Author + information
- Received February 27, 2019
- Revision received April 8, 2019
- Accepted April 22, 2019
- Published online July 8, 2019.
- Anders Dahl, MD, PhDa,b,∗ (, )@HerlevGentofte,
- Kasper Iversen, MD, DMSca,
- Niels Tonder, MDc,
- Nis Hoest, MD, PhDb,
- Magnus Arpi, MDd,
- Morten Dalsgaard, MD, PhDa,
- Mahtab Chehri, MDe,
- Lars L. Soerensen, MD, PhDa,
- Soren Fanoe, MD, PhDf,
- Soeren Junge, MDg,
- Ulla Hoest, MD, PhDg,
- Nana Valeur, MD, PhDb,
- Trine K. Lauridsen, MD, PhDa,
- Emil Fosbol, MD, PhDf,h,
- Thomas Hoi-Hansen, MD, PhDa and
- Niels E. Bruun, MD, DMSca,i,j,k
- aDepartment of Cardiology, Herlev Gentofte University Hospital, Copenhagen, Denmark
- bDepartment of Cardiology, Bispebjerg Hospital, Copenhagen, Denmark
- cDepartment of Cardiology, Nordsjællands Hospital, Hillerød, Denmark
- dDepartment of Clinical Microbiology, Herlev Gentofte University Hospital, Copenhagen, Denmark
- eDepartment of Clinical Microbiology, Hvidovre University Hospital, Hvidovre, Denmark
- fDepartment of Cardiology, Hvidovre University Hospital, Hvidovre, Denmark
- gDepartment of Cardiology, University Hospital Glostrup, Glostrup, Denmark
- hDepartment of Cardiology, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
- iDepartment of Cardiology, Zealand University Hospital, Roskilde, Denmark
- jClinical Institute, Copenhagen University, Copenhagen, Denmark
- kClinical Institute, Aalborg University Hospital, Aalborg, Denmark
- ↵∗Address for correspondence:
Dr. Anders Dahl, Department of Cardiology, Herlev Gentofte University Hospital, Kildegårdsvej 28, Building 8, Third Floor, 2900 Copenhagen, Denmark.
Background Enterococcus faecalis is the third most frequent cause of infective endocarditis (IE). Despite this, no systematic prospective echocardiography studies have examined the prevalence of IE in patients with E. faecalis bacteremia.
Objectives This study sought to determine the prevalence of IE in patients with E. faecalis bacteremia. The secondary objective was to identify predictors of IE.
Methods From January 1, 2014, to December 31, 2016, a prospective multicenter study was conducted with echocardiography in consecutive patients with E. faecalis bacteremia. Predictors of IE were assessed using multivariate logistic regression with backward elimination.
Results A total of 344 patients with E. faecalis bacteremia were included, all examined using echocardiography, including transesophageal echocardiography in 74% of the cases. The patients had a mean age of 74.2 years, and 73.5% were men. Definite endocarditis was diagnosed in 90 patients, resulting in a prevalence of 26.1 ± 4.6% (95% confidence interval [CI]). Risk factors for IE were prosthetic heart valve (odds ratio [OR]: 3.93; 95% CI: 1.76 to 8.77; p = 0.001), community acquisition (OR: 3.35; 95% CI: 1.74 to 6.46; p < 0.001), ≥3 positive blood culture bottles (OR: 3.69; 95% CI: 1.88 to 7.23; p < 0.001), unknown portal of entry (OR: 2.36; 95% CI: 1.26 to 4.40; p = 0.007), monomicrobial bacteremia (OR: 2.73; 95% CI: 1.23 to 6.05; p = 0.013), and immunosuppression (OR: 2.82; 95% CI: 1.20 to 6.58; p = 0.017).
Conclusions This study revealed a high prevalence of 26% definite IE in patients with E. faecalis bacteremia, suggesting that echocardiography should be considered in all patients with E. faecalis bacteremia.
Enterococcus faecalis is the third most frequent cause of infective endocarditis (IE), a disease with high morbidity and mortality (1,2). A recent review found enterococci to be the most frequent cause of endocarditis in patients undergoing transcatheter aortic valve replacement (3). Despite this, no systematic prospective echocardiographic studies have examined patients with E. faecalis bacteremia to determine the prevalence of IE. There are only a few retrospective and case-control studies showing an estimated IE prevalence ranging from 5% for mixed enterococcal bacteremia to 13% for monomicrobial E. faecalis bacteremia (4–7). All studies were limited by low rates of echocardiography, with transesophageal echocardiography (TEE) rates as low as 12% (6). The prevalence of IE was lowest in the studies with the lowest echocardiographic examination rates (4–7). Several studies have shown that E. faecalis IE is increasing (8–10), and the low prevalence of IE in some studies of E. faecalis bacteremia may be biased by low rates of echocardiography (6). The increase in E. faecalis IE is likely related to an aging population with more comorbidities, more implanted prosthetic material in the heart, and an increasing number of surgical procedures in the urinary and gastrointestinal tract (11–13). Echocardiography is the cornerstone imaging technique in diagnosing IE (14). The important question is whether current medical practice with rather low echocardiography rates results in missed cases of endocarditis. Given the limited evidence of the actual prevalence of IE in patients with E. faecalis bacteremia, we conducted the present study with systematic echocardiographic screening in all patients with E. faecalis bacteremia. The objectives of the study were to prospectively estimate the prevalence of IE in patients with E. faecalis bacteremia and to identify predictors of IE.
In a multicenter setup with 10 hospitals in the Capital Region of Denmark covering a catchment area of approximately 1.5 million people, all adult patients (>18 years of age) with E. faecalis bacteremia were considered for inclusion during the 3-year study period (January 1, 2014, to December 31, 2016). Consecutive inclusion of all patients with E. faecalis bacteremia was ensured by daily contact between the principal investigator and the different departments of clinical microbiology. Completeness of data was further validated by continuously drawing lists of patients with E. faecalis bacteremia from the microbiological database twice a week. In this way all patients with E. faecalis bacteremia were considered for inclusion. Patients dying before the results of the blood cultures were available and patients considered so terminally ill that clinicians terminated all treatment were excluded. Follow-up control blood cultures were performed only if clinicians suspected continued bacteremia and thus not performed routinely in all patients.
Two of the hospitals were referral centers for endocarditis, and all hospitals have videoconferences with a tertiary center. Patients with complicated IE were transferred to an IE center and evaluated for cardiothoracic surgery.
The diagnosis of IE versus non-IE was made by the multidisciplinary IE expert team, assessing all available information including clinical data, laboratory data, cardiac imaging, and data to evaluate the modified Duke criteria. The assessment of cardiac imaging was not done blinded to clinical data. The decision was made during admission when the necessary information was available to confirm or reject the diagnosis of IE. The expert team used the modified Duke criteria (15) and the current version of the guidelines of the European Society of Cardiology to evaluate the diagnosis (16,17). According to this, positron emission tomography/computed tomography was included in the expert team evaluation on the basis of the findings of Saby et al. (18). Cases with Duke-definite IE were all diagnosed with definite IE by the expert team. Cases with Duke-possible IE were included in the non-IE group except for 3 cases with obvious clinically confirmed IE (large vegetations but only 1 positive blood culture) resulting in the IE expert team’s diagnosing them with definite IE and including them in the IE group.
The mode of acquiring E. faecalis bacteremia was defined as previously described (19) on the basis of the most used international definitions in IE studies distinguishing between community acquired, health care–associated, and in-hospital acquired (1,20,21). The presumed portal of entry was determined as follows: 1) microbiologically established if E. faecalis was cultured from a clinically plausible entry site; 2) clinically established if there were clear signs and symptoms of localized infection entry and/or a clear temporal link to a proceeding surgical procedure; or 3) when none of these categories were fulfilled, the presumed portal of entry was registered as unknown. To assess comorbidity we calculated the Charlson comorbidity index on the basis of the algorithm proposed by Quan et al. (22). Immunosuppression was defined as active treatment with immunosuppressant medication such as corticosteroids (any dose), chemotherapy, or antirheumatic drugs affecting the immune system.
Relapse E. faecalis infection was defined as within 6 months, and reinfection with E. faecalis was defined as occurring between 6 and 12 months after primary infection.
In Denmark, all patients have a unique social security number, making 100% follow-up possible through electronic patient charts, microbiological databases, and the Danish registry of death. Relapse and reinfection with E. faecalis as well as death of any cause were registered.
Categorical data are expressed as frequencies and percentages and were compared between groups using chi-square and Fisher exact tests as appropriate. Continuous data are expressed as mean ± SD or median and interquartile range, and groups were compared using 2-sample Student’s t-tests or Mann-Whitney U tests, respectively. Clinical variables expected to be potentially important predictive variables of the outcome (IE) on the basis of published research were included in the univariate logistic regression. All variables were allowed entry in the multivariate logistic regression. Independent predictors were determined using backward elimination with a significance level of p < 0.05 for a variable to stay in the model. A 2-sided p value <0.05 was considered to indicate statistical significance in all analyses. All statistical analyses were completed using IBM SPSS version 24.0 (SPSS, Chicago, Illinois) and SAS version 9.4 (SAS Institute, Cary, North Carolina).
To limit the risk for tautology from allowing variables in the multivariate logistic regression that are also part of the modified Duke criteria, we conducted a 2-step sensitivity analysis. In the first part, we used a method from Tubiana et al. (23), excluding all patients with definite IE in whom the definite classification was dependent on the presence of the modified Duke criteria prosthetic valve or continuous E. faecalis bacteremia (included in the statistical model as potential predictors of IE). In the second part, we excluded the variables prosthetic valve and/or number of positive blood cultures to evaluate the other predictors without including the variables possibly influenced by tautology.
Sample size calculation
On the basis of earlier retrospective studies (5,7) finding an IE prevalence of 10% to 15% in E. faecalis bacteremia, we used an estimated IE prevalence of 15% to calculate the sample size. To obtain a 95% confidence interval around an IE estimate of 15 ± 5%, a minimum sample size of 196 patients with E. faecalis bacteremia was required.
Ethics and permissions
Treatment followed normal clinical routines referring to European Society of Cardiology and national guidelines, and echocardiography was offered and recommended as part of the examination program for E. faecalis bacteremia as advised by the Danish Society of Cardiology (24). Data were collected with the approval of the Danish Health and Medicines Authorities with journal numbers 3-3013-764/1, 3-3013-764/2, and 3-3013-764/3 and handled according to approval by the Danish Data Protection Agency with journal number GEH-2014-036.
Patient inclusion is shown in Figure 1. During a 3-year period (January 1, 2014, to December 31, 2016), a total of 421 adult patients were diagnosed with E. faecalis bacteremia. After exclusion, all remaining 344 patients were examined with echocardiography, including TEE in 253 patients (74%). Ninety patients were diagnosed with definite endocarditis by the IE team, resulting in an IE prevalence of 26.1 ± 4.6% (95% confidence interval). The diagnosis of IE versus non-IE was made on median day 3 after positive blood cultures (interquartile range: 2 to 6 days). Sixty-seven patients (19%) were classified as having possible IE according to the modified Duke criteria. Sixty-four of these Duke-possible IE cases (96%) were diagnosed as bacteremia only by the IE expert team and included in the non-IE group. The remaining 3 patients (4%) with Duke-possible IE were diagnosed as having certain IE by the IE expert team and included in the IE group. These 3 patients with IE were diagnosed on the basis of large vegetations (>10 mm) and 1 of 2 blood cultures with E. faecalis (Duke minor).
The mean age of the patients was 74 years, and 26% of patients were women, without any significant difference between patients with and those without IE (Table 1). Both groups carried a moderate to high degree of comorbidity, with a nonsignificantly higher Charlson comorbidity index in patients without IE. Patients diagnosed with IE had a significantly higher rate of prosthetic heart valves and cardiac devices (Table 1).
Characteristic of infection
The mode of acquiring the infection was dominated by health care–associated infections in both groups. Community acquired infection was more common in patients with IE, whereas in-hospital acquisition was more common in those with bacteremia without IE. The urinary tract was the most frequent presumed portal of entry, followed by the gastrointestinal tract. Unknown portal of entry was frequent and significantly more common in patients with IE (Table 1).
Symptoms and clinical and laboratory findings
Upon admission, the most frequently reported symptom was fever. Patients with endocarditis presented with more unspecific symptoms and a clinical picture with only moderately elevated white blood cell count and C-reactive protein and less often with systemic inflammatory response syndrome (Online Table 1). Microbiological findings showed that patients with IE more often had monomicrobial E. faecalis bacteremia (90%) and ≥3 positive blood culture bottles (84%) (Online Table 2). High-level gentamicin resistance (minimal inhibitory concentration ≥256 mg/l) was significantly more common in E. faecalis strains causing IE (17%) compared with strains causing bacteremia only (6%) (p = 0.009). Few E. faecalis strains (3%) were resistant to penicillin (minimal inhibitory concentration >8 mg/l), and none of the strains were resistant to vancomycin or ampicillin (Online Table 2).
Echocardiographic findings in patients with endocarditis
The 90 patients with definite IE were diagnosed primarily on the basis of vegetations (83%) (Table 2). Twenty-three patients had small vegetations (≤5 mm), 25 patients had medium-size vegetations (5 to 10 mm), and 27 patients had large vegetations (≥10 mm). The remaining patients were diagnosed on the basis of intracardiac abscess (n = 6), significant new paravalvular leak or dehiscent prosthetic valve (n = 4), pseudoaneurysm or perforation (n = 3), and findings on positron emission tomography/computed tomography (n = 2). Most frequent was aortic valve endocarditis, followed by mitral endocarditis and combined aortic and mitral valve endocarditis. Less frequent were isolated pacemaker lead endocarditis, tricuspid valve endocarditis, and pulmonary valve endocarditis. Of the 75 patients with vegetations, transthoracic echocardiography missed vegetations in 35 cases (47%) in which TEE found vegetations.
Predictors of endocarditis
Predictive factors for endocarditis were prosthetic heart valve, community acquisition, ≥3 positive blood culture bottles, unknown portal of entry, monomicrobial infection, and immunosuppression (Figure 2).
The first part of the sensitivity analysis performed in a subset of 318 patients resulted in a model with no differences regarding 5 of the independent variables: community acquisition, unknown origin, prosthetic heart valve, immunosuppression, and ≥3 positive blood culture bottles. Only monomicrobial infection (not part of the modified Duke criteria) was insignificant and therefore removed from the model (Online Table 3). The second part of the sensitivity analysis demonstrated that the predictive factors in the statistical model remained significant after excluding prosthetic valve and/or ≥3 positive blood culture bottles (Online Table 4).
Dividing the patients into a low-risk group (0 risk factors), an intermediate-risk group (1 or 2 risk factors), and a high-risk group (3 to 6 risk factors), the IE prevalence was 3.4%, 13.9%, and 56.1%, respectively (Central Illustration).
Treatment, complications, and outcomes
Patients with endocarditis were treated significantly longer with intravenous antibiotics (Table 2). Heart valve surgery because of endocarditis was completed in almost 20% of the patients, and approximately 10% of the patients with IE had their cardiac devices removed (Table 2). Relapse E. faecalis infection within 6 months and reinfection with E. faecalis between 6 and 12 months were comparable in the 2 groups. Patients with endocarditis had higher in-hospital mortality, whereas if discharged alive, the 1-year mortality was significantly lower in patients surviving endocarditis (Table 2). When combining mortality in-hospital and during the first year after discharge, there was no significant difference between the groups.
Intention to screen
None of the excluded patients underwent echocardiography, and therefore the echocardiography rate in the total cohort was 82% (344 of 421 patients). When including the patients declining echocardiography and those discharged before echocardiography could be performed, the total number of patients was 374. Two patients died in the hospital without undergoing echocardiography. None of the patients who missed screening died within 30 days of discharge or had a relapse of bacteremia within 6 months after discharge, and hence it could be assumed that they did not have IE. When including them in the non-IE group, the prevalence of IE would be 24.1 ± 4.3%. The remaining key results were not changed significantly by adding these patients to the non-IE group (data not shown).
The main finding of this study is a high IE prevalence of 26% in patients with E. faecalis bacteremia. This is the first study to systematically examine patients with E. faecalis bacteremia using echocardiography. It seems that the more extensively we examine using echocardiography, the higher the detection rate of IE. This is not surprising, but the question is, do we overlook cases of IE in the normal clinical routine? The discrepancy in IE prevalence between our prospective study and the previous studies (4–7) suggests a substantial underdiagnosis of E. faecalis endocarditis in the general everyday clinic represented by the earlier studies. This theory is supported by the fact that 2 of the earlier retrospective studies were performed in the same catchment area as the present study (5,7). In our study, the majority of the patients with endocarditis had extensive intracardiac infection with vegetations, abscesses, pseudoaneurysms, or dehiscent prosthetic valves. However, despite the severe infection, many patients with E. faecalis IE have unspecific symptoms and present with a complex clinical picture, with only moderately elevated white blood cell count and C-reactive protein and often without systemic inflammatory response syndrome. Therefore, these patients are challenging to diagnose, and there is risk for overlooking serious infection if they are not examined using echocardiography. If echocardiography is used more extensively, some cases of less severe cardiac infections will be diagnosed, and it can be difficult to distinguish between a small vegetation and a thickened degenerated valve. However, E. faecalis IE is a very serious disease, and diagnosing it in a milder and less invasive state and initiating appropriate treatment as early as possible may prevent serious complications such as valve destruction and abscess formation. The rate of complications in our study was relatively low compared with other studies (25–27). The relapse rate in the present study is comparable with a weighted average in a review based on 17 international E. faecalis IE studies (19). Earlier attempts have been made to create a risk score to reduce the need for TEE in E. faecalis bacteremia (6). The outcome of the NOVA score (incorporating the number of positive blood cultures, origin of the bacteremia, previous valve disease, and auscultation of heart murmur) study was that the only 2 determinant factors determining whether TEE should be performed were ≥3 positive blood cultures with E. faecalis and unknown origin of infection (6). In contrast to the actual TEE rate of 12% in the study by Bouza et al. (6), their NOVA score estimated that a TEE rate of >70% would be needed to find all cases of IE. In our study, we confirm some of the known risk factors for E. faecalis IE: prosthetic heart valve, unknown origin of infection, 3 or more positive blood culture bottles with E. faecalis, community acquisition, and monomicrobial infection (6,7,28,29). As a novel finding, we discovered immunosuppressant therapy to be independently associated with the development of IE in E. faecalis bacteremia. The likely explanation for why immunosuppression is not significant in the univariate analysis is that none of these patients can end up in the community-acquired group because of definitions. Therefore, because community acquisition is a known strong predictor of IE, immunosuppression becomes insignificant in the univariate analysis. However, in the multivariate analysis when adjusting for mode of acquisition, immunosuppression becomes significant. All of these risk factors are central to the clinician when assessing individual patients with E. faecalis bacteremia. Instead of creating another complex risk score, we advise that these factors be included in the combined clinical evaluation of each patient by the endocarditis team. In our study, <10% of the patients were in a true low-risk group (0 risk factors).
The higher in-hospital mortality found in patients with endocarditis is likely related to the acute course of infection with related complications and possible need for urgent high-risk surgery. However, if discharged alive, patients with IE had significantly better 1-year survival than those without IE. This could be explained partly by the slightly lower comorbidity score in the patients with IE as well as a selection bias, with only the strongest patients with IE surviving an extended admission.
On the basis of our findings, we recommend increasing the use of echocardiography in E. faecalis bacteremia. In Staphylococcus aureus bacteremia, the guidelines recommend TEE on the basis of screening studies finding an endocarditis prevalence of 22% to 25% (30,31). We find an endocarditis prevalence of 26% and propose that TEE should be considered in all patients with E. faecalis bacteremia, especially those with relevant risk factors.
Our study was done in a specific geographic region in Scandinavia, with the possible limitation that findings could be different in other areas of the Western world. The properties of the E. faecalis strains, including virulence traits and resistance patterns, are likely related to geography. The study was done in a population-based setting considering all patients with E. faecalis for inclusion in an unselective approach.
The study could have been improved by performing systematic control blood cultures in all patients during and after the treatment to investigate more thoroughly for continuous and relapse bacteremia.
We found a high prevalence of 26% IE in patients with E. faecalis bacteremia and suggest an increase in the use of echocardiography in patients with E. faecalis bacteremia.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Patients with E. faecalis bacteremia have a high prevalence of IE and should undergo echocardiography.
TRANSLATIONAL OUTLOOK: Further research is needed to compare the prevalence of IE in patients with E. faecalis bacteremia in various geographic regions and to ascertain whether strains of E. faecalis with specific genomic traits are more likely to cause endocarditis.
This work was supported by the Research Foundation of the Capital Region of Denmark and the Danish Heart Foundation (grant 13-04-R94-A4470-22761). The study was completely investigator initiated and performed through all its phases. The funders of the study had no influence on study design, data collection, analysis, and interpretation. The funders had no role in writing the manuscript or the decision to submit. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Listen to this manuscript's audio summary by Editor-in-Chief Dr. Valentin Fuster on JACC.org.
- Abbreviations and Acronyms
- infective endocarditis
- odds ratio
- transesophageal echocardiography
- Received February 27, 2019.
- Revision received April 8, 2019.
- Accepted April 22, 2019.
- 2019 American College of Cardiology Foundation
- Amat-Santos I.J.,
- Ribeiro H.B.,
- Urena M.,
- et al.
- Fernández Guerrero M.L.,
- Goyenechea A.,
- Verdejo C.,
- Roblas R.F.,
- de Górgolas M.
- Olmos C.,
- Vilacosta I.,
- Fernández-Pérez C.,
- et al.
- Habib G.,
- Hoen B.,
- Tornos P.,
- et al.
- Habib G.,
- Lancellotti P.,
- Antunes M.J.,
- et al.
- Saby L.,
- Laas O.,
- Habib G.,
- et al.
- Tubiana S.,
- Duval X.,
- Alla F.,
- et al.
- Danish Society of Cardiology
- Dahl A.,
- Rasmussen R.V.,
- Bundgaard H.,
- et al.
- Fowler V.G. Jr..,
- Li J.,
- Corey G.R.,
- et al.