Introduction

Enterococcus spp. are ubiquitous bacteria, as part of the intestinal microbiota, and may colonize the urinary tract. The most common species are E. faecalis and E. faecium. A rising incidence of invasive enterococcus infection has been reported since the past decades [1,2,3]. In France, the proportion of E. faecalis isolated from healthcare-associated infections (HCAI) is higher in 2017 than in 2012 (6.5% versus 4.6% respectively), being now the third pathogen after Escherichia coli and Staphylococcus aureus [4]. Accordingly, Enterococcus spp. are a significant cause of bacteraemia in the healthcare setting, accounting for 14% of all cases, and appear to be an opportunistic pathogen [5, 6].

The major sources of enterococcal invasive infections are the urinary and digestive tracts [7,8,9]. Enterococcal bacteraemias are also observed in patients exposed to invasive devices such as central venous line. Thus, depending on study designs, enterococcal bacteraemia are considered to be an HCAI in 7 to 23% of all cases [10]. Furthermore, between 8 and 25% of patients with enterococcal bacteraemia develop an infective endocarditis [1, 11, 12]. These data imply that enterococcal bacteraemias are associated with a high rate of unfavourable outcome, the rate of death being measured over 20% [1, 8, 9].

The antimicrobial treatment of enterococcal invasive infections has to take into account the natural resistance of the bacteria to cephalosporins, and an increasing prevalence of acquired resistance to penicillins and aminoglycosides is observed [3, 10]. E. faecalis, the main species isolated in human medicine remains most of the time susceptible to amoxicillin [7, 8]. E. faecium bacteraemias are associated with a higher mortality than E. faecalis ones, probably related to a lower efficacy of penicillins and subsequent inappropriate empirical antibiotic therapy [8, 9, 13].

The unfavourable outcome of enterococcal bacteraemia is probably caused by several factors, such as the burden of comorbidities and the inappropriate and/or delayed efficient therapy [14,15,16]. Moreover, another study also showed that vancomycin-susceptible E. faecium bacteraemias were associated with higher in-hospital mortality and prolonged length of stay than E. faecalis bacteraemias [17]. To the best of our knowledge, no published study has described the antibiotic therapy for enterococcal bacteraemia in real life, with an “antimicrobial stewardship approach”, evaluating the quality of the empirical therapy as well as its reassessment. Our aim was to describe such approach and to determine the risk factors associated with a favourable outcome.

Methods

Study setting and design

We conducted a retrospective multicentre cohort study of patients with enterococcal bacteraemia in several institutions in France working in a multidisciplinary professional network. The participating institutions worked together to develop antibiotic stewardship programme through common audits and clinical researches [18,19,20].

The French National Health Agency promotes antibiotic audits and patients, or relatives gave their written consent for computerizing their personal data for hospitalization purpose and potential clinical research.

Population

Patients were identified through the laboratory’s electronic database, which extracted all positive blood cultures for Enterococcus spp. between 1 January 2016 and 31 December 2016. We included all patients with at least one positive blood culture for Enterococcus spp. and for whom full records were available. Patients with multiple blood cultures positive for the same Enterococcus during the same hospitalization were included only once. Patients with polymicrobial blood cultures, including Enterococcus spp., were excluded, as we wanted to determine the quality of the antibiotic treatment against enterococcal infections.

Clinical characteristics

Only patients 18 years old or older were included. Patients’ data were gathered by reviewing hospital electronic records, and, when incomplete, by reviewing stored hard copy medical records. We reported the following: demographic data, comorbidities, immunosuppressive treatment, reasons for hospital admission, surgical procedures and invasive device exposure, source of bacteraemia as defined by the doctor in charge, creatinine value at the closest measurement to bacteraemia date, intensive care unit requirement and/or dialysis, hospital stay duration and final outcome. Data regarding antimicrobial treatments included antibiotic choice and treatment duration.

We used the definition proposed by Friedman et al. in 2002 for healthcare-associated bloodstream infection [21, 22]. Thus, the bacteraemia was considered as HCAI when diagnosed ≥ 48 h after hospital admission or for patients with regular hospital visits (haemodialysis, chemotherapy in the 30 days before the bloodstream infection), patients receiving intravenous therapy at home or wound care, within 90 days after two or more days of hospitalization, or for patients residing in a nursing home on a long-term care facility.

An enterococcal bacteraemia diagnosed within 48 h of hospital admission was considered to be community acquired.

Microbiological characteristics

The microbiological data included the following: enterococcal species causing bacteraemia, antibiotic susceptibility results and other relevant culture results within 30 days of the first positive blood culture.

Blood cultures were collected directly during the venepuncture procedure using aerobic (Bact/AERT® FA Plus, Biomérieux, France) and anaerobic (Bact/AERT® FN Plus, Biomérieux, France) blood culture bottles. Blood cultures were sent to the laboratory and processed with an automated Bactalert 3D system (BioMérieux, France). The date of the positivity of the blood culture was the date the blood culture was sampled.

Antimicrobial therapies

Empirical antimicrobial therapy was defined by those antibiotic(s) prescribed in front of a septic patient and/or those linked to the first notification of Gram positive read on blood culture.

Documented antimicrobial therapy was defined as the antibiotic(s) prescribed after the definitive bacterial identification and determination of drug susceptibility.

Efficient antimicrobial therapy was defined as any therapy containing at least one antibiotic compound efficient against Enterococcus spp.: amoxicillin, amoxicillin/clavulanic acid, piperacillin, piperacillin/tazobactam, imipenem, meropenem, vancomycin, daptomycin, linezolide, tigecycline, combined or not with gentamicin.

Inappropriate antimicrobial therapy was defined by any other antibiotic compounds, including cotrimoxazole or fluoroquinolones, combined or not with gentamicin.

Dosage and duration of treatment were not included in the definition of “appropriate” antimicrobial therapy, as it may be a source of disagreement between physicians and/or ID specialists.

Outcome

Primary outcome was the clinical consequence of the antibiotic treatment. We were interested in short-term unfavourable outcome defined by an intensive care requirement and/or in-hospital death after positive blood culture, during hospitalization. As a secondary analysis, we also evaluated the outcome at day 7 after positive blood culture. As long-time follow-up was insufficient for most of the patients, we decided therefore not to study long-term outcome.

Statistical analyses

First, we performed descriptive analysis, using absolute values and proportions for qualitative variables, and mean (with standard error) or median (with interquartile range (IQR)) for quantitative variables. Then, we performed a univariate analysis, comparing variables with a Chi-square test, or a Fisher test, and a Student t test, or a Wilcoxon–Mann–Whitney test where needed. Finally, we performed a multivariate analysis. All variables with a p value < 0.2 in univariate analysis were entered into the model.

Results

Population

The centres were two public general hospitals, one teaching hospital and 3 private institutions. The private institutions were clinics with 283, 198 and 225 beds. They had both medical and surgical wards and they all had an Emergency Department and an Intensive Care Unit. Four institutions had a dialysis centre.

A total of 131 patients presented monomicrobial enterococcal bacteraemia during the 1-year study period. Their main clinical characteristics are detailed in Table 1. The most frequent portal of entry for bacteraemia was the urinary tract (60/131; 46%), the digestive tract (21/131; 16%) and intravascular catheter (18/131; 14%). For 25/131 patients (19%), the portal of entry was unknown. Infection was a HCAI in 71 cases (54%), notably related to urologic surgery (n = 16; 23%).

Table 1 Main patients’ characteristics
Full size table

One hundred ten infections (84%) were caused by E. faecalis and 13 cases (10%) were caused by E. faecium. Nine infections (7%) were due to other Enterococcus spp.: E. durans (5/9) and E. dispar, E. avium, E. gallinarum and E. hirae (1/9 cases each).

In vitro susceptibility indicated that 7/131 strains (5%) were amoxicillin resistant (E. faecium strains exclusively), none was vancomycin resistant and 26/131 strains (20%) were gentamicin resistant (high level of resistance).

Empirical antimicrobial therapy

Twenty patients (15%) did not receive any antimicrobial therapy before the microbiological identification.

Among the 111 patients who received an empirical antimicrobial therapy, 45/111 (41%) benefited of a single antibiotic using 10 different options, 55/111 (50%) received simultaneously two antibiotics with 19 different options and the other 11/111 patients (10%) received simultaneously three antibiotics using 10 different options (see Table 2). According to definitions, an efficient empirical antibiotic therapy was prescribed in 63/131 (48%) cases.

Table 2 Antibiotic therapy for enterococcal bacteraemia
Full size table

Documented antimicrobial therapy

Four patients (4/131; 3%) did not receive any antimicrobial therapy after the availability of bacterial identification and determination of drug susceptibility. Also, despite the latter, for nine other patients, no antibiotic reassessment was reported.

Among the 118 patients who received a documented antimicrobial therapy, 84/118 (71%) received a single antibiotic using 11 different drugs, 32/118 (27%) received simultaneously two antibiotics using 15 different drugs and 2/118 (2%) patients received simultaneously three antibiotics (see Table 2). Out of 118, 98 (83%) patients received an efficient therapy; the efficient antimicrobial therapy was amoxicillin in 60/98 (61%) cases. In contrast, 20/118 (17%) patients did not benefit from an active antibiotic against Enterococcus spp.

Patient outcome

Among the 131 patients, 30/131 (23%) presented a short-term unfavourable outcome defined by an admission to intensive care and/or death. The median of hospitalization lengths of stay was 12 days IQR [5.5–18].

In univariate analysis, we found that an appropriate documented antimicrobial therapy was associated with a trend towards a better short-term outcome: p = 0.100 (see Table 1). Also, in univariate and multivariate analyses, the prescription of documented antimicrobial therapy that contains amoxicillin was significantly associated with a favourable short-term outcome: p < 0.001 (see Table 3). With a 7-day cut-off as definition of unfavourable outcome, documented antimicrobial therapy that contains amoxicillin was still significantly associated with a favourable short-term outcome: p = 0.016 (see Table 4).

Table 3 Multivariate analysis for variables associated with the short-term outcome during hospitalization
Full size table
Table 4 Multivariate analysis for variables associated with the short-term outcome at day 7
Full size table

Discussion

The antimicrobial stewardship programme aims to reduce misuse (and overuse) of antibiotic therapy to fight against the emergence of multidrug-resistant bacteria [23]. Focusing on enterococcal bacteraemia, we found that a huge diversity of antibiotic therapies was prescribed for these septic patients. Moreover, 50% and 17% of the patients did not benefit from an efficient empirical antimicrobial and efficient documented therapy respectively. Finally, we found that using amoxicillin during enterococcal bacteraemia was associated with a better short-term outcome (p < 0.001).

Our study has several limitations. This is a retrospective study and we faced some missing data, notably in the public teaching hospital in which the electronic patient records did not provide a date for the laboratory results, therefore limiting our effort to determine the reasons for treatment modifications. Also, antibiotic therapy durations were recorded for inpatient only. Lastly, we included all the patients with at least one positive blood culture. As suggested in the literature, Enterococcus species may be a contaminant in 10–15% of blood cultures [24, 25]. Accordingly, co-isolation with a skin organism suggests a contamination rather than a real blood stream infection [26].

Our study population was similar to previous reports concerning the rate of enterococcal bacteraemia observed in healthcare settings [4,5,6]. Of note, none of the isolates was vancomycin resistant in our study cohort. In 2017, in France, the percentage for vancomycin resistance in E. faecium was 0.8% while it was 14.9% in the European Union [27]. Indeed, the microbiological epidemiology is significantly different between countries, with national percentages of vancomycin resistance in 2017 ranging from 0 to 43.9% [27]. In contrast, American works report 82% and 9% of vancomycin-resistant E. faecium and E. faecalis strains respectively [28, 29].

Very few studies have focused on antimicrobial treatment of enterococcal bacteraemia. The diversity of antibiotics used for patients with bacteraemia has been already reported, suggesting that local or national guidelines are not respected in clinical practice [13, 16, 19]. Our data and other previous studies show that repeated audits with feedback information to clinicians are still major tools in antimicrobial stewardship programme.

In our work, nearly 50% of patients received an efficient empirical antimicrobial therapy, which is much more than elsewhere: it was 18% in Pinholt’s study [1], but our efficient treatment list was broader than in the latter, considering imipenem and meropenem as efficient against Enterococcus spp. Antimicrobial reassessment enhanced amoxicillin use (from 6 to 46%), and decreased the use of carbapenem (from 9 to 3 patients), vancomycin (from 9 to 5 patients), third-generation cephalosporin (from 41 to 12 patients) and aminoglycosides (from 38 to 10 patients). However, the rate of patients who benefited from amoxicillin (46%) as a documented antimicrobial therapy remains insufficient. We do not have any explanation for this low level of use of this reference therapy recommended for susceptible enterococcal species, its bactericidal effect being sufficient to treat this infection [10]; it is also the cornerstone of enterococcal endocarditis [30]. Physicians might have been scared to treat bacteraemia with a narrow-spectrum antibiotic on frail patients. Yet, in our experience, the traceability of the information about the positivity of the blood culture is often transmitted by the physician from the Department of Clinical Microbiology to the nurse and not to the physician in charge of the patient. Therefore, as we cannot assert that the physicians had seen the bacteriological results, our results suggest that microbiological information was not fully used or even seen.

Lastly, the mortality rate was 15%, which is similar to other studies that worked on vancomycin-susceptible enterococcal bacteraemia (from 13 to 31%) [8, 31]. McBride et al. found that an inefficient empirical antibiotic therapy against the enterococcal isolate was not associated with increased mortality, probably due to the low virulence of Enterococcus spp. [8]. In contrast, all other studies showed that inefficient documented antimicrobial therapy was associated with a higher mortality [8, 13, 32, 33].

Conclusion

A high level of inappropriate antimicrobial therapy, for both empirical and documented therapies, was observed during enterococcal bacteraemia. Among efficient antibiotics, amoxicillin use was associated with a better outcome. Since Enterococcus spp. are increasingly recognized as a causative agent of HCAI, the antimicrobial stewardship team should include enterococcal infections in their targets.