Antimicrobial resistance is a major public health concern in human medicine both in the community and in medical institutions [1]. Aged patients, especially frail elderly, constitute a subgroup of persons which seems more affected than other population groups. For instance, the carriage of methicillin-resistant Staphylococcus aureus (MRSA) in acute care hospitals has been well documented in Belgium since the early nineties [2]. Over the last years, large reservoirs of MRSA carriers have been identified among elderly subjects living in nursing homes [3]. A recent cross-sectional survey carried-out in 2005 among 60 Belgian nursing homes (NH) showed an average prevalence rate of MRSA carriage as high as 20% of the residents (with ranges between 2 and 42%) [4]. By contrast, epidemiologic surveys of multi-drug resistant gram-negative microorganisms (MDR) have been very scarce and the current prevalence of such organisms currently remains only partially known in most European structures of care, especially in long-term care facilities [5–7]. Initial reports of ESBL-producers in the eighties and early nineties essentially focused on the descriptions of nosocomial outbreaks, mainly in intensive care unit settings, caused by Klebsiella and Enterobacter spp. which produced various TEM- or SHV- ESBL subtypes [8]. More recently, novel ESBLs coding genes (e.g. of the CTX-M family) derived from plasmid mobilization of environmental or soil organisms have emerged worldwide in several enterobacteria species particularly in E. coli [9]. The CTX-M ESBLs are most often encountered in association with community-acquired infections rather than with hospital outbreaks [8]. In a nationwide surveillance program, the proportion of ESBL-producing E. coli isolated from clinical specimens referred to hospital-based laboratories was found to rise from 3.6% in 2005 to 4.8% in 2008 [10, 11].

Reports depicting the clinical pictures and profiles of patients colonized or infected by ESBL-producing Enterobacteriaceae isolates and admitted to hospital have mainly been published from the United States, Canada or Israel but are rather limited in Europe [8, 9, 12]. In order to gain a better insight of the epidemiology and the risk factors associated with ESBLs, we undertook a descriptive analysis of the clinical charts of patients in whom ESBL-producing microorganisms had been isolated from clinical samples.

Overall, 114 patients were included in the study. During the 20-month period of the study, the total number of admission was 26,727 and the number of hospitalisation-days was 198,476. The number of routine microbiological samples analysed during this period of time was 53,545 (urinary analyses: 14,342, blood cultures: 17,339, respiratory tract samples: 8,782 and wounds or swabs: 13,082). The prevalence of ESBL-producing Enterobacteriacae isolates recovered from clinical samples (excluding duplicate isolates from the same patient) was 4.5% [114/2525]. The incidence rate of colonization/infection caused by ESBL-positive Enterobacteriaceae isolates was 4.3 cases/1000 admitted patients and 0.57 cases/1000 hospitalization days.

Data of the 114 patients were categorized according to the probable healthcare-associated or community origin of acquisition of ESBL-producing Enterobacteriaceae. Because of the study design and of the hospital-case mix of the population (85-90% of clinical specimens submitted to the laboratory originated from hospitalized patients), the healthcare-associated ESBL+ group was the most important (n = 104 [91%]) while the community acquired ESBL+ group was smaller (n = 10 [9%]). Compared with the group of patients with healthcare-associated ESBLs, patients with community-acquired ESBLs originated mostly from the ambulatory clinic (n = 7), were significantly younger (52 ± 20 years; P < 0.001) and were more frequently asymptomatic carriers (Escherichia coli (n = 8) harbouring CTX-M1 group ESBL) (P: 0.005).

Hospitalized patients colonized or infected with an ESBL producer were admitted in all types of wards including surgery (n = 43), internal medicine (n = 17), intensive care unit (n = 11), oncology/haematology (n = 10), and geriatric wards (n = 8). A high number of co-morbid conditions were present in these patients (diabetes mellitus (23%), chronic renal insufficiency (23%), congestive heart failure (13%), chronic respiratory disease (16%), and haematological malignancies (8%)). The mean CIRS-G global score was 10.5 ± 3.9 (mean number of different categories: 5.2 ± 2.0) and the mean Norton Index was 14.8 ± 3.6. Intravascular, bladder catheter or gastrostomy tubes were present in half of the patients. Other risk factors commonly present for the acquisition of a MDR-bacteria were also found; two-thirds of the hospitalized patients had been previously exposed to antibiotics (with a mean of two different antibiotic regimens) prior to the isolation of an ESBL-producing organism, over 60% of them had been hospitalized or had had multiple medical contacts with the hospital within the last 12 months, and 29% of the patients had been hospitalized in the ICU during hospitalization before detection of an ESBL.

Sixty-eight (65%) of the hospitalized patients were considered as infected while 36 patients were asymptomatic carriers of an ESBL-producing Enterobacteriaceae isolate (35%). The four major documented infections were: urinary tract infection (UTI) (n = 38, [56%]), lower respiratory tract infection (RTI) (n = 18, [27%]), septicaemia (n = 6, [9%]) and intra-abdominal infections (n = 3, [4%]).

The median hospital length of stay (LOS) was 23 days (inter-quartile range: 13-39 days) and the overall mortality rate during hospitalization was 13% (Confidence Interval 95% [95% CI]: 7-19). During the study period, the average LOS in the institution was of 7.4 days while the number of death was 602 (2.3 cases/100 admitted patients).

A fatal outcome was documented only in patients with ESBL-producing bacteria of healthcare-associated onset (12 deceased patients out of 104 vs. 0 out of 10 in those with community acquired ESBL; P = 0.01). Compared to patients who were still alive at the end of their hospital stay, deceased patients had a significantly higher mean LOS (42 ± 30 days; P: 0.05). Six out of the 12 patients who died were hospitalized in the ICU at the time of death. Of the 12 patients with a fatal outcome, 10 patients were infected (5 lower respiratory tract infection, 3 bloodstream infection and 2 urinary tract infection) while 2 only were ESBL carriers. However, mortality could be directly attributed to the infection in only 5 of the 10 cases.

This study reports the clinical patterns in a cohort of subjects carrying ESBL-producing Enterobacteriaceae isolates over a 20-month period. We found that a two-thirds of the patients carrying ESBLs were elderly adults aged over 65 years and who presented a large variety of co-morbidities (on the average, 5 ± 2 different organs affected per patient), as illustrated by the high mean score of the CIRS-G index. Interestingly, the majority of these patients (whether being asymptomatic carriers or whether presenting an ESBL infection) had had several hospitalizations or multiple contacts with the hospital within the previous 12 months. The proportion of patients who had received one or more courses of antimicrobial therapy before the isolation of an ESBL-positive strain was high (58%, n = 60). It is most probable that the frequent exposure to previous antimicrobial treatment played a major role in the selection of ESBL-producing organisms as previously demonstrated in several case-control studies [6, 20–22]. Further, antibiotic exposure is also known as a risk factor for the development of community-acquired bacteraemia caused by ESBL-producing E. coli [7], but due to the small number of patients with true community-acquired ESBL infection in our cohort we were not able to confirm this in the present study. In their comprehensive review of the literature, Pitout et al. highlighted the previous usage of antimicrobial agents (in particular cephalosporins), the severity of illnesses, a longer duration of hospital stay especially in the intensive care unit and the presence of urinary or arterial catheterisation as independent risk factors for the development of health-care associated ESBL infections [8].

One recent study in France, also highlighted the importance of inter-hospital transfer, especially after hospitalization in a foreign country, the previous carriage of an ESBL-producing Enterobacteriaceae or the presence of a chronic illness as risk factors for the acquisition of ESBLs [23]. Interestingly, these investigators also found that 4% of admitted patients without any potential risk factors were ESBL+ at entry.

The retrospective design of our study did not allow to establish the risk of co-colonization by ESBL-producing gram-negative bacilli and by other drug resistant bacterial pathogens such as methicillin-resistant Staphylococcus aureus (MRSA) or vancomycin-resistant Enterococci (VRE). However it was interesting to note that 9% (n = 9) of the patients colonized by an ESBL-producing organism were also known as previous MRSA carriers. The risk of co-colonisation probably reflects the occurrence of common risk factors for colonization or infection by different types of nosocomial pathogens [24].

The median LOS of 23 days and the 13% mortality rate observed here appear as much higher figures than the expected average rates observed at our institution (7.4 days and 2.3 death/100 admitted patients respectively). Lautenbach et al. also reported a longer LOS (OR: 1.74) and they documented higher hospital costs (OR: 2.9) associated with infections caused by an ESBL-producing E. coli or K. pneumoniae organism in comparison to a matched-control group of patients infected with organisms belonging to these same species but that were not producing an ESBL [25]. Furthermore, a positive association has also been reported between the LOS and the risk of colonization by an ESBL-producing microorganism. For instance, Friedmann et al. found in a non-outbreak setting that one-third of patients in general medicine wards (mean age of 76 ± 13 years), were colonized by ESBL-producing Enterobacteriaceae after a course of hospitalization of 2 weeks [12].

Several case-control studies have suggested that infections caused by ESBL-producing Enterobacteriaceae species were associated with a higher mortality rate as compared to infections caused by the same organisms but lacking an ESBL [26–28]. A recent meta-analysis concluded that bacteraemia caused by ESBL-producing Enterobacteriaceae was associated with an increased mortality rate (pooled Relative Risk [RR]: 1.85, 95%CI: 1.39-2.47, P < 0.001) and that delayed effective therapy did further increase the mortality (pooled RR: 5.56, 95% CI: 2.94-10.51, P < 0.001) [29]. Nevertheless, the real attributable impact of ESBL infection on mortality remains controversial and a critical review of the literature does not provide clear answers to this matter [28]. The lack of well-powered studies, the importance and clinical severity of the underlying illnesses, the appropriate or inappropriate choice of empirical therapy, as well as variable performance of microbiology laboratories to detect and report ESBL microorganisms, might all be limiting factors or confounding variables for a correct interpretation of the data. Well designed prospective studies are thus clearly warranted to address this important issue [28].

As anticipated, we found that E. coli by far accounted as the most frequent ESBL-producing enterobacteria species. A recent study carried out in 2004 and involving 16 laboratories in London and in the South-Eastern part of England showed that among 1127 cephalosporin-resistant gram-negative isolates from both hospital and community settings, E. coli also accounted as the largest group (51%) and, in particular, CTX-M producing strains [30]. The high prevalence of CTX-M-1 group ESBLs (mainly comprising the CTX-M-15 type) is an emerging and well described problem worldwide [31–37]. While most published reports until the 1990's found a predominance of SHV- or TEM-type ESBLs associated with nosocomial outbreaks, the epidemiology dramatically changed over the last ten years. The extensive dissemination of E. coli in the community [37] and its subsequent spread in hospitals illustrates the importance of movement of patients between the community and the healthcare system, in particular between nursing homes and acute hospitals. Data from our study further highlight this emerging pattern. In France, a progressive increase in the occurrence of CTX-M enzymes linked to E. coli has also been reported [31, 37]. In Belgium, a three-fold increase of the prevalence of multi-resistant CTX-M producing E. coli was detected between 2000 and 2004 [38] and over 50% of the ESBL-producing Enterobacteriaceae isolates identified in human clinical isolates in Belgium in 2008 belong to the CTX-M [39]. CTX-M-15 (75% of all CTX-M ESBLs) and CTX-M-2 to a lesser extent currently account as the most predominant ESBLs of the CTX-M type in Belgium and this likely reflects both clonal dissemination as well as plasmid epidemics across different bacterial species (Salmonella enterica and E. coli) in animals and in humans [40].

There have been only few studies which did recently address the epidemiology of ESBL-producing Enterobacteriaceae in Belgium, and the present report gives some insights on the clinical patterns of healthcare associated colonization or infection caused by ESBL-producing Enterobacteriaceae.

We do however recognize the limitations of this retrospective study which was only based on the collection of routine microbiological samples with no screening being carried out to detect asymptomatic carriage upon hospital admission. Comparing and translating data obtained from studies carried out in epidemic or endemic settings with a systematic screening approach may therefore be difficult.

Further, we acknowledge the fact that the selection of patients from clinical samples submitted to a hospital-based laboratory may have introduced some bias. Indeed, almost 90% of the clinical specimens submitted for microbiological analysis to our bacteriology laboratory originate from hospitalized patients and/or from subjects with frequent contacts with the healthcare system. This may be explained by the fact that our hospital is geographically located in a rural area and that over 90% of the microbiological specimens received in our laboratory originates either from hospitalized patients or from patients regularly visiting the polyclinics; thus the group of patients with true community acquired infections clearly constitutes a minor proportion in our setting.

The lack of a control group constitutes a limiting factor excluding possible comparisons of odds ratios for clinical risk factors.

Despite all the limitations we think these data may be interesting for the following reasons: Apart from of one recent microbiological report of ESBLs isolated from humans in Belgium [40], our study constitutes one of the first attempt of a global clinical description of a large cohort of patients colonized or infected by ESBL-producing gram-negative bacilli at one university hospital. Since the majority of the patients were aged over 65 years, the patient's characteristics were also defined with geriatric tools such as the Cumulative Illness Rating Scale for Geriatrics and the Norton Index. The reliance on these objective scales and indexes may be useful to develop algorithms and screening strategies aiming to select at risk groups in this population upon hospital admission. Urinary tract and respiratory tract infections were the two major diagnostic categories and these indeed constitute the most frequent types infections among elderly patients admitted in general internal medicine or geriatric wards. While most clinicians do currently recognize the importance of MRSA screening upon hospital admission, few teams would consider assessing the potential risk factors for ESBL carriage upon hospital admission. Nevertheless, asymptomatic intestinal carriage of ESBLs could represent as for MRSA an important risk factor for the development of subsequent infections. One study carried out in a large cohort of elderly patients screened upon hospital admission in Israel showed an increase rate of subsequent bacteraemia for ESBL faecal carriers (15% versus 0.5%; odds ratio: 38.9; P < 0.001) [5]. A more comprehensive approach may also lead to improvement in the choice of empiric antibiotic treatment, infection control procedures and patient outcomes.

We strongly believe that multidisciplinary work involving geriatricians, infectious diseases specialists, infection control units and the microbiological laboratory could improve clinical practice and help reducing antibiotic pressure [41].

Prospective studies with systematic screening of asymptomatic ESBL carriage in subjects residing in nursing homes and in elderly patients upon admission to hospitals in geriatric wards are necessary to improve our knowledge on the dynamic of acquisition and transmission of ESBLs between different human sectors.