Extended-spectrum β-lactamases in ambulatory care: a clinical perspective



This article is corrected by:

  1. Errata: Errata concerning Volume 14, Supplement 1, January 2008 Volume 14, Issue 3, 293, Article first published online: 31 January 2008

Corresponding author and reprint requests: J. Rodríguez-Baño, Sección de Enfermedades Infecciosas, Hospital Universitario Virgen Macarena, Avda. Dr Fedriani 3, Seville 41071, Spain
E-mail: jrb@nacom.es


Extended-spectrum β-lactamase (ESBL)-producing organisms are emerging as a cause of infection in outpatients in many areas of the world. Most of these involve urinary tract infections caused by Escherichia coli harbouring CTX-M types of ESBLs. Risk-factors for such infections include older age and previous use of quinolones and cephalosporins. Oral alternatives for such infections are limited. These organisms may also cause some community-acquired life-threatening infections, such as bacteraemic urinary tract or intra-abdominal infections. Patients with these infections are at increased probability of receiving inappropriate empirical treatment. More studies addressing the risk-factors and therapeutic options for these infections are needed.

Until recently, the microorganisms producing extended-spectrum β-lactamases (ESBLs) have mainly caused problems in acute-care hospitals [1]. However, these organisms have been spreading to other healthcare facilities such as nursing homes [2], and, during recent years, infections caused by ESBL-producing organisms have been increasingly diagnosed in outpatients [3–5]. The leading role in this new situation is being played by ESBL-producing Escherichia coli, and, at the molecular level, by the genes encoding for CTX-M types of ESBLS [6], which are rapidly spreading throughout the world [7].

From a clinical perspective, the impact of this phenomenon is very relevant. ESBL-producing organisms are not only resistant to penicillins and cephalosporins, but are also frequently resistant to trimethoprim–sulphamethoxazole and fluoroquinolones [1]. Thus, these organisms may be resistant to most of the antimicrobial agents that are usually recommended for the treatment of community-acquired infections caused by enterobacteria. The situation is even more worrisome because they may also cause severe invasive infections. This review will summarise the epidemiology, risk-factors, clinical features and therapeutic options for infections caused by ESBL-producing organisms in outpatients.

Epidemiological overview of infections caused by ESBL-producing organisms in outpatients

ESBL-producing organisms have been emerging as relevant community pathogens during the last decade in many areas of the world. The evolution of ESBL-producing organisms was studied from 1995 to 2002 in an area of 500 000 inhabitants in Seville, Spain [8]. From 1995 to 1998, only 14 strains of E. coli were ESBL producers. The situation abruptly changed in 2000, when 53 isolates of ESBL-producing E. coli were detected, and there was a further increase during 2001 and 2002 (106 and 114 isolates, respectively). Thus, the percentage of E. coli isolates that were ESBL producers increased from 0.3% in 1995 to 4.8% in 2002. During the same period, there was a significant decrease in the rates of ESBL-producing Klebsiella pneumoniae isolates, following the control of nosocomial transmission of this organism. However, many data showed that ESBL-producing E. coli strains were not replacing K. pneumoniae strains, but constituted a new and distinct phenomenon; while most ESBL-producing K. pneumoniae strains were clonally related, produced ESBLs of the SHV and TEM families almost exclusively, and were mainly isolated from hospitalised patients, ESBL-producing E. coli strains were not clonally related, the predominant ESBLs produced belonged to the CTX-M family, and approximately half were isolated from outpatients [8,9]. These features were confirmed in a specific clinical and molecular investigation of the epidemiology of ESBL-producing E. coli in outpatients [3].

As an example of the explosive dissemination of the CTX-M enzymes, CTX-M-14 (an enzyme belonging to the CTX-M-9 group that was characterised in Taiwan [10]) emerged almost at the same time in such distant areas as Spain [11,12]. Also, intercontinental spread of many other CTX-M enzymes has rapidly occurred. Such efficacious dissemination has been related to the fact that the genes encoding for these enzymes are associated with highly transmissible genetic mobile elements [6]. In fact, most isolates of E. coli harbouring these enzymes have been found to be clonally unrelated [3,5,11–13]. However, clonal spread has also been described both in the community and in healthcare facilities [14], and it seems to be particularly frequent in the case of strains harbouring CTX-M-15 [15,16], which further contributes to the problem of dissemination.

The prevalence of colonisation with ESBL-producing enterobacteria in the general population is largely unknown. There are probably substantial geographical variations. Most data come from studies in which the presence of ESBL-producing enterobacteria has been investigated in faecal samples from patients with diarrhoea. In Barcelona, Spain, Mirelis et al. [17] studied the presence of ESBL-producing enterobacteria in faecal samples obtained from outpatients presenting at their hospital during three different periods. The percentage of carriers increased from 2.1% (February–May 2001) to 3.8% (April–June 2002) and to 7.5% (October 2002). In Madrid, Spain, Valverde et al. searched for the presence of ESBL-producing enterobacteria in faecal samples collected during 1999 and 2003 [18]. The percentage of outpatients from whom ESBL-producing enterobacteria were isolated significantly increased from 0.7% in 1999 to 5.5% in 2003, when all the ESBL-producing isolates found were E. coli. The most frequent types of ESBL found were CTX-M, followed by SHV. In 2003, faecal samples from healthy volunteers were also studied and the rate of colonisation was 3.7%. In York, UK, Munday et al. studied 1000 faecal samples (565 from outpatients) during October–December 2003 [5]. They found ESBL-producing E. coli in nine samples, eight of which were obtained from outpatients (rate in outpatients, 1.4%). All these isolates produced CTX-M enzymes (five CTX-M-15, three CTX-M-14 and one CTX-M-9). The rate for faecal carriage was 7% in a study performed in healthy executives in India [19] but was much lower in a study in healthy children in Bolivia and Peru (0.1%) [20].

Data concerning the population-based rates for infections due to ESBL-producing E. coli are scarce. In Calgary, Pitout et al. found a rate of 5.5 cases per 100 000 population during 2000–2002; only 29% of the cases involved nosocomial infections [14]. Similar rates were found in a multicentre study performed in Spain including only community-acquired infections. With regard to specific data concerning community-acquired urinary tract infections (UTIs), the percentage of ESBL-producing E. coli isolates ranged from 1.4% to 3.3% in studies from Seville, Barcelona and the Gaza Strip [3,21,22]. There are even fewer data concerning invasive infections. In Seville, 6.5% of community-acquired bacteraemia cases due to E. coli during 2001–2005 were caused by ESBL-producing isolates, with an increasing trend [23]. In a multicentre study on the susceptibility patterns of Enterobacteriaceae isolated from patients with intra-abdominal infections in Spain, 15 of 398 (3.7%) isolates from community-acquired infections were ESBL producers [24]. Again, the rates may be very different in different geographical areas.

Risk-factors for ESBL colonisation and infection in outpatients

Several groups have investigated the variables associated with the isolation of ESBL-producing E. coli in outpatients. Three case-control studies have been published, two of them from different areas in Spain and the other from Israel (Table 1) [3,21,25]. The methodological differences between them may account for the different results. Nevertheless, previous use of antimicrobials that may select for ESBL-producing isolates, such as quinolones (due to the high frequency of co-resistance) and cephalosporins, were found to be a risk-factor in all three studies. The fact that previous hospitalisation was a risk-factor in the two bigger studies [3,25] should not lead to the conclusion that the organisms were acquired during the hospital stay, although this could be the case for those patients with ESBL-producing K. pneumoniae isolates that were also included in the study from Israel. Our data strongly suggest that at least the CTX-M-producing E. coli isolates were truly community-acquired, and therefore previous hospitalisation could be a surrogate marker for other predisposing factors. In fact, previous hospitalisation was not a risk-factor when only CTX-M producers were analysed [3].

Table 1.   Case-control studies investigating the risk-factors for infections caused by extended-spectrum β-lactamase (ESBL)-producing Escherichia coli in outpatients
ReferenceCase definitionNumber of casesControl definitionRisk-factors (multivariate analysis)
  1. aESBL-Ec, extended-spectrum β-lactamase-producing E. coli.

[3]Outpatients with isolation of ESBL-Eca in any clinical sample, no hospital admission in last month49Outpatients with a sample sent for culture, no hospital admission in last monthAge in male, diabetes mellitus, previous hospital admission, previous fluoroquinolone use
[21]Community-acquired urinary tract infections caused by ESBL-Eca19Community-acquired urinary tract infections caused by non-ESBL-Ec (matched by gender, age and address)Oral cefuroxime
[25]Community-acquired urinary tract infections caused by ESBL-Eca or Klebsiella pneumoniae128 (74 E. coli)Community-acquired infections caused by non-ESBL-Ec or K. pneumoniaeAge >60 years, male sex, previous use of cephalosporins or quinolones, previous hospital admission, infection caused by K. pneumoniae

In their population-based surveillance study concerning ESBL-producing E. coli in Calgary, Pitout et al. [4] found that patients aged ≥65 years, as compared to younger patients, were at increased risk of infection caused by these organisms. Age was also identified as a risk-factor in two of the case-control studies [3,24], although only in male patients in the study from Seville [3]. The situation is less clear with regard to gender. While Pitout et al. [4] found that female subjects were at increased risk in all age groups, female sex was not found to be a risk-factor in any of the case-control studies. In fact, male gender was an independent risk-factor in the study by Colodner et al. [24].

A different approach was used by Ben-Ami et al. in Israel [26]. These authors screened 241 patients at hospital admission; 26 (10.7%) were found to be faecal carriers of ESBL-producing Enterobacteriaceae. The multivariate analysis of this cohort selected poor functional status, current antibiotic use, chronic renal insufficiency, liver disease and use of histamine2 receptor antagonists as independent risk-factors for faecal carriage. As long as the influx of these organisms from the community may potentially cause problems inside hospitals, these data might help to identify high-risk patients to whom infection control efforts, such as the performance of surveillance cultures, should be directed. It would be necessary to perform similar studies in other areas, as risk-factors may differ according to differences in antimicrobial prescribing practices, rates of colonisation in the population, etc. However, it is probably impossible to avoid the influx of these microorganisms into hospitals, and consequently, other measures, such as reducing the use of antimicrobials that select ESBL-producing organisms (mainly extended-spectrum cephalosporins and quinolones), should be considered.

There is less information concerning the specific risk-factors involved in the acquisition of CTX-M-producing E. coli in outpatients. In our case-control study, all CTX-M-producing isolates harboured enzymes from the CTX-M-9 group. Previous fluoroquinolone use, age and a more severe chronic underlying condition as measured by the Charlson index were independent risk-factors [3].

Ben-Ami et al. also performed a case-control study to investigate the risk-factors for non-nosocomial bacteraemia caused by ESBL-producing enterobacteria, using episodes caused by non-ESBL-producing enterobacteria as controls [26]. They included 38 cases (15 caused by E. coli, 17 by Klebsiella spp., and six by Proteus mirabilis). The multivariate analysis selected male gender and admittance directly from a long-term care facility as independent risk-factors. However, the model was only moderately predictive, probably reflecting the heterogeneity of the cases. Our own group recently reported the preliminary results of an ongoing case-control study in which the cases were patients with community-acquired bacteraemia caused by ESBL-producing E. coli [27]. The identified risk-factors were previous antimicrobial use, age and structural disease of the urinary tract. More studies in this field are necessary, because identification of risk-factors for community-acquired bacteraemia caused by ESBL-producing organisms is particularly relevant from a clinical perspective, as it may help to identify those patients for whom empirical therapy should include coverage against ESBL producers.

Clinical features and therapeutic options

Although many published studies describe the microbiological features of ESBL-producing organisms in outpatients, the information concerning clinical features and the prognosis of community-acquired infections caused by these organisms is surprisingly scarce. Many studies, aimed primarily at describing molecular aspects of the strains, include isolates obtained from outpatients and inpatients, and frequently lack epidemiological and clinical data, so it is not possible to know whether the infections were really community-acquired, or which types of infection were included. Obviously, collecting precise clinical and follow-up data from outpatients is not an easy task. Since most of the clinical microbiology laboratories where ESBL-producing organisms are identified are located in hospitals, a collaborative approach involving the hospital and primary-care levels on one side, and microbiological and clinical specialists on the other, is necessary.

Community-acquired pathogens that commonly cause diarrhoea may also harbour ESBLs [1]. ESBL-producing Salmonella isolates have been found in many parts of the world; also, occasional cases caused by Shiga toxin-producing E. coli, Shigella and Vibrio cholerae have been described. As expected, the majority of ESBL-producing E. coli strains isolated from outpatients come from urinary samples, with samples from other sources, e.g., blood, wounds, and respiratory tract, accounting for <40% of the samples [3,4,9,11,16]. Only one study provides specific clinical information concerning the patients; in that study, which included 49 cases, the organisms were isolated from urine samples in 47 patients and from blood in six (in two patients, both urine and blood cultures yielded ESBL-producing E. coli) [3]. Thirty-seven patients (76%) were considered to have a symptomatic UTI, 11 (22%) had asymptomatic bacteriuria, and one (2%) had acute cholangitis. Five of the patients with UTIs (13.5%) were bacteraemic. The patient with cholangitis was also bacteraemic. Among the 28 patients who received empirical antimicrobial therapy, only ten received agents to which the organism was susceptible in vitro. A UTI relapse occurred in 13% of patients. None of the patients in the series died as a consequence of the infection. Obviously, more studies are needed in order to gain better knowledge of ESBL-producing E. coli as a cause of community-acquired UTIs. In particular, more information about the importance of these isolates as a cause of complicated UTIs is awaited. Regarding therapy, there are no published studies analysing the efficacy of different regimens for UTIs caused by these strains. Where available, oral fosfomycin is a reasonable alternative for uncomplicated UTIs [28], as most isolates are susceptible in vitro [29], and nitrofurantoin may also be useful [29]. Amoxycillin–clavulanate [30], quinolones and trimethoprim–sulphamethoxazole are useful when the organism is susceptible.

From a clinical perspective, the situation is more worrisome when patients present with clinical sepsis. Among all patients from whom an ESBL-producing E. coli strain is isolated from a clinical sample, c. 10% are bacteraemic [3,4,9]. In the area of Seville, 6.5% of all community-acquired episodes of bacteraemia caused by E. coli have involved ESBL producers during recent years, but the percentage seems to be increasing [23]. In the BSAC bacteraemia study performed in the UK during 2002, 8/210 blood isolates of E. coli (3.8%) were ESBL producers [31]. In Israel, Ben-Ami et al. found that 4/50 (8%) of non-nosocomial bacteraemias due to E. coli involved ESBL producers during 2003; three of these cases did not involve previous healthcare contact [26].

Most of the clinical information about bacteraemia caused by ESBL-producing organisms is from studies including only, or predominantly, nosocomial episodes caused by Klebsiella species. Two studies that specifically address episodes caused by ESBL-producing E. coli also include a majority of nosocomial episodes [32,33]. There are some descriptions of anecdotal cases of community-acquired bloodstream infections caused by ESBL-producing enterobacteria [34,35]. We have recently analysed the epidemiology, prognosis and clinical features of nosocomial and community-acquired bacteraemia caused by ESBL-producing E. coli in the area of Seville during 2001–2005 [23]. Forty-three cases were included; 22 (51%) were non-nosocomially acquired, and eight (19%) were strictly community-acquired (the patients had no significant healthcare experience during the previous year). Considering only the non-nosocomial cases, 45% and 18% of the patients had obstructive diseases of the urinary and biliary tract, respectively, so it was not surprising that the sources of the episodes were the urinary tract in 55% of the patients and an intra-abdominal infection in 36% (3/8 of the latter were biliary tract infections); the source was unknown in 9%. Half of the patients had recently received antimicrobials, fluoroquinolones being the most frequent. Regarding clinical severity, 18% of the patients presented with severe sepsis or septic shock. Empirical antimicrobial therapy was appropriate according to in-vitro susceptibility testing in 64% of the cases, and the crude mortality rate was 18%. These data emphasise the clinical relevance of ESBL-producing E. coli as a potential cause of severe community-acquired infections. The analysis of the influence of the empirical antimicrobial treatment in the prognosis included both community and nosocomial cases; the adjusted mortality was higher in patients who received cephalosporins or quinolones than in those treated with β-lactam–β-lactamase inhibitors or carbapenems. In the series by Ho et al., seven patients with apparent ceftazidime-susceptible isolates were treated with this agent and four of them failed to recover [32]. All of these data suggest that, similar to what happens with other ESBL-producing enterobacteria [1], cephalosporins should not be used for the empirical treatment of severe infections potentially caused by these organisms. Appropriate antimicrobial therapy was not associated with improved survival in another study, but 85% of the cases received appropriate regimens in that series [33]. The results of a multicentre study on bloodstream infections caused by ESBL-producing E. coli conducted by the Spanish Network for Research in Infectious Diseases (REIPI) are currently being analysed; available data indicate that inappropriate empirical therapy (including cephalosporins) is associated with increased rates of mortality. However, until more data are available, it seems prudent for protocols for the empirical treatment of sepsis potentially caused by E. coli to be revised in areas where these organisms are emerging, unless for patients with severe infections and risk-factors for ESBL-producing isolates.

Carbapenems are considered to be the drugs of choice for the treatment of severe infections caused by ESBL-producing organisms [1], and among them, ertapenem has been suggested as a reasonable possibility for treatment of these community-acquired infections, as long as its in-vitro activity against these isolates is comparable to that of imipenem and meropenem [36], and it should exert the least selection pressure for resistance in Pseudomonas aeruginosa, as compared with other carbapenems. However, the emergence and spread of carbapenem resistance is a concern, and alternatives should be studied. The rate of aminoglycoside resistance (amikacin, particularly) among CTX-M-producing E. coli isolates is usually low; thus, the addition of an aminoglycoside to the empirical regimen for treatment of community-acquired sepsis originating in the urinary tract might be a reasonable option for selected patients at low risk for renal insufficiency, until the susceptibility data of the causative agent are available (usually 48–72 h). In many areas, the rates of in-vitro resistance to β-lactam–β-lactamase inhibitor combinations in ESBL-producing isolates are high enough to preclude their empirical use when it is necessary to cover these organisms [1,3,26]. Some data indicate that temocillin, a semi-synthetic derivative of ticarcillin that is available in some countries, has good in-vitro activity against ESBL-producing E. coli [37], but no clinical experience with this antimicrobial has been published. Also, tigecycline, a new compound that has been approved for complicated intra-abdominal and soft-tissue infections, has shown excellent in-vitro activity against ESBL-producing isolates and is another alternative for treatment of such infections [38].


ESBL-producing enterobacteria, and particularly E. coli, have emerged as significant pathogens in outpatients almost worldwide, mainly due to the spread of CTX-M enzymes. ESBL-producing E. coli strains mainly cause UTIs in predisposed patients, but severe infections may also occur. Since there are not many alternatives for the treatment of these infections, more studies addressing the risk-factors, clinical significance and therapeutic options for community-acquired infections caused by these organisms are needed. Local surveillance of ESBL-producing enterobacteria, and particularly of those causing community-acquired bacteraemia, is critical in order to adequately design recommendations for the empirical therapy of severe urinary tract and intra-abdominal infections.

Conflict of interest statement

J. Rodríguez-Baño has been a consultant for Wyeth and Merck, and has served as a speaker for Wyeth, Merck and GlaxoSmithKline.