resistance genes in the community and healthcare setting endanger patients at highest risk for nosocomial difficult-to-treat infections, especially in the ICU or on transplant units.Known infection control measures such as hand hygiene and antimicrobial stewardship urgently need to be implemented all over the world.New infection control measures need to be studied in order to halter further spread of resistant bacteria.The concurrent paucity of new antibiotics being developed stresses the importance of preventive measures even more.Especially horizontal infection control strategies could gain in im-

Figures (large format)
Infection control strategies in intensive care units: A continuum.Infection control in intensive care units can involve vertical and horizontal strategies [134,135].While vertical interventions aim to reduce colonisation and infection with a certain pathogen, horizontal interventions try to minimise the spread of pathogens in general by using a universal approach [134,135].For prevention of bacterial infection in ICU patients, vaccines may be of interest in the future.

Table 1 :
Studies on vancomycin-and ampicillin-resistant enterococci: rates and risk factors for colonisation/infection in adult patients (intensive care unit setting only).

design and reference Study year Country Total of colonised patients Rate of co- lonisation on admission a , % Rate of HA colonisation a , % Rate of infections, % Risk factors for VRE/ARE colonisation and/or infection Vancomycin-resistant enterococci
Rectal/perirectal colonisation if not stated otherwise.HA colonisation was defined as negative culture within the first 48h of ICU admission and subsequent positive culture; b among patients with HA VRE colonisation; c among patients with VRE colonisation on admission; d total VRE colonization rate (on admission or HA); e based on rectal, faecal, and/or urine cultures; f based on rectal, faecal, respiratory, and/or urine cultures; g based on rectal, integumental (groin and arm), oropharyngeal, tracheal, and/or gastric cultures; h ICU and/or medical ward.

Table 2 :
Studies on extended-spectrum ß-lactamase producing gram-negative bacteria: rates and risk factors for colonisation/infection in adult patients (intensive care unit setting only).
[116]hospital-acquired; ESBL = extended-spectrum ß-lactamase; GNB = gram-negative bacteria; POS = prospective observational study; ROS = retrospective observational study; VAN = vancomyin; QN = quinolone; MN = metronidazole; CAR = carbapenem; P/T = piperacillin/tazobactam; CEPH = cephalosporin; CFM = cefepime; IMI = imipenem; ICU = intensive care unit; MRSA = methicillin-resistant Staphylococcus aureus; NA = not assessed.aRectal/perirectalcolonisationif not stated otherwise.HA colonisation was defined as negative culture within the first 48h of ICU admission and subsequent positive culture; b HA colonisation and/or infection; c total VRE colonisation rate (on admission or HA); d based on cultures from the nares, oropharynx, and/or rectum; e other pathogens than ESBL-GNB included in analysis; f based on an ecological correlation; g colonisation within the first 72 hours after admission.Swiss Medical Weekly • PDF of the online version • www.smw.chimportance of nosocomial transmission of C. difficile has been questioned by a recent study from Oxfordshire, United Kingdom[116].Using whole-genome sequencing, 45% of patients with C. difficile had genetically distinct strains compared to patients previously diagnosed with C. difficile.Noteworthy, even within a single patient, diverse subtypes were detected indicating different transmission events.

Table 3 :
Studies on Clostridium difficile: infection rate, recurrence rate, mortality, and risk factors for infection/death in adult patients (intensive care unit setting only).

Mortality in CDI, % Risk factors for CDI or CDI-associated mortality
e Long hospital stay, enteral tube feeding, mechanical ventilation, Pseudomonas aeruginosa bacteremia, VRE colonisation or infection, gastric acid suppressive therapy, C. difficile colonisation pressure, antimicrobial therapy CDI = Clostridium difficile infection; ROS = retrospective observational study; APACHE = acute physiology and chronic health evaluation; ICU = intensive care unit; NA = not assessed; CT = computed tomography; SOFA = sequential organ failure assessment; VRE = vancomycin-resistant enterococci.a in-hospital mortality rate; b 30-day in-hospital mortality rate; c 3 consecutive surveillance periods; d crude 30-day mortality rate; e mortality rate during ICU stay.Swiss Medical Weekly • PDF of the online version • www.smw.ch Swiss Med Wkly.2014;144:w14009 Swiss Medical Weekly • PDF of the online version • www.smw.chportance as new multidrug-resistant pathogens constantly emerge.