Incidence, clinical characteristics, and outcomes of nosocomial Enterococcus spp. bloodstream infections in a tertiary-care hospital in Beijing, China: a four-year retrospective study

Background Enterococcus spp. are the common cause of nosocomial bloodstream infections (BSIs) with high morbidity and mortality. The purpose of this study was to characterize the incidence, clinical and microbiological features, and mortality of nosocomial enterococcal BSIs at a large Chinese tertiary-care hospital in Beijing, China. Methods A retrospective cohort study on adult patients with nosocomial BSIs due to Enterococcus spp. was performed between January 1, 2012, and December 31, 2015 at the Chinese People’s Liberation Army General Hospital. Patients’ data were gathered by reviewing electronic medical records. Results A total of 233 episodes of BSI due to Enterococcus spp. occurred among 224 patients during these 4 years. The overall incidence was 3.9 episodes per 10,000 admissions. Enterococcus faecium (E. faecium) was the major pathogen (74%, 95% CI 68–80%), followed by Enterococcus faecalis (E. faecalis) (20%, 95% CI 15–25%). E. faecium showed higher antimicrobial resistance than E. faecalis. The 30-day mortality of nosocomial enterococcal BSI was 24% (95% CI 18–29%). Predictors for mortality included the Acute Physiology and Chronic Health Evaluation II (APACHE II) score, Charlson comorbidity index (CCI), impaired renal function, prior use of immunosuppressive agents, and appropriate empirical antimicrobial treatment. Conclusions This study emphasizes that Enterococcus spp. were major pathogens for nosocomial BSIs and associated with high mortality. Appropriate empirical antimicrobial treatment can improve outcomes. Vancomycin is the best choice for patients with E. faecium BSIs. Penicillins, aminoglycosides, fluoroquinolones, and vancomycin can be considered for patients with E. faecalis BSIs.


Background
BSIs can cause high mortality and result in heavy social and economic burdens. As one of the common grampositive pathogens, Enterococcus spp. are the fourth leading cause of BSIs in North America [1], and are rated as the third most prevalent pathogens of nosocomial BSIs in the United States [2], accounting for >9.0% of the BSIs. Several multicentre studies from China and Japan also reported that Enterococcus spp. were the fourth most common BSI pathogens [3][4][5]. And a systematic review focused on community-acquired BSIs in south and southeast Asia showed that Enterococcus spp. were the third most prevalent gram positive bacteria [6]. The crude mortality rates of enterococcal BSIs ranged between 21.4% and 64.2% [7][8][9][10][11]. Risk factors for developing enterococcal BSIs include the presence of comorbidities such as malignancy, diabetes; more severe illness; invasive devices, such as a central intravenous catheter; complicated surgery; solid organ transplantation; and hematopoietic stem cell transplantation [2,[12][13][14][15][16].
Enterococci are innately resistant to cephalosporins, and the commonly used antimicrobial agents such as fluoroquinolones and carbapenems are not recommended. The increasing prevalence of acquired resistance to penicillins and aminoglycosides has been observed in many countries [17,18]. Vancomycin-resistant enterococci (VRE) are a great challenge in clinical treatment because there are limited bactericidal options to choose [19].
E. faecalis and E. faecium are the most frequently isolated species, and incidences of both have shown a rising trend, especially for E. faecium [11,20]. E. faecium has significantly higher antibiotic resistance rates than E. faecalis and may lead to more serious disease and worse outcome [11,21,22].
Although there has been much research on enterococcal BSI, the incidence, species distribution, clinical features, and prognosis vary in different periods and different regions. In addition, nosocomial BSI is different from community-acquired BSI in several aspects, including a poor prognosis [11]. Little data exists regarding nosocomial enterococcal BSIs in China. The purpose of the present study was to characterise the incidence, clinical, and microbiological features, and to identify the predictors of crude mortality in patients with enterococcal BSIs at our hospital. The clinical and microbiological characteristics of E. faecalis and E. faecium were also compared.

Study design, hospital setting, and patients
We performed a retrospective cohort study on adult patients with nosocomial BSIs due to Enterococcus spp. between January 1, 2012, and December 31, 2015, at the Chinese People's Liberation Army General Hospital (PLAGH), a 2200-bed tertiary-level healthcare hospital in Beijing, China. It is one of the biggest comprehensive hospitals in China, with medical, health care, teaching, and scientific research that serves national military and nonmilitary personnel from across the country.
Eligible patients included all patients aged ≥18 years with at least 1 positive blood culture for Enterococcus spp. In patients with persistent BSIs caused by the same organism, only the first episode was included. If the patients had 2 or more separate BSIs, each infection was considered individually. All patients were identified by searching the real-time nosocomial infection surveillance system (RT-NISS) [23]. This platform utilises data from electronic medical record systems, such as hospital stay, temperature changes, and microbiology results with the application of clinically validated algorithms to identify and classify all of the patients' infections. RT-NISS was developed by the Infection Management and Disease Control Department of PLAGH.

Data collection
The patients' data were gathered by reviewing electronic medical records. We recorded the demographic data including age and gender. The clinical data collected included: body mass index (BMI), underlying diseases, the CCI score, the APACHE II score in the first 24 h following the onset of BSIs, the hospitalisation wards, previous exposures (prior hospital stay, previous treatment such as surgical procedures, immunosuppressive agents, chemotherapeutic agents, total parenteral nutrition, mechanical ventilation, renal replacement therapy, antibiotics, or invasive devices within 30 days prior to BSIs), treatment (antibiotic choice), and outcomes (length of hospital stay, and all-cause mortality at 7 days and 30 days). The microbiological data collected included: species of Enterococcus, likely source of BSIs (identified by treating doctors and/or physicians of the Infection Management and Disease Control Department), whether cultures were polymicrobial, and antimicrobial susceptibility results. We collected the annual admission data to calculate incidence rates, which are expressed as the number of BSI episodes per 10,000 hospital admissions.

Definitions
The diagnosis of enterococcal BSI should meet the following criteria: 1)isolation of Enterococcus spp. from one or more blood cultures; and one of the following 2) fever (>38°C), chills, or hypotension; and 3) eliminated the possibility of contamination during the collection and cultivation of blood samples [24]. Nosocomial BSI was defined as the first positive blood culture obtained ≥48 h after hospital admission and with no evidence of infection at admission [15,24]. An episode was defined as the positive isolation of enterococcus from at least one blood culture sample from a patient and without the prior blood culture isolating the same bacteria within the previous 30 days [2,25]. Onset of BSIs was defined as the date when the blood culture was collected. Polymicrobial BSIs were defined as 2 or more clinically important organisms isolated from 1 single blood culture sample or different blood culture samples within 48 h. Appropriate antimicrobial treatment was defined as an active antimicrobial choice and an adequate dosage within 5 days from the onset. Empirical antimicrobial therapy was defined as the antimicrobial agents given within 24 h after the onset of BSIs. Active agents were confirmed according to the antimicrobial susceptibility test.

Identification and antibiotic susceptibility testing
Blood was cultured using BacT/ALERT 3D system (Becton-Dickinson, Sparks, MD, USA) in the microbiology laboratory. Species identification was performed using the VITEK 2 system (BioMérieux, Marcy 1′Étoile, France). Antibiotic susceptibility testing was performed using the VITEK 2 system or the Kirby-Bauer Disk Diffusion method (Oxoid, UK) according to the recommendations proposed by the Clinical and Laboratory Standards Institute (CLSI).

Statistical analysis
Categorical variables were expressed as frequency counts and percentages with 95% confidence interval (95% CI). Continuous variables were expressed as median and interquartile ranges (IQRs). Comparison of categorical variables was performed using Pearson's chi-squared test or Fisher's exact test, and comparison of continuous variables was performed using the Mann-Whitney U test. In order to identify the risk factors associated with mortality, a multivariate logistic regression model with backward method was generated to control the effects of confounding variables. Variables statistically related (p < 0.10) to mortality in the univariate logistic analyses were chosen to build the multivariate model. Results with a 2-tailed p-value < 0.05 were considered to be significant. All statistical analyses were run using SPSS 20.0 software (IBM Corp., Armonk, NY, USA).

Incidence and species distribution
In total, 233 episodes of nosocomial BSIs caused by Enterococcus spp. occurred among 224 patients during the 4-year study period. Of the 224 patients who developed enterococcal BSIs, 6 had infections with 2 different species >30 days apart and 3 had 2 infections with the same species >30 days apart. Seven patients who had infections with 2 different species of enterococcus isolated within 48 h were excluded because of the interpretative problem.
The overall incidence of nosocomial enterococcal BSIs was 3.9 episodes per 10,000 admissions and the rate fluctuated from 3. . The overall incidence rates of E. faecium and E. faecalis were 2.9 episodes and 0.8 episodes per 10,000 admissions, respectively. Incidence of E. faecalis BSI showed a rising trend, while that of E. faecium was on the decline, as shown in Fig. 2.
A comparison of nosocomial BSIs according to E. faecium and E. faecalis in demographics and clinical characteristics is shown in Table 1. Patients with E. faecalis BSIs presented a higher BMI, but the median was in the normal range. Patients with E. faecium BSIs showed a higher CCI score than those with E. faecalis BSIs (median, 3 vs 2, p = 0.014). The median APACHE II score of patients with E. faecium BSIs was higher than in those with E. faecalis BSIs, but these were not statistically significant.
Before obtaining the report of antibiotic susceptibility test reports, 98 (43%, 95% CI 37-50%) patients were treated with effective antibiotics. Patients with E. faecalis BSIs were more likely to get effective treatment by the empirical use of antibiotics (57% vs 37%, p = 0.018).
Univariate analyses of predictors for 7-day mortality and 30-day mortality are shown separately in Table 3. The 7-day mortality was associated with pneumonia, impaired renal function, CCI, APACHE II score, ICU residence, prior exposure to immunosuppressive agents and central intravenous catheter, and appropriate empirical antimicrobial treatment. The predictors for 30-day mortality were similar to those for 7-day mortality, except for surgery and renal replacement therapy prior to the onset of BSI within 30 days.
In the multivariate logistic regression model (

Discussion
This study focused on the incidence and characteristics of nosocomial enterococcal BSI in one of the biggest comprehensive hospitals in China. Two multicentre studies in China reporting on the species distribution and antibiotic resistance of clinical isolates from blood cultures showed that Enterococcus spp. were the fourth most common pathogens [3,4]. But, these 2 studies did not involve the species distribution, clinical information, and prognosis. A single-centre study published recently,  [26]. But the sample size (64 episodes) was smaller than our study, and all nosocomial, health care associated, and community acquired enterococcal BSIs were included. In our 4-year study, the incidence rate of nosocomial enterococcal BSIs fluctuated from 3.3 to 4.4 episodes per 10,000 admissions. This is similar to an earlier study conducted from 1995 to 2002 at 42 US hospitals [2], but is higher than 2 recent reports from Denmark (1.96/ 10,000) and Canada (0.69/10,000) [11,15]. But these were both population-based studies. To our knowledge, there are no such reference data about the incidence of nosocomial enterococcal BSIs in China.
In accordance with other studies, patients with enterococcal BSIs were commonly associated with complications, such as malignant, cardiovascular diseases, diabetes mellitus, or chronic kidney disease [11,15,22,27]. We      Variables entered into multivariable logistic regression model of risks factors for 30-day mortality found that pneumonia was also a major comorbidity of nosocomial enterococcal BSIs at our hospital. E. faecalis was reported to be the most common pathogen for nosocomial enterococcal BSIs in most investigations, and the ratio of E. faecium was no more than 50% [1,11,20,28]. However, a constant increase in E. faecium BSI rates was observed [11,29], and the incidence of E. faecium BSIs exceeded E. faecalis BSI in 2009 in a 14-year study conducted at a Swiss tertiary hospital [29]. At our hospital, the incidence of nosocomial E. faecium BSIs was significantly higher than that of E. faecalis BSIs (2.9/10,000 vs 0.8/10,000). The total vancomycin resistance rate of all isolates was 3%, and 4% for E. faecium in our study. All of the vancomycinresistant isolates were E. faecium. This vancomycin resistance rate is higher than those in the Danish study (1.9%) [11], but lower than many studies in other countries. The SCOPE project carried out between 1995 and 1996 at 49 US hospitals reported the VRE rate was between 9.5% to 20.6% in nosocomial enterococcal BSIs from different hospitals [30]. The 1997 SENTRY program reported that 14.1% of enterococcal BSIs were VRE in the United States [1]. A prospective nationwide surveillance study at Brazilian hospitals, which published in 2011, showed that the VRE rate was as high as 25% [9]. Therefore, the species distribution and antimicrobial resistance varies geographically.
Abdominal infections, central venous catheters, and unknown sources were the most common foci of enterococcal BSIs in the present study, is similar to the prior studies [28]. However, no infective endocarditis was observed in our study. There may be 2 reasons to explain this difference. First, the study population was different, as we were focused on nosocomial infection. Another reason was that echocardiography was not a routine examination for patients with BSIs in our hospital, so this could lead to missed diagnosis.
Our data showed that patients with nosocomial E. faecalis BSIs were more likely to get appropriate empirical treatment than those with E. faecium BSIs. This may be explained by the low resistance rate of E. faecalis to many antimicrobial agents, such as ampicillin (9%), gentamicin (33%), and ciprofloxacin (39%).
In the present study, the 30-day mortality of nosocomial enterococcal BSIs was 24%, and the early mortality (7-day) accounted for half. This is lower than many prior reports [5,8,9,12,31], but similar to a study focusing on vancomycin-susceptible enterococcal BSIs [22]. This may be due to selection bias of the study population, or may be a reflection of medical progress. Mortality was higher in patients with higher CCI and APACHE II scores, prior use of immunosuppressive agents, and complicated with impaired renal function. In other reports, the risk factors for mortality of patients with enterococcal BSIs may also include advanced age, pulmonary infection, malignancy, species, polymicrobial infection, ampicillin resistance, high-level gentamicin resistance, and vancomycin resistance [11,15,22]. Although controversial, we confirm that the severity of underlying diseases is the most important factor, especially for early morbidity.
The present study has several limitations that should be taken into consideration. First, our study was a retrospective study, the collection of clinical data depended on medical records rather than interviews and clinical examinations at the onset of infection by unified training doctors of our research group. Second, as a single-centre study, it could lead to an inevitable selection bias, and the multivariate logistic analysis might be affected by the small sample size. Third, not all isolates did all the same antimicrobial agent sensitivity test, so we could not incorporate all the antimicrobial agents resistance in predicting mortality risk factors in order to avoid selection bias.

Conclusions
Enterococci were major pathogens for nosocomial BSIs and were associated with high mortality,especially for