Community Infections Caused by Extended-Spectrum β-Lactamase–Producing Escherichia coli

Background  Extended-spectrum β-lactamase (ESBL)-producing Escherichia coli is an increasingly important group of community pathogens worldwide. These organisms are frequently resistant to many of the antimicrobial agents usually recommended for the treatment of infections caused by E coli, such as penicillins, cephalosporins, fluoroquinolones, and trimethoprim-sulfamethoxazole. Data concerning risk factors, clinical features, and therapeutic options for such infections are scarce.

Methods  A case-control study was performed to investigate the risk factors for all types of community-acquired infections caused by ESBL-producing E coli in 11 Spanish hospitals from February 2002 to May 2003. Controls were randomly chosen from among outpatients with a clinical sample not yielding ESBL-producing E coli. The clinical features of these infections were investigated in the case patients. The efficacy of fosfomycin tromethamine and amoxicillin-clavulanate potassium was observationally studied in patients with cystitis.

Results  A total of 122 cases were included. Risk factors selected by multivariate analysis included the following: age older than 60 years; female sex; diabetes mellitus; recurrent urinary tract infections (UTIs); previous invasive procedures of the urinary tract; follow-up in outpatient clinic; and previous receipt of aminopenicillins, cephalosporins, and fluoroquinolones. Urinary tract infections accounted for 93% of the cases; 6% of the patients were bacteremic and 10% needed hospitalization. The cure rate of patients with cystitis was 93% with fosfomycin therapy (all isolates were susceptible); among patients treated with amoxicillin-clavulanate, cure rates were 93% for those with susceptible isolates (minimum inhibitory concentration ≤8 μg/mL) and 56% for those with intermediate or resistant isolates (minimum inhibitory concentration ≥16 μg/mL) (P = .02).

Conclusions  In predisposed patients, ESBL-producing E coli is a notable cause of community-acquired infection, and particularly UTI. Fosfomycin and amoxicillin-clavulanate appear to be effective for cystitis caused by susceptible isolates.

Extended-spectrumβ-lactamase (ESBL)-producing Escherichia coli has emerged worldwide as a significant group of community pathogens.^1,2 This is of clinical importance because E coli is the most common cause of community-acquired urinary tract infections (UTIs), and ESBL-producing isolates are frequently resistant to many of the antimicrobial agents recommended for the treatment of such infections; this is particularly true of oral agents.³ While many studies have reported the microbiological characterization of community-acquired ESBL-producing E coli, published data that are clinically orientated are scarce; the few studies published concerning risk factors or the clinical significance of such infections involved only small samples of cases and were carried out in individual centers.^4-6 As the epidemiologic features of ESBL-producing organisms may show significant local variations, multicenter studies are particularly important to obtain comprehensive knowledge of risk factors for these infections. Likewise, studies that evaluate the efficacy of the few potentially useful oral drugs available for treatment are also needed.

The objectives of our multicenter study were to provide a comprehensive view of the epidemiologic characteristics and clinical relevance of community-acquired infections caused by ESBL-producing E coli and to observationally analyze the efficacy of fosfomycin tromethamine and amoxicillin-clavulanate potassium in the treatment of UTI caused by these organisms.

Eleven public Spanish hospitals, located in 5 geographically distinct areas of Spain at some distance from each other (Barcelona, Madrid, Majorca, Seville, and Valencia), participated in the study. The microbiology laboratories of the hospitals concerned receive samples from hospitalized patients, emergency departments, specialized outpatient clinics, and primary care centers. In total, a population of more than 4 million people is served. Each laboratory processes more than 90% of its samples from its respective health care area.⁷

The risk factors for infections caused by ESBL-producing E coli in outpatients were investigated by a previously used case-control design.⁴ The base population for the case-control study consisted of outpatients in the catchment areas of the participant centers, who had a clinical sample taken and sent for culturing and who were suspected of having a community-acquired infection. The samples might have been obtained in primary care settings, an outpatient clinic, or an emergency department. A case patient was defined as a person whose clinical sample yielded ESBL-producing E coli. For each case patient, 2 controls were randomly chosen (using a computerized method) from among those outpatients with a clinical sample that had been sent for culturing to the same laboratory during the following week, provided that ESBL-producing E coli had not been isolated in the sample. Case patients and controls who had been hospitalized for more than 48 hours when the sample was collected, those admitted to a hospital for longer than 48 hours during the preceding month, and those with ESBL-producing E coli isolated during the previous year were excluded. Case patients were recruited prospectively by daily review of microbiological reports. The study period was from February 2002 through May 2003.

The following variables were collected from all cases and controls using a structured questionnaire: demographic data, underlying diseases, severity of underlying diseases according to the Charlson Comorbidity Index,⁸ history of recurrent UTI, previous admission to any acute or long-term care center (including nursing homes), visits to specialized outpatient clinics, presence of indwelling devices (urinary or vascular catheters), dialysis, specialized home care, ambulatory intravenous therapy, invasive procedures, surgery, and antimicrobial use during the preceding 3 months. The questionnaire had been previously tested⁴ and was filled in by a trained investigator from each center. The data were obtained from medical records and by interviewing the primary care physicians and the patients or their closest relatives. Neither the patient (case or control) nor their physician was informed about the purpose of the study until the end of the interview. The use of antimicrobial agents was determined in several ways; in brief, each patient was asked whether they remembered having received any antimicrobial agents or therapy for any specific infection and whether they remembered receiving specific antimicrobials. Infections were considered as health care associated⁹ if they fulfilled any of the following criteria: hospitalization in acute care center for longer than 48 hours during previous year; residence in a nursing home or long-term care facility; receipt of hemodialysis treatment, antineoplastic chemotherapy, intravenous therapy at home, or specialized home care in the previous 3 months; and major ambulatory surgery or endoscopic procedure performed in the preceding 3 months.

The clinical significance of ESBL-producing E coli was investigated in the case patients. Infection type was classified using clinical and laboratory data. Patients with positive urine cultures (pure culture of a single microorganism and a colony count higher than 10⁵ CFU [colony-forming units]/mL) and clinical signs of UTI (dysuria, urgency, frequency, suprapubic tenderness, or flank pain) were considered as having UTI unless proven otherwise¹⁰; UTI location was defined using clinical criteria.¹⁰ Asymptomatic patients with positive urine cultures, including those with indwelling bladder catheters, were considered to have asymptomatic bacteriuria. Patients with more than 2 episodes of UTI over a period of 6 months were considered to have recurrent UTI. Because the study was observational, it was the attending physician who decided on indications for blood cultures (typically, for feverish patients attended to in emergency departments), other tests, and therapy. The presence of bacteremia, need of hospital admission, and antimicrobial therapy were also recorded.

The efficacy of antimicrobial treatment was evaluated in patients with cystitis caused by ESBL-producing E coli using previously described criteria,¹¹ modified for the present study; patients were followed for 4 weeks after completion of therapy. Patients were considered to be clinically cured if they showed no persistent symptomatic or recurrent UTI. Patients with persistent symptoms after completion of therapy were considered to have persistent symptomatic UTI. Patients were considered to have a recurrent UTI if, after resolution of initial UTI symptoms, they developed recurrent symptoms warranting retreatment. Persistent or recurrent UTI cases were considered to be confirmed by culture if the urine culture performed after completion of therapy yielded ESBL-producing E coli.

The study protocol was approved by the local ethics committees of the participating centers. Because of the observational character of the study, written informed consent was not required.

Only the first isolate per case patient was studied. Extended-spectrum β-lactamase production of isolates was studied in the participating centers according to Clinical and Laboratory Standards Institute (CLSI) recommendations.¹² All isolates were sent to a reference laboratory (Servicio de Microbiología, Hospital Universitario Virgen Macarena, Seville, Spain), where identification to species level was confirmed by the API 20E system (bioMérieux, Marcy l’Etoile, France) and ESBL production was confirmed by broth microdilution according to CLSI guidelines; antibiotic susceptibility was evaluated by microdilution in accordance with CLSI guidelines.¹² Fosfomycin susceptibility was evaluated by the E-test.¹³

We used conditional logistic regression to compute crude odds ratios (ORs) with 95% confidence intervals (CIs). Multivariate analyses were also performed by conditional logistic regression.¹⁴ Variables introduced in the multivariate analysis included those with a statistically significant association in the univariate comparisons and were selected using a stepwise backward process. Interactions between paired variables were investigated. Incidence rates were calculated by dividing the number of cases in each area by its catchment's populations; the result was extrapolated for 1 year. Data were analyzed using the STATA 9.2 (StataCorp, College Station, Texas) software package.

During the study period, ESBL-producing E coli was isolated from 122 outpatients, all of whom were included. The estimate population-based pool incidence of community-acquired infections due to these organisms was 2.2 cases per 100 000 population per year, with a range between 1.5 and 3.9 for the 5 areas studied.

Extended-spectrum β-lactamase–producing E coli was isolated from urine culture in 113 cases (92%), blood culture in 6 (5%), abscess in 2 (2%), and peritoneal fluid in 1 (1%). The characteristics of the case patients are given in Table 1; 57 patients (47%) had a chronic comorbid condition, and 41 (34%) were considered to have health care–associated infections: 27 patients had been admitted to an acute care hospital, 27 had undergone some invasive procedure (4 had undergone surgery; 4, biliary-tract endoscopy; and 19, invasive procedure of the urinary tract), 3 were nursing home residents, 1 received chronic hemodialysis, and 1 was a transplant recipient. Finally, 21 patients (17%) had no chronic underlying condition, no recurrent UTI, and their infections were not health care associated.

To analyze the risk factors, the 122 case patients were compared with 242 control patients (for 2 cases, only 1 control was found). Crude analysis showed age older than 60 years, female sex, certain underlying diseases, recurrent UTI, health care–associated infection, indwelling bladder catheter, invasive urinary tract procedure, and previous antimicrobial use to be associated with ESBL-producing E coli (Table 1). We performed 3 multivariate models (Table 2). In the first (general) model, we included variables such as “health care–associated infection” and “previous antimicrobial use” but not their individual components (such as different forms of previous health care contact or specific antimicrobials); independent risk factors selected in this model were age older than 60 years, female sex, diabetes mellitus, recurrent UTIs, health care–associated infection, and previous antimicrobial use. In the second model, we introduced specific antimicrobial groups instead of “previous antimicrobial use,” and the same variables as the first model plus 3 antimicrobial groups (aminopenicillins, cephalosporins, and fluoroquinolones) were selected. In the third model, we introduced “previous hospital admission,” “nursing home residence,” “home care,” and “invasive procedures” instead of “health care–associated infection,” and the same variables as the first model were again selected, although only 1 specific type of previous health care contact (invasive urinary tract procedure) showed an independent association with ESBL-producing E coli. Considering the risk factors found in the general model, only 4 cases (3%) had no risk factors, 20 (16%) had 1, 41 (34%) had 2, and 57 (47%) had 3 or more.

Extended-spectrum β-lactamase–producing E coli caused UTI in 113 patients (93%), acute cholangitis in 3 (3%), and secondary peritonitis, cellulitis, and pneumonia in 1 patient each (2%, respectively). Among patients with UTI, 73 had cystitis (60% of the whole series), 33 asymptomatic bacteriuria (27%), 5 pyelonephritis (4%), and 2 prostatitis (2%). Seven patients (6%) were bacteremic (3 with pyelonephritis, 3 with cholangitis, and 1 with cellulitis), and 12 (10%) required hospitalization as a consequence of the infection; none died. Resistance rates of the isolates to different antimicrobial agents are given in Table 3.

Of the 73 patients with cystitis, 37 were treated with amoxicillin–clavulanate potassium (500 mg/125 mg every 8 hours, for 5-7 days); 31 of them were considered clinically cured (84%), and the cure rate was lower in patients with higher minimum inhibitory concentration (MIC) (17 of 18 [94%] in those with an MIC of ≤4 μg/mL; 9 of 10 [90%] in those with an MIC of 8 μg/mL; 3 of 4 [75%] in those with a MIC of 16 μg/mL; and 2 of 5 [40%] in those with an MIC of ≥32 μg/mL). The difference was statistically significant between those with susceptible isolates (MIC ≤8 μg/mL) and those with intermediate or resistant isolates (MIC ≥16 μg/mL) (26 of 28 [93%] vs 5 of 9 [56%]; P = .02, Fisher test). Twenty-eight patients were treated with fosfomycin tromethamine (3 g, 1 dose), and 26 (93%) were clinically cured; persistent or recurrent UTI was microbiologically confirmed in the 2 patients in whom treatment failed. The number of cases for other types of infections or antimicrobial agents used was too low to merit specific analysis.

Extended-spectrum β-lactamase–producing E coli and community-acquired methicillin-resistant Staphylococcus aureus are the 2 most relevant antimicrobial-resistant organisms to have emerged in recent years as a cause of community-acquired infections. Community-acquired infections caused by ESBL-producing E coli have been reported worldwide,^{1,2,4-6,15,16} including the United States,¹⁷ and continue to increase, although there are regional differences in rates.¹⁶ The rates found in our study, ranging from 1.5 to 3.9 cases per 100 000 population per year in the different areas, should be interpreted as minimum rates, since participating laboratories process most (>90%), but not all, of the samples in their catchment area.⁷ The treatment of infections caused by such isolates is a challenge because they are resistant to penicillins, cephalosporins (with the exception of cephamycins), and aztreonam and are frequently resistant to trimethoprim-sulfamethoxazole and fluoroquinolones.³ To our knowledge, ours is the first multicenter study that includes a detailed clinical and epidemiological investigation and may provide a comprehensive view of the problems posed by ESBL-producing E coli as community pathogens. Although the clonal relatedness of the isolates was not investigated in this study, previous studies have shown that most ESBL-producing E coli isolated from outpatients in Spain were clonally unrelated during that period.^4,18

We have only found 3 published case-control studies investigating the risk factors associated with community acquisition of ESBL-producing E coli; all 3 studies were carried out in individual hospitals in Spain and Israel and used different methods.^4-6 In addition, a population-based surveillance study has been carried out in Canada.¹⁵ Finally, another study in Israel investigated the risk factors for fecal carriage of ESBL-producing organisms at hospital admission.¹⁹ Our data confirm some of the risk factors found in those previous studies, such as older age,^4,5,15 diabetes mellitus,⁴ recurrent UTI,⁴ and previous use of fluoroquinolones^4,5 or cephalosporins.^5,6 Herein, we report for the first time, to our knowledge, that previous receipt of aminopenicillins is also a risk factor. The implication of aminopenicillins as a risk factor may be related to the fact that all ESBL-producing E coli are, by definition, resistant to ampicillin and, in our series, 29% were also resistant to amoxicillin-clavulanate; in Spain, the frequencies of ampicillin and amoxicillin-clavulanate resistance in UTI caused by E coli are approximately 60% and 9%, respectively.²⁰ Thus, aminopenicillin use in previously colonized patients may select not only for aminopenicillin-resistant strains but also for ESBL producers.

We chose not to select control patients from among patients with infections caused by non–ESBL-producing E coli, since this would have overestimated the risk associated with previous antimicrobial use.²¹ Since recurrent UTI, female sex, older age, and invasive urinary tract procedures are associated with an increased risk of having an UTI,^22,23 their specific association with ESBL-producing E coli isolates might have been overestimated. However, our results apply to patients suspected of having a community-acquired infection and who have a sample sent for culturing.

Health care–associated infections were associated with an increased risk in our analysis. In some previous studies, recent hospital admission was found to be a risk factor^4,5; however, we hypothesize that ESBL-producing E coli are usually truly community-borne and not acquired during a previous admission and that previous admission could be a surrogate marker for other variables. Thus, we specifically studied different types of previous health care contact; when invasive urinary tract procedure was included, previous admission was no longer associated with ESBL-producing E coli.

Of all risk factors found, only previous exposure to antimicrobial agents is potentially susceptible to intervention. Because many infections caused by ESBL-producing isolates occur in patients with recurrent UTI and considering that fosfomycin and nitrofurantoin are frequently active against ESBL-producing isolates, it should be studied whether therapy (or prophylaxis when needed) of recurrent UTI with these agents is helpful in avoiding the selection pressure posed by fluoroquinolones, aminopenicillins, and cephalosporins.

The distribution by type of infections was as expected for E coli. Lower UTI accounted for the majority of cases, while some 10% of patients presented complicated or invasive infections and needed hospitalization. It is noteworthy that 6% of the cases were bacteremic, as we have recently shown that appropriate therapy in patients with bacteremia due to these organisms is clinically important.²⁴ The frequency of intra-abdominal infections in which ESBL-producing E coli are involved is probably underestimated, since samples from patients with community-acquired secondary peritonitis are not routinely sent for culturing. In a recent multinational study, 9% of 2620 E coli strains isolated from intra-abdominal infections were ESBL producers.²⁵ The impact of ESBL production in such infections has yet to be determined.

With regard to the treatment of cystitis caused by ESBL-producing E coli, to our knowledge, no randomized controlled trial has been performed to date. The susceptibility profile of our isolates confirms that oral alternatives for such infections are considerably limited; apart from cephalosporins, to which these isolates are inherently resistant, resistance to ciprofloxacin and trimethoprim-sulfamethoxazole occurred in 64% and 57%, respectively, of the isolates. Such high prevalence of resistance to these antimicrobial agents, particularly fluoroquinolones, is a well-known feature of ESBL-producing E coli.^3,4,15,18 Only fosfomycin shows consistent in vitro activity.²⁶ Our results and those of a previous study²⁷ suggest that this might be the preferred therapy. According to in vitro data, nitrofurantoin could be an option because approximately 70% of isolates are susceptible,²⁸ although the drug is not well tolerated by some patients and should not be used in patients with renal insufficiency. Another option could be amoxicillin-clavulanate; in our series, only 29% of the isolates were resistant to amoxicillin-clavulanate in vitro. However, the clinical efficacy of β-lactam/β-lactamase inhibitor combinations has been questioned.³ Our data suggest that amoxicillin-clavulanate might also be useful for cystitis caused by isolates with MIC values in the susceptibility range (MIC ≤8 μg/mL). Unfortunately, resistance to amoxicillin-clavulanate among ESBL-producing E coli is frequent in some areas.²⁹

In conclusion, our study confirms that ESBL-producing E coli are a notable cause of community-acquired infections in predisposed patients. Should this resistance mechanism spread further in E coli, community-borne ESBL may become a public health concern over the next few years.

Correspondence: Jesús Rodríguez-Baño, MD, PhD, Sección de Enfermedades Infecciosas, Hospital Universitario Virgen Macarena, Avda Dr Fedriani, 3, 41009 Seville, Spain (jrb@nacom.es)

Accepted for Publication: March 9, 2008.

Author Contributions: Dr Rodríguez-Baño had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Rodríguez-Baño, Alcalá, Cisneros, Cantón, and Pascual. Acquisition of data: Rodríguez-Baño, Alcalá, Cisneros, Grill, Oliver, Horcajada, Tórtola, Mirelis, Navarro, Cuenca, Esteve, Peña, Llanos, Cantón, and Pascual. Analysis and interpretation of data: Rodríguez-Baño and Alcalá. Drafting of the manuscript: Rodríguez-Baño, Alcalá, Cisneros, Grill, Tórtola, Navarro, and Cantón. Critical revision of the manuscript for important intellectual content: Oliver, Horcajada, Peña, Llanos, Cantón, and Pascual. Statistical analysis: Rodríguez-Baño. Obtained funding: Rodríguez-Baño, Esteve, Peña, and Pascual. Administrative, technical, and material support: Alcalá, Grill, Tórtola, Mirelis, Navarro, Cuenca, and Esteve. Study supervision: Rodríguez-Baño, Cisneros, Horcajada, and Cantón.

Financial Disclosure: Dr Rodríguez-Baño has been a consultant for Chiron, Wyeth, Merck, and Pfizer and has served as speaker for Wyeth, Merck, Pfizer, and GlaxoSmithKline. Dr Pascual has been a consultant for Merck and Pfizer; has served as speaker for Wyeth, Astra-Zeneca, Merck, and Pfizer; and has received research support from Merck, Pfizer, and Wyeth. Dr Cantón has been a consultant for Chiron; has served as speaker for Wyeth, Sanofi-Aventis, and GlaxoSmithKline; and has received research support from Sanofi-Aventis and Wyeth.

Funding/Support: This study was supported by grant REIPI C03/14 from the Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III–FEDER, Spanish Network for the Research in Infectious Diseases; grant REIPI RD06/0008 from the Spanish Network for the Research in Infectious Diseases; grant PI070190 from Fondo de Investigación Sanitaria; and grant 75/04 from Junta de Andalucía.

Role of the Sponsors: The funding organizations had no role in the design and conduct of the study; in the collection, management, analysis, and interpretation of the data; nor in the preparation, review, or approval of the manuscript.