Antibiotic Stewardship Related to Delayed Diagnosis and Poor Prognosis of Critically Ill Patients with Vancomycin-Resistant Enterococcal Bacteremia: A Retrospective Cohort Study.

Purpose: Patients with septicemia caused by vancomycin-resistant Enterococcus (VRE) bacteremia have higher mortality rates than patients infected by VSE. Vancomycin or teicoplanin is selected as the antibiotic stewardship intervention to cover methicillin-resistant Staphylococcus aureus infections before blood culture reveals VRE bacteremia in critically ill patients with Gram-positive cocci (GPC) bacteremia; this may require linezolid or daptomycin treatment instead. We thus evaluated antibiotic stewardship practices, such as appropriate timing of antibiotic use in GPC bacteremia, and clinical outcomes of critically ill patients with VRE infection. Patients and
Methods: This retrospective study enrolled 191 critically ill patients with enterococcal bacteremia at the Taipei Tzu Chi Hospital during January 1, 2019-December 31, 2020. Demographic and clinical characteristics, as well as disease outcomes and appropriate antibiotic use after GPC bacteremia diagnosis, were compared between the VRE and VSE groups.
Results: Of 191 patients, 55 had VRE bacteremia (case group) and 136 had VSE bacteremia (control group). The rate of antibiotic change after initial antibiotic use for GPC bacteremia was higher in the VRE bacteremia group (100% vs 10.3%; p<0.001). The time to appropriate antibiotic administration after GPC bacteremia diagnosis was longer in the VRE bacteremia group (3.3±2.1 vs 1.5±1.8 days; p<0.001). Patients with VRE bacteremia had higher 28-day mortality rates (relative risk, 1.997; 95% confidence interval [CI], 1.041-3.83). Multivariate Cox regression analysis showed that delayed appropriate antibiotic administration of >3 days after GPC bacteremia diagnosis increased the risks of 28-day all-cause mortality (adjusted hazard ratio, 2.045; 95% CI, 1.089-3.84; p=0.026) in patients with VRE infection.
Conclusion: Patients with VRE bacteremia with delayed appropriate antibiotic administration of >3 days after GPC bacteremia diagnosis had increased 28-day mortality risks. New strategies for early VRE detection in GPC bacteremia may shorten the time to administer appropriate antibiotics and lower mortality rates.

Introduction

Enterococci, which are facultative anaerobic and Gram-positive catalase-negative bacteria, generally exhibit low levels of virulence owing to their presence as natural members of human intestinal flora. Although normally distributed in the human gut, they may cause various infections, including bacteremia or sepsis.1 Owing to their ability to survive in harsh environments (including high salt concentrations) and at wide temperature ranges (from 10 °C to >45 °C), they easily spread through Caregiver hand contamination and Medical catheter-associated transmission.2

In the United States, Enterococcus species accounted for 11–13% of healthcare-associated infections during 2015–2017.3 The emergence of vancomycin-resistant enterococci (VRE) has been alarming in the past two decades owing to high mortality rates.4 Based on the National Nosocomial Infections Surveillance System Report from Taiwan in 2004, among patients admitted to intensive care units (ICUs), the resistance rate was around 20–30% for nosocomial VRE.5 Moreover, longitudinal surveillance (1996–2009) of annual rates at National Taiwan University Hospital revealed a gradual rise in the prevalence of VRE (from 1.2% to 25.1%). These rates included all enterococcal isolates from patients with healthcare-associated infections.6 The causes of multidrug-resistant enterococci emergence may include intrinsic resistance to several antimicrobial agents. Another cause may be acquired resistance through mobile elements, such as transposons and plasmids against glycopeptides. Furthermore, VRE can transfer genetic material to other Gram-positive pathogens, thus further inducing antibiotic resistance.7–9

Patients with septicemia caused by VRE have a higher mortality rate than those with vancomycin-sensitive Enterococcus (VSE) bacteremia.10 This is the case even in the era of effective VRE therapy.4 Some studies indicate that VRE increases length of hospital stay and mortality.4,10–12 Therefore, it is important to identify the risk factors that could be applied in critical care to prevent disastrous outcomes. Additionally, the appropriateness of antimicrobial therapy has a prognostic impact in patients with VRE bacteremia.13–16 A vast majority of enterococcal bacteremia studies include different groups of patients such as patients who are immunosuppressed,17 with cancer,18 with VRE colonization,19,20 or patients admitted in the ICU.21,22 Linezolid and daptomycin are suitable treatments for confirmed VRE bacteremia. However, before blood culture reveals VRE bacteremia in critically ill patients with GPC bacteremia, vancomycin or teicoplanin is administered under antibiotic stewardship to cover methicillin-resistant Staphylococcus aureus (MRSA) infections. This delayed reporting of VRE leads to inappropriate antibiotic treatment, as linezolid or daptomycin treatment is required for VRE bacteremia instead. Therefore, this retrospective study aimed to determine clinical risk factors in patients with VRE bacteremia related to antibiotic stewardship application, such as appropriate timing of antibiotic administration, as well as determining clinical outcomes of patients with Enterococcus bacteremia.

Patients and Methods

This single-center, retrospective cohort study was approved by the Institutional Review Board (IRB) of Taipei Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation (Protocol No.: 10-XD-097) and was conducted in accordance with the guidelines of the amended Declaration of Helsinki. An informed consent waiver was received from the IRB, and patient privacy rights, including any individual person’s data in any form (such as individual details, images, or videos), have been observed.

Study Population and Design

This retrospective cohort study enrolled in-patients with Enterococcus bacteremia at the Taipei Tzu Chi Hospital from January 1, 2019 to December 31, 2020 (Figure 1). In these two years, 5144 positive blood cultures were screened. The study excluded patients with Gram-negative bacilli bacteremia, contamination, only having one set of blood cultures, and Gram-positive cocci (GPC) bacteremia excluding enterococcus species. All patients with Enterococcus bacteremia had at least two sets of blood cultures and underwent rectal swab examinations on admission. Records of patients with Enterococcus bacteremia were reviewed. Demographic and clinical characteristics were compared between the two groups (VRE and VSE bacteremia) according to variables defined as risk factors previously.15–26 These possible risk factors included antibiotic exposure, catheter use, or comorbidities. In addition, clinical outcomes were also compared between the two groups (VRE and VSE bacteremia). Four subgroups of antibiotics were categorized according to VRE or VSE infection. The timing related to appropriate antibiotic administration after GPC bacteremia diagnosis was also considered during subgrouping. If risk factors were present in critically ill patients, special consideration would be provided regarding antibiotic selection for MRSA or VRE for patients with GPC bacteremia. As part of our hospital antibiotic stewardship program, vancomycin and teicoplanin are the first treatment choices for MRSA bacteremia.27 However, linezolid or daptomycin is selected after confirmed VRE bacteremia.28 Furthermore, the maintenance vancomycin dose was adjusted according to the trough level (target 15 to 20 mcg/mL), which was measured before the fourth dose following the most recent dose adjustment, within 30 min before infusion, to ensure the effectiveness of the treatment.

Figure 1

Flow chart of patients enrolled in this study. We screened 5144 cases with blood culture data and excluded 3582 cases of GNB bacteremia. After further exclusion of 452 contaminated cultures, 233 single set blood cultures, and 2 cases of out-of-hospital cardiac arrest (OHCA) without return of spontaneous circulation (ROSC), 875 cases of GPC bacteremia remained. Furthermore, 191 Enterococcus bacteremia cases were analyzed after excluding 450 Staphylococcus and 231 Streptococcus cases. A total of 191 patients with enterococcal bacteremia (55 patients with VRE bacteremia [case group] and 136 patients with VSE bacteremia [control group]) were included in the analysis.

Study Setting

Clinical information of patients in both groups (VRE and VSE bacteremia) was recorded in Excel spreadsheets. Demographic and clinical data of each patient included age, sex, vital signs, catheter use, comorbid diseases, Charlson comorbidity index,29 laboratory data, antibiotic experiences, disease severity index, and infection events (other sites of Enterococcus infection and other bacteremia coinfections). Further data collected included initial antibiotic use with confirmed GPC bacteremia, antibiotic change rate after initial antibiotic use after GPC bacteremia diagnosis, VRE rectal swab, residence in a long-term care facility, antacid use, probiotic use, history of surgery within 6 months, and time to appropriate antibiotic administration after enterococcal bacteremia. The clinical outcome analysis after admission included in-hospital mortality, 28-day mortality, length of hospital stay, ICU admission rate, septic shock, use of mechanical ventilation, and ventilator days.

Disease severity indexes included sepsis-related organ failure assessment scores, Acute Physiology and Chronic Health Evaluation scores,30 and oxygenation use. Comorbidities included malignancy, diabetes mellitus, renal insufficiency, chronic obstructive pulmonary disease, liver disease, heart disease, and history of transplantation. Antibiotic use was defined as exposure to vancomycin, glycopeptide, carbapenem, cephalosporin, penicillin, colimycin, fluoroquinolones, Baktar (trimethoprim-sulfamethoxazole), metronidazole, clindamycin, macrolides (erythromycin, azithromycin), aminoglycosides, or tetracycline for >3 days within 3 months before the bacteremic episode. VRE bacteremia was defined as a blood culture growing Enterococcus species with a vancomycin minimum inhibitory concentration of ≥32 mcg/mL. The antibiotic susceptibility, catalase, and hemolysis tests were all performed at the clinical laboratory department. The duration of appropriate antibiotic treatment was recorded from the time to GPC bacteremia diagnosis to the time to appropriate antibiotic administration, in which final cultured microorganisms were susceptible to the antibiotics used. The 28-day mortality rate was determined as the number of patients who died within 28 days after enterococcal bacteremia divided by the total number of hospitalized patients.

Statistical Analyses

Data was analyzed using SPSS version 26.0 (IBM Corp, Armonk, NY, USA), and a P value of <0.05 indicated statistical significance. Categorical variables are expressed as frequencies and percentages and were compared using the chi-square test or Fisher’s exact test, if the expected values were below 5. Continuous variables are expressed as mean ± standard deviation and were analyzed using Student’s t-test. A multivariate analysis was performed using logistic regression to determine risk factors for VRE bacteremia. All risk factors with a significance level of <0.05 in the univariate analysis were included in the multiple logistic regression model. The Kaplan–Meier survival curve was used to analyze mortality in the VRE and VSE bacteremia subgroups. Multivariate analysis using Cox regression analysis was performed to determine the factors associated with all-cause 28-day mortality, after adjusting for other confounding factors.

Results

Characteristics and Laboratory Data of Participants

In our study, after the exclusion of patients with contaminated blood cultures, only one set of blood cultures, and OHCA without ROSC, 875 cases of GPC bacteremia were identified in the 2 study years (Figure 1). Staphylococcus bacteremia (positive catalase test) was the most common infection (51.4%), followed by Enterococcus (negative catalase test and γ-hemolysis) and Streptococcus (negative catalase test and α- or β-hemolysis) bacteremia (21.8%).

A total of 191 patients with Enterococcal bacteremia (55 patients with VRE bacteremia [case group] and 136 patients with VSE bacteremia [control group]) were included in the analysis. The VRE bacteremia case group included 15 women and 40 men, with a mean age of 72.1 years (±9.9 years), while the VSE bacteremia control group included 53 women and 83 men with a mean age of 72.4 years (±15.9 years). Demographic characteristics and clinical data of patients are presented in Table 1 and . Age, sex, and disease severity index were not significantly different between the case and control groups. Comorbidities were not significantly different between the two groups, except for hematological malignancy, gastrointestinal disease, and solid organ transplant. Laboratory data showed that hemoglobin, albumin, and lactate levels were lower while HCO3 and Na levels were higher in the VRE group than in the VSE group. Additionally, the VRE group had more patients with VRE rectal swabs, antacid use, history of surgery within 6 months, and mechanical ventilation use. Moreover, catheter use and antibiotic use were higher in the VRE group than in the VSE group. Prior use of glycopeptides, carbapenem, cephalosporin, penicillin, colimycin, fluoroquinolones, trimethoprim-sulfamethoxazole, metronidazole, and tetracycline was significantly higher in the VRE bacteremia group (p<0.05). The first line antibiotics to treat GPC bacteremia were selected for 7.3–10.3% of patients. Other initial antibiotic treatments included vancomycin and teicoplanin, where vancomycin was selected for 12.5–14.5% of patients and teicoplanin was selected for 77.2–78.2% of patients. Furthermore, the antibiotic change rate after initial antibiotic administration for GPC bacteremia was higher in the VRE bacteremia group (100% vs 10.3%; p<0.001; Table 1).

Table 1

Comparisons of Demographic Characteristics Between Critically Ill Patients with VRE Bacteremia and Those with VSE Bacteremia

Demographic Characteristics	VRE Bacteremia (N=55)	VSE Bacteremia (N=136)	p value
Sex			NS
Female	15 (27.3%)	53 (39%)
Male	40 (72.7%)	83 (61%)
Age	72.1±9.9	72.4±15.9	NS
Comorbidities
Charlson comorbidity index	6.9±2.5	6.3±3.2	NS
Any cancer	22 (40%)	37 (27.2%)	NS
Solid organ tumor	15 (27.3%)	37 (27.2%)	NS
Hematological malignancy	7 (12.7%)	0 (0%)	<0.001a
DM	17 (30.9%)	57 (41.9%)	NS
COPD	7 (12.7%)	16 (11.8%)	NS
Chronic renal failure	24 (43.6%)	54 (39.7%)	NS
Heart disease	30 (54.5%)	62 (45.6%)	NS
Liver cirrhosis	7 (12.7%)	7 (5.1%)	NS
Gastrointestinal disease	29 (52.7%)	40 (29.4%)	0.002a
Hepatobiliary disease	13 (23.6%)	29 (21.3%)	NS
Bone marrow transplant	0 (0%)	0 (0%)	NS
Solid organ transplant	5 (9.1%)	1 (0.7%)	0.003a
Catheter
Catheter total number	2.4±1.8	1.0±1.3	<0.001a
Central venous catheter	29 (52.7%)	22 (16.2%)	<0.001a
Urinary catheter	30 (54.5%)	38 (27.9%)	0.001a
NG tube	31 (56.4%)	36 (26.5%)	<0.001a
Drainage tube	11 (20%)	6 (4.4%)	0.001a
Endo tube	16 (29.1%)	9 (6.6%)	<0.001a
Tracheostomy	2 (3.6%)	6 (4.4%)	NS
Double lumen (or perm)	15 (27.3%)	19 (14%)	0.03a
Antibiotic experiences
Antibiotic exposure within 3 months	46 (83.6%)	64 (47.1%)	<0.001a
Vancomycin	5 (9.1%)	5 (3.7%)	NS
Glycopeptide	27 (49.1%)	10 (7.4%)	<0.001a
Carbapenem	26 (47.3%)	17 (12.5%)	<0.001a
Cephalosporin	40 (72.7%)	49 (36%)	<0.001a
Penicillin	30 (54.5%)	27 (19.9%)	<0.001a
Colimycin	9 (16.4%)	5 (3.7%)	0.002a
Fluoroquinolones	23 (41.8%)	28 (20.6%)	0.003a
Baktar	6 (10.9%)	2 (1.5%)	0.003a
Metronidazole	4 (7.3%)	2 (1.5%)	0.037a
Clindamycin	0 (0%)	2 (1.5%)	NS
Macrolides	0 (0%)	0 (0%)	NS
Aminoglycosides	2 (3.6%)	4 (2.9%)	NS
Tetracycline	4 (7.3%)	2 (1.5%)	0.037a
Infection events
Other sites of Enterococcus infection	17 (30.9%)	43 (31.6%)	NS
Other bacteremia coinfections	35 (63.6%)	74 (54.4%)	NS
Others
Rectal swab test result positive for VRE	14 (25.5%)	6 (4.4%)	<0.001a
Antacid use	34 (61.8%)	52 (38.2%)	0.003a
Probiotic use	1 (1.8%)	7 (5.1%)	NS
History of surgery within 6 months	24 (43.6%)	30 (22.1%)	0.003a
From long-term care facility	5 (9.1%)	19 (14%)	NS
Septic shock	8 (14.5%)	13 (9.6%)	NS
Mechanical ventilation use	27 (49.1%)	37 (27.2%)	0.004a
Initial antibiotic during confirmed GPC bacteremia
First line antibiotics	4 (7.3%)	14 (10.3%)	NS
Vancomycin	8 (14.5%)	17 (12.5%)	NS
Teicoplanin	43 (78.2%)	105 (77.2%)	NS
Linezolid	0	0	NS
Daptomycin	0	0	NS
Tigecycline	0	0	NS
Antibiotic change rate	55 (100%)	14 (10.3%)	<0.001a
Time to appropriate antibiotic administration after GPC bacteremia diagnosis	3.3±2.1	1.5±1.8	<0.001a

Notes: aStatistically significant, p<0.05.

Abbreviations: VRE, vancomycin-resistant Enterococcus; VSE, vancomycin-sensitive Enterococcus; NG, nasogastric; DM, diabetes mellitus; COPD, chronic obstructive pulmonary disease; GPC, Gram-positive cocci; NS, non-significant.

All significant univariate variables were included and adjusted for in the final multivariate model (Table 2; other insignificant variables are present in ). However, multivariate logistic regression showed that central venous catheter (CVC) use (odds ratio [OR], 3.116; 95% confidence interval [CI], 1.386–7.008), glycopeptide exposure (OR, 5.734; 95% CI, 2.297–14.312), and cephalosporin exposure (OR, 2.923; 95% CI, 1.346–6.345) were associated with VRE bacteremia after adjusting for other confounding factors. These findings suggest that CVC use, glycopeptide exposure, and cephalosporin exposure for >3 days within 3 months are significant risk factors for VRE bacteremia in patients with Enterococcus bacteremia.

Table 2

Univariate and Multivariate Logistic Regression Analyses to Identify Variables Associated with Vancomycin-Resistant Enterococcus (VRE) Bacteremia in Critically Ill Patients with Enterococcus Bacteremia

	Univariate		Multivariate
	OR (95% CI)	p	Adjusted OR (95% CI)	p
Gastrointestinal disease	2.677 (1.404–5.103)	0.003a
Solid organ transplant	13.5 (1.539–118.404)	0.019a
Total bilirubin	1.085 (1.004–1.171)	0.038a
Albumin	0.496 (0.306–0.802)	0.004a
Lactate (mmol/L)	0.784 (0.632–0.974)	0.028a
HCO3	1.088 (1.006–1.177)	0.036a
A-a DO2	1.002 (1–1.004)	0.03a
Overall catheter use	1.801 (1.437–2.256)	<0.001a
Central venous catheter	5.78 (2.873–11.627)	<0.001a	3.116 (1.386–7.008)	0.006a
Urinary catheter	3.095 (1.616–5.926)	0.001a
NG tube	3.588 (1.864–6.908)	<0.001a
Drainage tube	5.417 (1.892–15.507)	0.002a
Endo tube	5.789 (2.373–14.126)	<0.001a
Double lumen (or perm)	2.309 (1.073–4.969)	0.032a
Antibiotic exposure within 3 months	5.575 (2.61–12.666)	<0.001a
Glycopeptide	12.150 (5.281–27.951)	<0.001a	5.734 (2.297–14.312)	<0.001a
Carbapenem	6.276 (3.013–13.072)	<0.001a
Cephalosporin	4.735 (2.377–9.43)	<0.001a	2.923 (1.346–6.345)	0.007a
Penicillin	4.844 (2.46–9.54)	<0.001a
Colimycin	5.126 (1.633–16.087)	0.005a
Fluoroquinolones	2.772 (1.407–5.462)	0.003a
Baktar	8.204 (1.602–42.016)	0.012a
Antacid use	2.615 (1.373–4.983)	0.003a
History of surgery within 6 months	2.735 (1.400–5.343)	0.003a
Rectal swab test result positive for VRE	7.398 (2.671–20.492)	<0.001a
Mechanical ventilation use	3.314 (1.604–6.850)	0.001a

Notes: Dependent variable: VRE bacteremia. aStatistically significant, p<0.05. All significant univariate variables were included and adjusted for in the final multivariate model.

Abbreviations: OR, odds ratio; CI, confidence interval; NS, non-significant.

Clinical Outcomes

Patients with VRE bacteremia had higher in-hospital mortality (relative risk [RR], 2.595; 95% CI, 1.366–4.933) and 28-day mortality (RR, 1.997; 95% CI, 1.041–3.83) rates than patients with VSE bacteremia (Table 3). The mean length of hospital stay after bacteremia was longer for patients with VRE bacteremia than for those with VSE bacteremia (25.1±26.8 days vs 15.0±14.5 days). Moreover, the days of ventilation usage were longer in the VRE group (p<0.05).

Table 3

Comparisons of Clinical Outcomes Between Patients with VRE Bacteremia and Those with VSE Bacteremia

	VRE Bacteremia (N=55)	VSE Bacteremia (N=136)	RR (95% CI)	p value
All-cause hospital mortality	30 (54.5%)	43 (31.6%)	2.595 (1.366–4.933)	0.004a
All-cause 28-day mortality	24 (43.6%)	38 (27.9%)	1.997 (1.041–3.830)	0.037a
AAU ≤3 D	9/33 (27.3%)	29/107 (27.1%)	0.991(0.413–2.382)	0.985
AAU >3 D	15/22 (68.2%)	9/29 (31%)	4.762 (1.444–15.703)	0.01a
Length of stay in hospital	25.1±26.8	15.0±14.5	1.025 (1.009–1.041)	0.002a
Ventilator use days	29.8±29.1	15.9±20.2	1.024 (1.001–1.047)	0.04a

Notes: aStatistically significant, p<0.05.

Abbreviations: VRE, vancomycin-resistant Enterococcus; VSE, vancomycin-sensitive Enterococcus; RR, relative risk; ICU, intensive care unit; NS, non-significant; AAU, appropriate antibiotic use.

Time to Appropriate Antibiotic Use

The time to appropriate antibiotic use after confirmed GPC bacteremia was longer in the VRE bacteremia group (3.3±2.1) than in the VSE bacteremia group (1.5±1.8) (p<0.001; Table 1). The median time to appropriate antibiotic administration after confirmed GPC bacteremia in patients with VRE bacteremia was 3 days. The four subgroups were categorized according to VRE or VSE infection and appropriate antibiotic use of ≤3 days or >3 days after GPC bacteremia diagnosis and were analyzed for all-cause 28-day mortality (Table 3). The VRE group with appropriate antibiotic administration of >3 days after GPC bacteremia diagnosis had the highest mortality rate (68.2%) (p<0.05). The survival analysis was performed using the Kaplan–Meier method (Figure 2) (Log rank test, p=0.033). VRE with delayed appropriate antibiotic administration (>3 days) was a significant risk factor for all-cause 28-day mortality. Moreover, because CVC use, glycopeptide, and cephalosporin were not associated with all-cause 28-day mortality, they were excluded from the final multivariate Cox regression analysis (Table 4). The significant univariate variables, such as Acute Physiologic Assessment and Chronic Health Evaluation II score, Sequential Organ Failure Assessment score, and enterococcal bacteremia with appropriate antibiotic use (AAU), were included in the final multivariate model analysis. The multivariate Cox regression analysis (Table 4) showed that patients with VRE and delayed AAU of >3 days (adjusted hazard ratio, 2.045; 95% CI: 1.089–3.84, p=0.026) had increased risks of all-cause 28-day mortality.

Figure 2

Survival curves according to VRE or VSE bacteremia and the time to appropriate antibiotic administration after GPC bacteremia diagnosis (four groups). The Kaplan–Meier survival analysis was adopted for patients with (1) VSE bacteremia with time to appropriate antibiotic administration ≤3 days, (2) VSE bacteremia with time to appropriate antibiotic administration >3 days, (3) VRE bacteremia with time to appropriate antibiotic administration ≤3 days, and (4) VRE bacteremia with time to appropriate antibiotic administration >3 days during their hospital stay (cut-off point: 28 days). Log rank test: p<0.05.

Table 4

Cox Regression Model to Determine Factors Associated with All-Cause 28-Day Mortality

	Univariate		Multivariate
	HR (95% CI)	p value	Adjusted HR (95% CI)	p value
VSE with AAU ≤3 D	Reference	0.047	Reference	0.048
VSE with AAU >3 D	0.948 (0.449–2.005)	NS	1.053 (0.493–2.249)	NS
VRE with AAU ≤3 D	0.744 (0.35–1.58)	NS	0.679 (0.308–1.498)	NS
VRE with AAU >3 D	2.096 (1.123–3.913)	0.02a	2.045 (1.089–3.84)	0.026a
APACHE II	1.04 (1.009–1.072)	0.011a	1.040 (1.009–1.072)	0.011a
SOFA score	1.083 (1.014–1.158)	0.018a	1.057 (0.962–1.16)	NS
Central venous catheter use	1.452 (0.872–2.418)	NS
Antibiotic exposure for 3 days within 3 months	1.474 (0.839–2.589)	NS
Glycopeptide	0.89 (0.496–1.595)	NS
Cephalosporin	1.299 (0.78–2.164)	NS
Septic shock	1.995 (1.062–3.748)	0.032	1.249 (0.604–2.585)	NS
Mechanical ventilation use	1.207 (0.700–2.083)	NS	0.796 (0.400–1.585)	NS

Notes: Events: all-cause 28-day mortality. Time: days since enterococcal bacteremia diagnosis to 28 days. aStatistically significant, p<0.05.

Abbreviations: AAU, appropriate antibiotic use; HR, hazard ratio; CI, confidence interval; NS, non-significant.

Discussion

In this single-center, retrospective study, we found that CVC use, glycopeptide exposure, and cephalosporin exposure were significant risk factors for VRE bacteremia in patients with Enterococcus bacteremia after adjusting for other confounding factors. CVC use is an important risk factor for VRE bacteremia, and the association is well established.23,31 Our findings further suggest that VRE is spread from environmental surfaces to the CVC and then into the bloodstream, therefore emphasizing the importance of infection control practices. Glycopeptide exposure for >3 days within 3 months was a significant risk factor for VRE bacteremia (p<0.001) in our study. However, prior vancomycin exposure was not a significant risk factor for VRE bacteremia, which differs from the study findings of Furtado et al.16 For the prevention of adverse events (skin rash and red man syndrome), teicoplanin was used more frequently than vancomycin in our study, and for that reason, only 10 patients had prior exposure to vancomycin (Table 1). Regarding cephalosporin, as demonstrated in the current study, several previous studies also showed that prior use of cephalosporins for >3 days within 3 months was a significant risk factor for VRE bacteremia.19,32 In our study, CVC use, previous glycopeptide treatment, and previous cephalosporin treatment were associated with VRE bacteremia (Table 2). However, these risk factors for VRE bacteremia were the same as those for MRSA bacteremia.33,34 This result, however, does not offer any benefit for antibiotic stewardship.

Comorbidities and laboratory data were not significant risk factors for VRE bacteremia after adjusting for other confounding factors in our study. However, Lautenbach et al15 found that renal insufficiency and neutropenia were independent risk factors for VRE bacteremia in 72 patients with VRE bacteremia and 188 patients with VSE bacteremia. Moreover, Peel et al23 observed that neutropenia, allogeneic bone marrow transplantation, and hypoalbuminemia were independent risk factors for VRE bacteremia after adjusting for other confounding factors in 80 patients with VRE bacteremia and 80 matched control patients. Furthermore, Johnstone et al25 demonstrated that bone marrow transplantation, solid organ transplantation, cancer, and heart disease were independent risk factors for VRE bacteremia after adjusting for other confounding factors in 217 patients with VRE bacteremia and 651 matched control patients. As our study population was relatively small, further multicenter studies with longer follow-up periods are needed to confirm the present findings.

Despite the availability of effective VRE therapy, VRE bacteremia remains an important risk factor for in-hospital mortality and length of hospital stay when compared with VSE bacteremia.4 In our cohort, 30 (54.5%) patients with VRE bacteremia died in the hospital. The in-hospital mortality rate (RR, 2.595; CI 95%, 1.366–4.933) and the 28-day mortality rate (RR, 1.997; CI 95%, 1.041–3.83) were evaluated for VRE- and VSE-infected patients. The results show that the rates were higher for patients with VRE bacteremia than for patients with VSE bacteremia. This might be owing to inappropriate initial antimicrobial coverage.13,14 Moreover, in critical care, when GPC bacteremia is combined with shock or critical illness, MRSA may be considered, and vancomycin or teicoplanin is administered. On the other hand, not only pathogen and antibiotic susceptibility but also antibiotic dosing and appropriate time of administration need to be considered in critically ill patients. For instance, for the effectiveness of the treatment, appropriate vancomycin dosing requires consideration of the type and severity of infection, patient weight, kidney function, and even pharmacokinetic/pharmacodynamic efficacy.35–37 However, VRE is often ignored in these instances. In reality, appropriate antibiotic treatment of VRE is always delayed until VRE confirmation. Notably, a large dataset cohort study showed that delay in initial antibiotic administration was associated with increased in-hospital mortality.38 This study also showed a linear increase in the risk of mortality for each hour of delay in antibiotic administration. Another retrospective study conducted in Italy39 found that time (since blood culture collection) to appropriate antibiotic therapy was an independent predictor of 30-day mortality in patients with carbapenemase-producing Klebsiella pneumoniae bacteremia. Another single-center retrospective study14 revealed that receiving appropriate therapy within the first 48 hours (from blood culture to receiving the first dose of appropriate therapy) was associated with reduced mortality in patients with hospital-onset Enterococcus bacteremia. In our study, patients with VRE bacteremia with delayed appropriate antibiotic administration (>3 days) after GPC bacteremia diagnosis had an increased risk of 28-day mortality. This is the first study to investigate antibiotic stewardship application in patients with VRE bacteremia in Taiwan. CVC use, previous glycopeptide treatment, and previous cephalosporin treatment were associated with VRE bacteremia but not with all-cause 28-day mortality (Tables 2 and 4). Because risk factors for VRE bacteremia were the same as for MRSA bacteremia,33,34 these results do not offer any benefit for antibiotic stewardship. Therefore, early detection of VRE bacteremia and timely use of appropriate antibiotics may be more important in antibiotic stewardship practices.

Recently, film array and matrix-assisted laser desorption/ionization-time of flight mass spectrometry may help with the rapid detection of VRE bacteremia within hours.40–43 Hence, our findings demonstrate that the only effective factor in improving appropriate antibiotic administration in VRE bacteremia and improving clinical outcomes is the rapid VRE detection in GPC bacteremia.

Our study had some limitations. Because this is a single-center retrospective study, it is important to evaluate the study validity carefully before making any decision or changing clinical practice. Moreover, selection bias may be a concern in the subgroup analyses (VRE or VSE bacteremia and the time to appropriate antibiotic administration after GPC bacteremia diagnosis) and the sample size may be insufficient. However, we are still glad to share our study result and to offer the issue of delayed appropriate antibiotic treatment for patients with VRE bacteremia. Further prospective studies are needed to evaluate early detection of VRE and clinical outcomes.

Conclusion

In this retrospective cohort study, patients with VRE bacteremia with delayed treatment (antibiotic administration >3 days after GPC bacteremia diagnosis) had higher risks of all-cause 28-day mortality. Our findings suggest that early detection of VRE bacteremia in GPC bacteremia may shorten the duration to appropriately administer antibiotics and therefore reduce mortality rates. Further prospective studies are needed to validate these present findings.