Staphylococcus aureus Endocarditis:  A Consequence of Medical Progress FREE

For decades, infective endocarditis (IE) caused by Staphylococcus aureus has been viewed primarily as a community-acquired disease, especially associated with injection drug use.^1- 7 In contrast, patients with nosocomial or intravascular catheter–associated S aureus bacteremia were considered to be at low risk for IE.^5- 6,8- 11S aureus IE is relatively infrequent at any individual institution, and observations of its characteristics were based primarily upon relatively small samples,^{1,3,6,9,12- 14} single-center experiences,^{5- 6,8- 9,13- 16} or retrospectively identified patients.^{2,7- 8,15- 16} Patient characteristics, treatment practices, and outcomes in these single-center studies often differed considerably. Moreover, because no large, prospectively collected, and geographically diverse cohort of patients with IE existed before now, the impact of regional variations on the characteristics, treatment, and outcome of S aureus IE was unknown.

Recent substantive changes in health care delivery and in antimicrobial resistance patterns have changed the epidemiology of S aureus infections. Rates of S aureus infection, particularly bacteremia associated with health care contact, have increased among hospitalized patients^17- 18 and among those receiving outpatient medical therapy.¹⁹ Rates of infections due to methicillin-resistant S aureus (MRSA) in both hospital¹⁸ and community settings²⁰ have also increased dramatically. Finally, the number of patients with implanted medical devices (eg, prosthetic heart valves, grafts, hemodialysis catheters, pacemakers), a population at high risk for S aureus bacteremia and endocarditis, has also risen in the past 2 decades.^21- 23 However, the extent to which these findings can be generalized is unknown.

The current investigation aimed (1) to document the characteristics of IE caused by S aureus, including IE associated with health care contact and IE due to MRSA, in different parts of the world; and (2) to assess regional differences and the effect of these differences on clinical outcomes among patients with S aureus IE.

Data from the International Collaboration on Endocarditis (ICE) were used for this study. ICE began in June of 1999 after the fifth meeting of the International Society of Cardiovascular Infectious Diseases in Amsterdam, the Netherlands. ICE investigators initiated the ICE-Prospective Cohort Study (ICE-PCS),²⁴ which enrolled 1779 patients in 39 centers in 16 countries between June 15, 2000, and December 31, 2003.

Patients with definite IE were determined by Duke criteria. All patients with IE from sites that met criteria for participation were included in this study. These site criteria included (1) minimum enrollment of 12 cases per year in a center with access to cardiac surgery, (2) patient identification procedures in place to ensure consecutive enrollment and to minimize ascertainment bias,^24- 25 (3) high-quality data with query resolution, and (4) institutional review board (IRB)/ethics committee approval or waiver based upon local standards. Sites not meeting these criteria were excluded from this analysis (3 sites representing 23 patients). The ICE-PCS database is maintained at the Duke Clinical Research Institute, which is the coordinating center for ICE studies with IRB approval. Informed consent (oral or written) was obtained from all patients according to local IRB/ethics committee instructions.

Patients were identified prospectively using site-specific procedures to ensure consecutive enrollment.^24- 25 Physical examination findings and clinical demographic data were recorded at the time of patient enrollment. Patients were enrolled in ICE-PCS if they met criteria for possible or definite IE, based upon the modified Duke criteria.²⁶ Only patients with definite IE were admitted (n = 30 excluded) in the current investigation. To preserve the assumption of independence of observations, only the first episode of S aureus IE recorded for an individual patient was included in the analysis.

A standard case report form was used at all sites to collect data. The case report form included 275 variables and was developed by ICE according to standard definitions.²⁴ Domains included specific data on demographics, medical history, medications, clinical procedures, presence of intravascular devices, predisposing factors, clinical examination, microbiology, serology, nosocomial or community acquisition, antibiotic therapy, echocardiography, surgery, complications, and outcome. Data were collected during the index hospitalization and were then entered at the coordinating center or by the site investigators using an Internet-based data entry system. Queries were developed on all critical variables and were distributed to the sites for reconciliation when necessary. Query responses were reviewed and entered.

Chronic immunosuppressive therapy was defined as the administration of recognized immunosuppressive agents (including oral corticosteroids or other agents such as those used in solid organ transplantation or rheumatologic disorders) for more than 30 days at the time of IE diagnosis.

A cardiac device was defined as a permanent pacemaker, cardioverter-defibrillator, and/or prosthetic cardiac valve. Intravascular access devices were defined as an arterial-venous fistula or an indwelling vascular catheter. A chronic indwelling central catheter was defined as a tunneled, cuffed catheter, or a subcutaneous port catheter. An intravascular access device was presumed to be a possible source of IE if it was present at the onset of IE symptoms.

Vascular evidence of IE was defined as conjunctival hemorrhages, vascular embolic events, or Janeway lesions.²⁷ Immunologic evidence of IE was defined as the presence of Osler nodes or Roth spots.²⁷ Heart murmurs were defined as present or absent according to the physical examination performed by the investigators at the time of initial evaluation. A stroke was defined as an acute neurological deficit of vascular etiology lasting more than 24 hours.²⁸ Systemic embolization was defined as an embolic event outside of the central nervous system. Congestive heart failure was defined according to the New York Heart Association classification system.²⁹ Persistent bacteremia was identified using Duke Endocarditis Service criteria.²⁷

Health care–associated IE was defined as either nosocomial infection or nonnosocomial health care–associated infection. Nosocomial infection was defined as IE developing in a patient hospitalized for more than 48 hours prior to the onset of signs/symptoms consistent with IE.

Nonnosocomial health care–associated infection was defined as IE diagnosed within 48 hours of admission in an outpatient with extensive health care contact as reflected by any of the following criteria: (1) received intravenous therapy, wound care, or specialized nursing care at home within the 30 days prior to the onset of S aureus–IE; (2) attended a hospital or hemodialysis clinic or received intravenous chemotherapy within the 30 days before the onset of S aureus IE; (3) was hospitalized in an acute care hospital for 2 or more days in the 90 days before the onset of S aureus IE; or (4) resided in a nursing home or long-term care facility.¹⁹

Community-acquired IE was defined as IE diagnosed at the time of admission (or within 48 hours of admission) in a patient not fulfilling the criteria for health care–associated infection. Infections were considered to be injection drug use (IDU)–associated if the patient actively used these substances at the time of IE diagnosis and was admitted from the community without an alternate presumed source. Vancomycin therapy was defined as being present if it was identified by the investigator as the predominant antibiotic used in the treatment of the infection.

Geographic regions participating in ICE included the following: United States (9 sites), South America (2 sites each from Brazil, Argentina, and Chile), Australia/New Zealand (7 sites), and Europe/Middle East (17 sites). For multivariate modeling, the European/Middle Eastern region was further subdivided a priori into Northern/Central Europe (a total of 10 sites in France, United Kingdom, Sweden, Germany, Croatia, and the Netherlands) and Southern Europe/Middle East (a total of 7 sites in Italy, Spain, Israel, and Lebanon). Because it contained the highest number of participating centers, Northern/Central Europe was used as the referent category for all multivariate analyses in which the geographic region of the infection was considered.

Continuous variables were represented as medians with 25th and 75th percentiles. Categorical variables were represented as frequencies and percentages of the specified group. Univariate comparisons were made with the Wilcoxon rank-sum test or the χ² test as appropriate. A generalized estimating equation method was used to determine factors that predict S aureus IE among IE patients, in-hospital death among native valve S aureus IE patients, and MRSA among native valve S aureus IE patients.³⁰ Variables found to have a univariate association with the outcome of interest (P<.10) were considered for the final model in a stepwise fashion. The variables included in the final adjusted regression models were selected based on a combination of statistical significance (P<.05) and clinical judgment. The generalized estimating equation method produces consistent parameter estimates that measure association between in-hospital death and the baseline covariates while accounting for the correlation in outcomes of patients from the same hospital. Final parameter estimates were converted to odds ratios (ORs) with corresponding 95% Wald confidence intervals (CIs). The fit of the multivariable models to the data was assessed by the concordance index, which estimates the probability that a patient who had the outcome of interest was assigned a greater probability of sustaining the outcome of interest than was a patient who did not sustain the outcome of interest. For all tests, statistical significance was determined at the .05 level. All statistical analyses were performed using SAS software version 8.2 (SAS Institute, Cary, NC).

During the 48-month study period, 1779 patients with definite IE from 16 countries were enrolled into ICE-PCS. S aureus was the most commonly identified pathogen and was present in 558 patients (31.4%) (Table 1). By univariate analysis, patients with IE due to S aureus were more likely than patients with IE due to other pathogens to be female (P<.001), hemodialysis dependent (P<.001), to have diabetes mellitus (P = .009), or to have other chronic illnesses (P<. 001) (Table 2). Patients with S aureus IE were also significantly more likely to have health care–associated IE (P<.001) than were patients with non-S aureus IE. Cardiac surgery was significantly less frequent among patients with S aureus IE than among patients with non–S aureus IE (P<.001), and it was more likely to be performed due to embolization (P<.001) or persistent bacteremia (P<.001). Relative to patients with non–S aureus IE, patients with S aureus IE also had higher rates of stroke (P<.001), systemic embolization (P = .001), persistent bacteremia (P<.001), and death (P<.001).

Next, a multivariate model was fitted using the entire cohort and the candidate variables listed in Table 2 to identify clinical features independently associated with S aureus IE (concordance index, 0.78). Patient characteristics associated with S aureus IE by multivariate modeling included IDU (OR, 9.3; 95% CI, 6.3-13.7), first clinical presentation less than 1 month after first symptoms (OR, 5.1; 95% CI, 3.2-8.2), health care–associated infection (OR, 2.9; 95% CI, 2.1-3.8), persistent bacteremia (OR, 2.3; 95% CI, 1.5-3.8), presence of a presumed intravascular device source (OR, 1.7; 95% CI, 1.2-2.6), stroke (OR, 1.6; 95% CI, 1.2-2.3), and diabetes mellitus (OR, 1.3; 95% CI, 1.1-1.8).

S aureus IE occurred in 3 distinct clinical settings in this investigation: health care–associated infection (218 patients, 39.1%), community-acquired infection in persons with no history of IDU (209 patients, 37.5%), and community-acquired IDU-associated infection (117 patients, 21.0%) (Table 2). The presumed place of infection was missing in 14 patients (2.5%). Of the 218 patients with health care–associated IE, 131 (60.1%) had nosocomial infection, and 87 (39.9%) had nonnosocomial health care–associated infection. Patients with health care–associated IE were more likely to be from the United States (46.3%) than were patients with community-acquired IDU-associated IE (40.2%) or community-acquired non-IDU–associated IE (21.1%) (P<.001); they were also older (median age, 64.6 vs 36.2 vs 60.6 years, P<.001), more likely to be infected with MRSA (49.1% vs 10.3% vs 13.4%, P<.001), and more likely to have mitral valve involvement (55.8% vs 19.6% vs 49.2%, P<.001). However, they were less likely to have a new/worsening cardiac murmur (33.5% vs 52.1% vs 41.2%, P = .004).

Patient outcomes differed according to the clinical setting. Patients with health care–associated IE had higher rates of in-hospital mortality (29.4%) than did patients with community-acquired IDU-associated IE (11.1%) or community-acquired non-IDU–associated IE (21.1%) (P<.001). Patients with health care–associated IE had higher rates of mortality within each of the 4 geographic regions represented in this study (Figure). They also had higher rates of persistent bacteremia (P<.001). Despite these observations, patients with health care–associated IE were less likely to undergo surgical therapy (33.0%) than were patients with community-acquired IDU-associated IE (35.0%) or community-acquired non-IDU–associated IE (44.5%, P = .04) (Table 3).

Cardiac devices were present in 130 (23.3%) of the patients in the overall cohort with S aureus IE (permanent pacemakers in 36 patients, cardioverter-defibrillators in 5 patients, prosthetic cardiac valves in 64 patients, and more than one device present in 25 patients). Patients with cardiac devices had higher rates of cardiac surgery (46.9% vs 35.6%, P = .02) and in-hospital mortality (27.7% vs 20.0%, P = .07) than did patients without cardiac devices.

A total of 153 patients (27.4%) with S aureus IE were known to be infected with MRSA. We compared the characteristics and outcome of these patients with those of patients with methicillin-susceptible S aureus (MSSA) IE. Patients with IDU were excluded from these analyses because the predominance of MSSA and the low mortality in this population would have confounded outcome comparisons. Of the 424 patients with definite S aureus IE and no history of active IDU, 141 (33.3%) were infected with MRSA (Table 4). Patients with MRSA IE had more chronic comorbid conditions and were more likely to have health care–associated infection (75.9% vs 37.1%, P<.001). MRSA- and MSSA-infected patients had similar rates of cardiac surgery. Complications differed between the 2 groups: MSSA-infected patients had higher rates of systemic embolization (26.2% vs 17.7%, P = .06), whereas MRSA-infected patients experienced higher rates of persistent bacteremia (42.6% vs 8.8%, P<.001). MRSA-infected patients tended to have higher mortality rates than did MSSA-infected patients, although this difference was not statistically significant (29.8% vs 23.3%, P = .14).

A multivariate model was then fit using the patient characteristics and variables presented in Table 4 to identify clinical features independently associated with MRSA IE. Patient characteristics associated with MRSA IE included persistent bacteremia (OR, 6.2; 95% CI, 2.9-13.2), chronic immunosuppressive therapy (OR, 4.1; 95% CI, 2.0-8.6), health care–associated infection (OR, 3.4; 95% CI, 2.1-5.5), a presumed intravascular device source (OR, 2.1; 95% CI, 1.2-3.7), and diabetes mellitus (OR, 2.0; 95% CI, 1.1-3.7).

The characteristics of patients with S aureus IE are presented by geographic region in Table 5. Compared with patients from other regions, patients from the United States were more likely to be hemodialysis dependent, to be diabetic, to have a hemodialysis fistula or a chronic indwelling central catheter as a presumed source of infection, and to receive vancomycin therapy (P<.001 for all comparisons). US patients were also more likely to have nonnosocomial health care–associated IE. Patients from the United States (37.2%) and Brazil (37.5%) were more likely to have MRSA IE than were patients from Europe/Middle East (23.7%) and Australia/New Zealand (15.5%) (P<.001). Although in-hospital mortality rates were similar among regions, United States patients were significantly more likely to develop persistent bacteremia (25.6%, P<.001).

Determinants of outcome of patients with S aureus IE were next evaluated using a multivariate model (Table 6). Patients were included in the model if they had definite native valve S aureus IE and no history of active IDU. Characteristics independently associated with mortality among patients with S aureus IE included stroke (OR, 3.67; 95% CI, 1.94-6.94), persistent bacteremia (OR, 3.06; 95% CI, 1.75-5.35), diagnosis in Southern Europe/Middle East (OR, 3.21; 95% CI, 1.17-10.56), and age in 10-year intervals (OR, 1.49, 95% CI, 1.23-1.81).

Changes in health care delivery and in antimicrobial resistance patterns have altered the epidemiology of S aureus infections. This investigation documents several new and important aspects of S aureus IE in the current era.

S aureus was the most common cause of IE in the overall ICE-PCS cohort, and IE due to S aureus exhibited distinct characteristics as compared with IE due to other pathogens. The finding of S aureus as the leading cause of IE differs from previous reports^12,15 and may be due in part to increasing rates of staphylococcal bacteremia related to health care contact in industrialized nations. For example, rates of sepsis due to gram-positive organisms increased by an average of 26.3% annually from 1979 to 2000, making these bacteria the leading cause of sepsis in the United States.¹⁷ In addition, S aureus was the second most common pathogen isolated in 2 recent, large studies of bloodstream infections in the United States¹⁸ and Europe.³¹ Our findings emphasize that, in addition to sepsis and bacteremia, S aureus is now also the leading cause of IE in many regions of the globe.

The current investigation definitively establishes the global emergence of health care contact as a critical risk factor for S aureus IE. In contrast to older, smaller, and often retrospective reports, health care–associated infection was the single most common form of S aureus IE in this study. Indeed, health care-associated infection accounted for one quarter to one half of S aureus IE cases reported in the represented geographic regions. In a significant proportion of these patients, an intravascular device was the presumed source of bacteremia, confirming previous single-center observations.^{13- 14,16,32- 34} Prosthetic cardiac devices (pacemakers, defibrillators, and prosthetic cardiac valves) were present in almost one quarter of the patients in this investigation. This finding is consistent with a recent report using Medicare claims data to demonstrate increasing numbers of cardiac device infections over the previous decade.²¹S aureus IE was also associated with immunosuppression in this and prior reports.²⁵ This association may be due to either the health care for underlying diseases that are being treated with immunosuppressive therapy or the immunosuppression itself. Taken together, these findings indicate an important shift in the epidemiology of S aureus IE: it is increasingly a consequence of medical progress.

Based upon the findings in this investigation, we suggest that health care–associated IE is a distinct clinical subtype of S aureus IE, distinguished by a relative infrequency of classic clinical stigmata of IE, a predominance of mitral valve involvement, and a high mortality rate. The lower rates of surgery among patients with health care-associated S aureus IE may reflect a higher rate of comorbid conditions or advanced age, a lower rate of clinical recognition of IE, or the higher rate of stroke among patients with S aureus IE in this study. These findings underscore the need to reduce the number of health care–associated infections through improved compliance with infection control procedures, as well as the need for new interventions and preventive strategies.

This investigation also conclusively establishes MRSA as a significant cause of IE internationally, accounting for almost 40% of the IE caused by S aureus in certain regions. Not surprisingly, patients with MRSA IE were significantly more likely to have preexisting chronic conditions and health care–associated IE by both univariate and multivariate modeling. One distinct characteristic of MRSA IE suggested in the current investigation was persistent bacteremia. This finding is consistent with previous reports^35- 36 and may be due in part to the use of vancomycin, an agent associated with delayed bactericidal activity against S aureus.^35- 37 Despite a higher rate of persistent bacteremia, the difference in mortality between MRSA- and MSSA-infected subjects did not achieve statistical significance. This observation may be due in part to the overall high mortality of S aureus IE (regardless of the antimicrobial resistance profile of the infecting pathogen) or due simply to an insufficient sample size.

It is important to note that approximately 20% of patients with MRSA IE developed their infection in the absence of identifiable health care contact. Community-acquired MRSA cases differ demographically from health care–associated MRSA cases; MRSA isolates from community-acquired infections commonly possess distinct exotoxin gene profiles (eg, Panton-Valentine leukocidin genes) compared with health care–associated isolates.²⁰ The findings of the current investigation, coupled with the increasing number of reports of community-acquired MRSA infection, suggest that these strains may be an emerging cause of IE in many regions of the world.

One clinical profile seen almost exclusively in US patients was the development of IE among non-hospitalized subjects with extensive health care contact, often in the context of long-term intravascular access. Almost one third of US patients developed nonnosocomial health care–associated IE, and in approximately 20%, the presumed source was a chronic indwelling central catheter. This finding may reflect the recent increasing trend in the United States toward outpatient medical therapy, such as long-term antibiotics, parenteral nutrition, and hemodialysis. United States patients with S aureus IE were also significantly more likely to be infected with MRSA and to develop persistent bacteremia. The higher rates of persistent bacteremia in US patients may be due in part to the fact that they were more likely to receive vancomycin.

While the ICE-PCS represents the largest and most geographically diverse collection of prospectively identified patients with S aureus IE ever assembled, this investigation has limitations. This is an observational study of patients with definite IE from centers that have self-selected to participate in the ICE-PCS. Therefore, a population sample was not obtained from any specific region, limiting any epidemiologic inferences. Similarly, as only a select group of centers within each geographic region participated in the current investigation, regional characterization is based solely on data from these centers. In addition, these hospitals are typically referral centers that have cardiac surgical programs. Therefore, the results of this study may not generalize to other patient populations with S aureus IE or to patients receiving care at other types of medical centers.

S aureus is now the most common cause of IE in many areas of the developed world. Patients with IE due to S aureus exhibit distinct characteristics compared with patients with IE due to other pathogens. Health care–associated IE is emerging as the most common form of S aureus IE and has distinct features compared with more familiar forms of S aureus IE, such as community-acquired IDU-associated infection. MRSA is now encountered internationally as a relatively common cause of IE and is associated with persistent bacteremia. Future investigations are required to identify better treatment and prevention strategies for this serious and common consequence of medical progress.