Infective Endocarditis Due to Staphylococcus aureus

106 Infective Endocarditis Due to Staphylococcus aureus: 59 Prospectively Identified Cases with Follow-up Vance G. Fowler, Jr., Linda L. Sanders, Li Kuo Kong, R. Scott McClelland, Geoffrey S. Gottlieb, Jennifer Li, Thomas Ryan, Daniel J. Sexton, Georges Roussakis, Lizzie J. Harrell, and G. Ralph Corey From the Department of Medicine, Divisions of Infectious Diseases and Cardiology, the Department of Pediatrics, and the Department of Microbiology and Clinical Microbiology Laboratory, Duke University Medical Center, Durham, North Carolina Fifty-nine consecutive patients with definite Staphylococcus aureus infective endocarditis (IE) by the Duke criteria were prospectively identified at our hospital over a 3-year period. Twenty-seven (45.8%) of the 59 patients had hospital-acquired S. aureus bacteremia. The presumed source of infection was an intravascular device in 50.8% of patients. Transthoracic echocardiography (TTE) revealed evidence of IE in 20 patients (33.9%), whereas transesophageal echocardiography (TEE) revealed evidence of IE in 48 patients (81.4%). The outcome for patients was strongly associated with echocardiographic findings: 13 (68.4%) of 19 patients with vegetations visualized by TTE had an embolic event or died of their infection vs. five (16.7%) of 30 patients whose vegetations were visualized only by TEE (P õ .01). Most patients with S. aureus IE developed their infection as a consequence of a nosocomial or intravascular device – related infection. TEE established the diagnosis of S. aureus IE in many instances when TTE was nondiagnostic. Visualization of vegetations by TTE may provide prognostic information for patients with S. aureus IE. Employing these two diagnostic aids, we prospectively evaluated all patients at Duke University Medical Center (Durham, NC) who developed S. aureus bacteremia during the period from September 1994 to January 1998. In contrast to earlier investigators, we found that most of our patients with IE had nosocomial and/or intravascular catheter – related infections. The following study of all patients with S. aureus IE from our registry was undertaken to examine the risk factors, clinical characteristics, diagnosis, and outcome for 59 consecutive patients with definite S. aureus IE by the Duke criteria. See editorial response by Watanakunakorn on pages 115 – 6. Methods Most previously reported series of patients with S. aureus IE were retrospective and/or included relatively few patients. These series also used clinical definitions in the diagnosis of IE. Because of the difficulty of clinically diagnosing S. aureus IE [7], evaluation of the data in these reports is difficult [8, 9]. Two recent advances have significantly improved the clinician’s ability to identify IE: the development of the Duke criteria for the diagnosis of IE [10, 11] and the widespread use of transesophageal echocardiography (TEE) for the detection of vegetations on cardiac valves [12, 13]. Received 29 April 1998; revised 4 September 1998. Financial support: V.G.F. was supported by a Health Services Research and Development Fellowship from the Veterans Administration Medical Center, Durham, North Carolina. Reprints or correspondence: Dr. Vance G. Fowler, Jr., P.O. Box 3824, Duke University Medical Center, Durham, North Carolina 27710 (fowle007@mc. duke.edu). Clinical Infectious Diseases 1999;28:106–14 q 1999 by the Infectious Diseases Society of America. All rights reserved. 1058–4838/99/2801–0016$03.00 / 9c5e$$ja11 12-30-98 16:01:10 Subjects and Setting The clinical microbiology laboratory notified a member of our research group of all patients at Duke University Medical Center for whom blood cultures were positive for S. aureus during the period from September 1994 to January 1998. One of the authors reviewed the records and clinical features of each case within 36 hours of notification. Patients with S. aureus bacteremia who were outpatients or younger than 18 years of age, as well as those with polymicrobic infection, those with neutropenia (WBC count, õ1.0 1 109/L), or those who died before positive results of blood cultures were known, were excluded from the study. Twenty-five of the patients described in the present analysis were included in a previous study [14]. Clinical Features Each eligible patient was evaluated for the presence of a potential source of bacteremia. Clinical signs suggesting IE in each patient were specifically sought. Staphylococcal tissue cida UC: CID Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 Infective endocarditis (IE) due to Staphylococcus aureus is a potentially lethal disease [1 – 3]. For decades, S. aureus IE was assumed to be primarily a community-acquired disease, especially when there was no identifiable source of infection. Although patients with community-acquired S. aureus bacteremia remain at high risk for IE, the recent increase in the frequency of nosocomial and ‘‘intravascular device – associated’’ S. aureus bacteremia [4 – 6] has resulted in an increase in the number of patients at risk for IE. CID 1999;28 (January) S. aureus Infective Endocarditis infection was considered to be the source of bacteremia if clinical signs of local infection antedated the bacteremia. An intravascular catheter was considered to be the portal of entry if inflammation was present around the catheter insertion site and/or a catheter-tip culture was positive for S. aureus and no other source was evident [15]. IE was considered to be community-acquired if blood specimens for positive cultures were obtained within 72 hours of admission. IE was considered to be hospital-acquired if a blood specimen for a positive culture was obtained ú72 hours after hospitalization. Definition of IE IE was defined according to the Duke criteria [11]. 107 Pulsed-Field Gel Electrophoresis Pulsed-field gel electrophoresis (PFGE) was performed on all isolates of S. aureus from patients in whom recurrence was documented to confirm genetic similarity, as described by Kong et al. [16]. Single isolates of S. aureus grown overnight at 377C in broth medium were embedded into small agarose plugs, digested with lysostaphin and lysozyme, and deproteinized with use of a reagent kit (GenePath I, Bio-Rad Laboratories, Hercules, CA). Restriction enzyme digestion was done with SmaI, and slices of the plugs underwent electrophoresis on 1% agarose gels and were stained with ethidium bromide; the results were interpreted according to previously reported guidelines [17]. Statistical Analysis All patients were contacted 12 weeks after the date of the first blood culture. When the patient could not be interviewed directly, a family member or the patient’s primary care physician was contacted and questioned. Four primary end points were defined as follows: cure, no evidence of recurrent staphylococcal infection within the 12-week follow-up period; relapse, clinical resolution of the initial episode of infection after treatment but culture-confirmed recurrent S. aureus infection documented within the follow-up period; death due to S. aureus bacteremia, persistent signs or symptoms of infection, positive blood cultures, or a persistent focus of infection at the time of death in the absence of another explanation for death; and death due to underlying disease, death due to a defined underlying disease other than staphylococcal bacteremia. For patients who died of other causes during hospitalization, death due to underlying causes was defined based on evaluation by one of the investigators. After discharge from the hospital, death due to underlying causes was defined based on the judgment of the patient’s primary physician, review of hospitalization records, and/or the listed cause of death on the patient’s death certificate. To preserve the statistical assumption of independence of observations, only the initial episodes of S. aureus IE were included in the study. Descriptive statistics for continuous variables were summarized in terms of medians and interquartile ranges (IQRs). Categorical variables were reported in terms of the number and percentage of patients affected. Wilcoxon ranksum and Fisher’s exact tests were used to evaluate group differences for continuous and categorical variables, respectively. The medians and IQRs reported for the duration of therapy and time to defervescence included a small number of patients (10 and three, respectively) who died before their therapy ended or before their fever abated. A separate analysis (data not shown) excluding these patients did not change the results from the group comparisons. Results were considered significant at a P value of £.05. Echocardiography Specific echocardiographic findings were analyzed as individual indicators of endocarditis as previously described [14]. All echocardiograms were read at the time of procedure by an experienced echocardiographer. In most cases, TEE was performed after transthoracic echocardiography (TTE). Because of the potential bias created by sequential readings, all TEEs were later reinterpreted in a blinded fashion by one of the investigators (T.R. or G.R.). To minimize the potential for bias created by knowing that all patients in the series had S. aureus bacteremia, 20 randomly selected TEEs were included for review. This exercise was designed to reduce the pretest likelihood of disease in the series of echocardiograms being blindly interpreted. The results of the initial clinical interpretation were used for all subsequent data analyses. / 9c5e$$ja11 12-30-98 16:01:10 Results Patient Characteristics A total of 477 adult patients had one or more blood cultures positive for S. aureus during the period from September 1994 to January 1998. Of these 477 patients, 59 (12.4%) had definite IE. Six patients had histologically confirmed lesions or a positive culture of resected heart valves or vegetations; the remaining 53 patients were considered to have definite IE on the basis of the Duke clinical criteria. Forty-five of these 53 patients had echocardiographic evidence of IE and persistently positive blood cultures (two major criteria). The eight remaining patients had persistently positive blood cultures plus at least three minor criteria, such as fever, predisposing conditions, and vascular phenomena. Follow-up data were obtained for all 59 patients. The median age of the 59 patients with IE was 58 years (IQR, 42 – 67 years); 59.3% were male, and 47.5% were white (table 1). Twenty-seven patients (45.8%) had hospital-acquired IE and 32 (54.2%) had community-acquired IE. cida UC: CID Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 Outcome for Patients 108 Fowler et al. CID 1999;28 (January) Table 1. Characteristics of 59 patients with IE due to Staphylococcus aureus. Characteristic Total (n Å 59) Median age (range) in y 58.0 (42.0 – 67.0) Male sex 35 (59.3) Race White 28 (47.5) Black 27 (45.8) Comorbidity Hemodialysis 22 (37.3) Diabetes 13 (22.0) Steroid therapy 6 (10.2) Neoplasm 7 (11.9) HIV infection 6 (10.2) Known preexisting risk factor for IE Any 27 (45.8) Valvular regurgitation 13 (22.0) Injection drug use 9 (15.3) Prosthetic valve 9 (15.3) Previous IE 4 (6.8) Pacemaker 4 (6.8) Other* 3 (5.1) Hospital-acquired IE Community-acquired IE (n Å 27) (n Å 32) P value 60.0 (50.0 – 67.0) 14 (51.9) 44.5 (39.0 – 67.0) 21 (65.6) NS NS 15 (55.6) 10 (37.0) 13 (40.6) 17 (53.1) NS 6 5 1 4 1 (22.2) (18.5) (3.7) (14.8) (3.7) 16 8 5 3 5 (50.0) (25.0) (15.6) (9.4) (15.6) .03 NS NS NS NS 10 5 1 4 1 2 1 (37.0) (18.5) (3.7) (14.8) (3.7) (7.4) (3.7) 17 8 8 5 3 2 2 (53.1) (25.0) (25.0) (15.6) (9.4) (6.3) (6.3) NS NS .03 NS NS NS NS Sources of Infection Clinical and Laboratory Features The most common source of staphylococcal infection was an intravascular device (30 patients [50.8%]) (table 2). Importantly, 16 patients (53.3%) with intravascular device – associated IE had a community-acquired infection. Eleven patients (34.4%) with community-acquired device-associated IE had cuffed indwelling catheters, and five patients (15.6%) had hemodialysis grafts. Cuffed indwelling catheters (e.g., permanent and Hickman catheters) were the presumed source of bacteremia in 25.4% (15) of all cases of IE. Temporary catheters such as central, peripheral, and arterial catheters inserted during hospital admission were the presumed source of bacteremia in 13.6% (eight) of the cases. Fifteen patients (25.4%) developed IE after surgical procedures. In 14 (93.3%) of these 15 patients, a surgical wound was the presumed source of S. aureus bacteremia. Median sternotomy wounds (seven patients) were the most common source of infection in these postoperative cases; other sources included abdominal surgical wounds (4 patients), orthopedic surgical wounds (2), and a urologic surgical wound (1). An intravascular catheter was the source of infection in one postoperative case. Eight (53.3%) of the patients with postoperative IE had underlying valvular disease: two of these eight patients had preexisting valvular regurgitation, and the remaining six patients had prosthetic cardiac valves. In four of the six patients with prosthetic cardiac valves, postoperative prosthetic valve IE involved newly placed valves. No definite source of infection could be identified in 13 patients (22.0%). All of these 13 patients had communityacquired infection, and eight were injection drug users. / 9c5e$$ja11 12-30-98 16:01:10 Symptoms and signs in patients with community- and hospital-acquired IE were similar (table 3). All patients were febrile at the time of diagnosis. Cardiac murmurs were common at the time of initial examination (46 patients [78.0%]); however, new murmurs were recognized in only seven patients (11.9%). Thirteen patients with IE (22.0%) had no murmur at the time of diagnosis. The median number of positive blood cultures for patients with hospital-acquired IE and those with communityacquired IE was four (IQR, 3 – 5). Three significant differences between patients with community- and hospital-acquired IE were noted. Patients with community-acquired IE were significantly more likely than patients with hospital-acquired IE to have vascular phenomena (56.3% vs. 25.9%, respectively; P Å .03) and no evident source of their bacteremia (40.6% vs. zero, respectively; P õ .01). Patients with hospital-acquired IE were significantly more likely than patients with community-acquired IE to have IE due to methicillin-resistant S. aureus (MRSA) (55.6% vs. 18.8%, respectively; P õ .01). Echocardiography Clinical and investigator interpretative agreement was very good (49 of 52; k Å 0.70). None of the 20 randomly selected TEEs was interpreted as demonstrating IE. For three patients, the investigator interpretation was negative for IE, while the clinical interpretation was positive. For all three patients, the clinical interpretation was used for subsequent analysis. cida UC: CID Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 NOTE. IE Å infective endocarditis. Data are no. (%) of patients with IE unless stated otherwise. * Rheumatic heart disease, valvular calcifications, and valvular prolapse. CID 1999;28 (January) S. aureus Infective Endocarditis 109 Table 2. Characteristics of infection in 59 patients with IE due to Staphylococcus aureus. Total (n Å 59) Characteristic Presumed source of infection Intravascular catheter Cuffed indwelling* Central Peripheral Arterial Hemodialysis graft† Surgical wound None‡ Other§ Median duration of symptoms before therapy (range) in d Time to defervescence after therapy (range) in d Laboratory data Methicillin-resistant S. aureus WBC count at diagnosis (range) in 1109/L Hematocrit at diagnosis (range) in % 23 15 4 3 1 7 14 13 2 Hospital-acquired IE (n Å 27) (39.0) (25.4) (6.8) (5.1) (1.7) (11.9) (23.7) (22.0) (3.4) 21 (35.6) 12 4 4 3 1 2 12 0 1 (44.4) (14.8) (14.8) (11.1) (3.7) (7.4) (44.4) (3.7) Community-acquired IE (n Å 32) 11 11 0 0 0 5 2 13 1 P value (34.4) (34.4) NS (15.6) (6.3) (40.6) (3.1) NS õ.01 õ.01 NS 3.0 (1.0 – 4.0) 3.0 (1.5 – 5.5) NS 4.0 (3.0 – 9.0) 5.0 (3.0 – 8.0) NS 15 (55.6) 6 (18.8) õ.01 12.8 (10.4 – 18.9) 12.1 (8.3 – 17.3) NS 30.0 (28.0 – 36.0) 30.0 (25.0 – 34.5) NS Echocardiography for all three of these patients demonstrated significant new valvular regurgitation. One of the three patients had a perforated valve and died of S. aureus IE. TEE revealed an aortic valve vegetation and new aortic insufficiency in the second patient. This patient ultimately underwent valve replacement after the 12-week follow-up interval because of perforation of the aortic valve. TEE disclosed that the third patient had new aortic regurgitation and multiple small oscillating masses after developing a relapse of S. aureus bacteremia (as confirmed by PFGE). Echocardiographic findings contributed to the diagnosis of definite IE in 54 (91.5%) of the 59 patients (table 4). Fifty-eight patients underwent TTE, which revealed diagnostic findings of IE in 20 patients (34.5%). Fifty-one patients underwent both TTE and TEE. TTE revealed vegetations in 15 (29.4%) of these patients, while TEE detected findings leading to the diagnosis of IE in 48 patients (94.1%) (vegetations, 44 patients; paravalvular abscesses, 4; perforated valves, 3). In one of these 51 patients, a mitral valve vegetation was revealed by TTE but not by TEE. One patient underwent only TEE. Therapy Twenty-nine patients (49.2%) were treated with vancomycin; 21 (72.4%) of these 29 patients had infection due to MRSA. / 9c5e$$ja11 12-30-98 16:01:10 Thirty patients (50.8%) were treated with a b-lactam antibiotic: 14 received nafcillin (usual dose, 2 g every 4 – 6 hours), 14 received cefazolin (usual dose, 2 g q8h), and 2 received other b-lactam agents. The median time to defervescence after initiation of treatment was 5 days. The median duration of therapy for all patients was 42 days (IQR, 29 – 44 days). Patients with hospital-acquired IE were treated for a median of 42 days (IQR, 28 – 43 days). Patients with community-acquired IE received therapy for a median of 43 days (IQR, 29 – 44.5 days). Five patients underwent valvular resection, and pacemaker devices were removed from three patients. Eight additional patients were evaluated for valvular replacement, but surgical consultants concluded that they were not candidates for operative procedures. Six (75%) of these eight patients died. Two other patients declined valve replacement; both were alive at follow-up 12 weeks later. Outcome At the time of follow-up 12 weeks after the initial positive blood culture, 38 patients (64.4%) were cured, 6 (10.2%) relapsed, and 2 (3.4%) died of underlying conditions. A total of 13 patients (22.0%) died of their infection, 7 (11.9%) had a cerebrovascular event, and 7 (11.9%) developed signs or symptoms of congestive heart failure (table 5). Patients with cida UC: CID Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 NOTE. IE Å infective endocarditis. Data are no. (%) of patients with indicated type of IE unless stated otherwise. * Permanent catheter (9 patients), Port-a-catheter (Becton Dickinson Microbiology Systems, Cockeysville, MD; 2), and Hickman catheter (2). † Synthetic vascular graft. ‡ Includes eight injection drug users. § Decubitus ulcer (1 patient) and foot ulcer (1). 110 Fowler et al. CID 1999;28 (January) Table 3. Clinical findings at the time of diagnosis for 59 patients with IE due to Staphylococcus aureus. No. (%) of patients Finding Fever Chills Cardiac murmur New Sepsis syndrome Congestive heart failure Any vascular phenomenon Peripheral embolic lesions* Pulmonary infarcts Cerebrovascular emboli Conjunctival hemorrhage Other site Autoimmune phenomenon Glomerulonephritis Roth’s spots or Osler’s nodes Total (n Å 59) 59 41 46 7 26 17 25 8 8 7 2 7 Hospital-acquired IE (n Å 27) (100.0) (69.5) (78.0) (11.9) (44.1) (28.8) (42.4) (13.6) (13.6) (11.9) (3.4) (11.9) 27 17 22 4 12 8 7 4 0 2 1 3 2 (3.4) 0 Community-acquired IE (n Å 32) (100.0) (63.0) (81.5) (14.8) (44.4) (29.6) (25.9) (14.8) 32 24 24 3 14 9 18 4 8 5 1 4 (7.4) (3.7) (11.1) 2 (7.4) 0 (100.0) (75.0) (75.0) (9.4) (43.8) (28.1) (56.3) (12.5) (25.0) (15.6) (3.1) (12.5) 0 0 P value NS NS NS NS NS .03 NS õ.01 NS NS NS NS S. aureus IE who had a cerebrovascular event were significantly more likely to die of their infection than were those who did not develop this complication (four [57.1%] of seven vs. nine [17.3%] of 52, respectively; P õ .04, Fisher’s exact test). Six patients developed recurrent bacteremia during the 12week follow-up period. Five (83.3%) of these six patients were initially treated with vancomycin. Of these six patients, two were undergoing chronic hemodialysis, and three were infected with MRSA. Four of these six patients relapsed after 6 weeks of intravenous antibiotic therapy. PFGE was performed on isolates from five of these six patients as reported elsewhere [16]. All five paired isolates were identical, thus suggesting relapse rather than reinfection. Twenty-four (40.7%) of the 59 patients had an indwelling foreign body prior to developing IE (prosthetic heart valve, 9 patients; orthopedic devices, 3; pacemaker devices, 4; vascular Table 4. Echocardiographic features of 54 cases of IE due to Staphylococcus aureus. No. (%) of cases Feature(s) Total (n Å 54) Vegetation(s) Aortic Mitral Tricuspid Other* More than one site† Vegetation seen only by TEE Abscess‡ Valve perforation§ 49 13 20 8 4 4 30 4 6 (90.7) (24.1) (37.0) (14.8) (7.4) (7.4) (55.6) (7.4) (11.1) Hospital-acquired IE (n Å 25) 22 7 9 3 2 1 16 2 2 (88.0) (28.0) (36.0) (12.0) (8.0) (4.0) (64.0) (8.0) (8.0) Community-acquired IE (n Å 29) 27 6 11 5 2 3 14 2 4 (93.1) (20.7) (37.9) (17.2) (6.9) (10.3) (48.3) (6.9) (13.8) P value NS NS NS NS NS NS NS NS NS NOTE. IE Å infective endocarditis. * Includes vegetations on pacemaker wire (3 patients) and intracardiac vegetations (1). † Includes vegetations on tricuspid valve and pacemaker wire (2 patients), vegetations on aortic valve and mitral valve (1), and vegetations on mitral valve and tricuspid valve (1). ‡ Includes abscess and mitral valve vegetation (1 patient), abscess and perforated mitral valve (1), and prosthetic mitral valve abscess (1). § Includes aortic valve perforation with mitral and aortic valve vegetations (2 patients) and mitral valve perforation (1). / 9c5e$$ja11 12-30-98 16:01:10 cida UC: CID Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 NOTE. IE Å infective endocarditis. * Includes Janeway lesions. CID 1999;28 (January) S. aureus Infective Endocarditis 111 Table 5. Complications in 59 patients with IE due to Staphylococcus aureus. No. (%) of patients Complication Total (n Å 59) Death due to S. aureus infection Recurrent bacteremia Cerebrovascular event Congestive heart failure Any metastatic infection Meningitis Arthritis Epidural abscess Osteomyelitis Other 13 6 7 7 20 3 5 4 4 6 Hospital-acquired IE (n Å 27) (22.0) (10.2) (11.9) (11.9) (33.9) (5.1) (8.5) (6.8) (6.8) (10.2) 7 2 2 3 6 0 0 2 1 3 (25.9) (7.4) (7.4) (11.1) (22.2) (7.4) (3.7) (11.1) Community-acquired IE (n Å 32) 6 4 5 4 14 3 5 2 3 3 P value (18.8) (12.5) (15.6) (12.5) (43.8) (9.4) (15.6) (6.3) (9.4) (9.4) NS NS NS NS NS NS NS NS NS NS NOTE. IE Å infective endocarditis. mial IE has often been assumed to be uncommon [18], almost one-half of our patients acquired IE while in the hospital. There are several possible explanations for this high rate of nosocomial S. aureus IE. First, the incidence of nosocomial S. aureus bacteremia has increased in the United States [4 – 6] and abroad [19]. For example, since 1980, investigators from the National Nosocomial Infections Surveillance System of the Centers for Disease Control and Prevention reported increases in the rate of S. aureus bacteremia that ranged from 122% to 283% in individual hospitals [4]. Second, during this same period, the prevalence of both S. aureus IE [2, 9] and nosocomial IE [1, 18, 20, 21] also increased. For example, Fernández-Guerrero et al. [20] reported that the number of cases of nosocomial IE (most of which were due to S. aureus) from 1978 to 1992 was 10fold greater than the number of cases occurring from 1960 to 1975. In addition, there has been a significant increase in Discussion NOTE. IE Å infective endocarditis; TEE Å transesophageal echocardiography; TTE Å transthoracic echocardiography. Only detection of a vegetation was included. Other echocardiographic evidence of endocarditis such as paravalvular abscess or valvular perforation was excluded. To our knowledge, this study is the first prospective analysis of S. aureus IE with use of the Duke criteria. Although nosoco- / 9c5e$$ja11 12-30-98 16:01:10 Table 6. Outcome for 49 patients with IE due to Staphylococcus aureus by echocardiographic findings of a vegetation: TTE vs. TEE. No. (%) of patients with vegetation Visualized by TTE (n Å 19) Outcome Death due to S. aureus infection Any major embolic event Cerebrovascular Pulmonary Other Major embolic event or death due to S. aureus infection cida UC: CID 6 11 5 5 4 (31.6) (57.9) (26.3) (26.3) (21.1) 13 (68.4) Visualized only by TEE (n Å 30) 2 3 1 2 2 P value (6.7) (10.0) (3.3) (6.7) (6.7) .04 õ.01 .03 NS NS 5 (16.7) õ.01 Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 graft, 8). In 16 (66.7%) of these 24 patients, the foreign body was thought to be infected prior to the recognition of IE. Twelve patients developed nosocomial S. aureus IE associated with an intravascular catheter. Four of these patients had tunneled intravascular catheters. The remaining eight patients (13.6% of all study patients) developed IE as a complication of bacteremia associated with a temporary intravascular catheter placed during their hospitalization (table 2). For these eight patients, the median duration from placement of the intravascular catheter to the onset of catheter infection was 6 days (IQR, 2 – 8.5 days). For six of the eight patients, diagnostic echocardiographic findings were demonstrated only by TEE. Three (37.5%) of these eight patients died of their infection. The rates of mortality, relapse, and cure among patients with hospital- and community-acquired IE were similar (table 5). Rates of complications (16 [42.1%] of 38 vs. 12 [57.1%] of 21, respectively; P Å NS) and mortality rates (eight [21.1%] of 38 vs. five [23.8%] of 21, respectively; P Å NS) among patients with infection due to methicillin-susceptible S. aureus (MSSA) and MRSA were similar. Rates of treatment failure (defined as mortality due to S. aureus bacteremia or relapse) were not significantly different between patients treated with vancomycin and those treated with a b-lactam antibiotic (12 [41.4%] of 29 vs. seven [23.3%] of 30, respectively; P Å NS). There was no difference in outcome between patients who did and did not receive adjunctive therapy with an aminoglycoside or rifampin. Outcome was associated with echocardiographic findings at the time of diagnosis (table 6). Of the 19 patients whose vegetations were visualized by TTE, 13 (68.4%) had an embolic event or died of their infection. In contrast, only five (16.7%) of the 30 patients whose vegetations were visualized only by TEE had an embolic event or died of their infection (P õ .01). 112 Fowler et al. / 9c5e$$ja11 12-30-98 16:01:10 were visible by TTE were significantly more likely to have embolic phenomena or die of their infection than were patients whose vegetations were detectable only by TEE. Although some investigators failed to detect significant correlations between vegetation size and frequency of complications [31, 32], other researchers showed that vegetation size was associated with clinical outcome [33 – 37]. To our knowledge, a prospective comparison of the clinical outcome for patients with S. aureus IE according to the echocardiographic method by which the valvular vegetation was detected has not been previously reported. Our findings support the hypothesis that the predictive value of echocardiography may be due to the fact that TEE is better able to detect small (£8 mm) vegetations that in turn are less likely to be associated with complications [34]. Such small vegetations may also represent an earlier stage of S. aureus IE that is more likely to respond to antibiotic therapy. Although we found that the form of echocardiographic detection of vegetations correlated well with outcome, confirmation of this observation by other investigators is needed. There were several questions concerning S. aureus IE that our study was unable to address. For example, some investigators reported the clinical outcome for patients with MRSA bacteremia to be worse than that for patients with MSSA bacteremia [27, 38, 39]; other researchers failed to detect significant correlation between methicillin susceptibility and clinical outcome [40]. Our results failed to find any difference in the outcome for patients with IE due to MRSA or MSSA, but the number of patients in each group was insufficient to resolve this important question. Another question our study could not answer was the impact of therapy on the outcome of IE. The efficacy of vancomycin as an antistaphylococcal agent was recently questioned by several investigators [41 – 44]. We were unable to detect a statistically significant difference in outcome between patients who were treated with vancomycin and those treated with b-lactam agents, an observation also made by previous investigators [3]. However, a trend toward an increased relapse rate among patients treated with vancomycin was apparent. Indeed, 83.3% of our patients who relapsed were treated with vancomycin. Therefore, further investigation into the possible limitations of vancomycin in the treatment of S. aureus IE is warranted. Our study has several limitations. First, the diagnosis of definite IE in our study was based on clinical, microbiological, and echocardiographic findings rather than pathological features. It is possible that we overdiagnosed IE on the basis of TEE findings. Many patients had small vegetations that we used as a major criterion to diagnose IE. Second, referral bias and the small percentage of injection drug users (15.3%) also may have affected our results. Third, we did not study all patients with S. aureus bacteremia by means of echocardiography. Patients with complications were in general more likely to undergo echocardiography. Because patients with S. aureus bacteremia and complications were more likely to undergo echocardiography that subsequently detected vegetations, the cida UC: CID Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 the number of patients with risk factors for nosocomial IE, such as those patients with prosthetic cardiac valves who develop nosocomial bacteremia [22]. Finally, because TEE is more sensitive than TTE in the diagnosis of endocarditis [12], its use has very likely contributed to the trend toward both earlier and more accurate diagnosis of hospital-acquired S. aureus IE. For example, we [14] recently reported a series of 103 consecutive patients with S. aureus bacteremia who underwent both TTE and TEE; of the 26 patients found to have definite IE by the Duke criteria, only seven patients had echocardiographic evidence of IE by TTE. The high diagnostic yield of TEE suggests that clinicians evaluating staphylococcemic patients may often wish to proceed directly to this diagnostic test. When we compared patients with community-acquired S. aureus IE with those with hospital-acquired S. aureus IE, we found both similarities and differences. Our patients with hospital-acquired IE were older and most had normal heart valves, findings similar to those of other case series [3, 18, 20, 21]. Also similar to the findings in other reports, vascular phenomena were significantly less common in patients with hospital-acquired IE [23], thus suggesting a shorter time to diagnosis [24]. All of our patients with hospital-acquired IE had an identifiable source for their bacteremia [25], while many patients with community-acquired IE (13 [40.6%] of 32 patients) had no apparent source for their infection. Our findings also reinforce the observations of more recent investigations [3, 6, 26] that intravascular device – related S. aureus bacteremia is an emerging problem in both hospital and nonhospital settings. Over one-half (50.8%) of the 59 patients with definite IE identified over a 3-year period developed S. aureus IE as a consequence of an infected intravascular device. Furthermore, intravascular devices were the dominant source of community-acquired IE (50.0% of cases). Finally, 13.6% of our patients developed S. aureus IE as a consequence of an infected intravascular catheter placed while the patient was hospitalized for another medical condition. Our observation that intravascular devices now represent an important source for S. aureus IE is consistent with the findings of some [3, 21, 23, 24] but not all [27, 28] previous series of patients with S. aureus bacteremia and IE. Recently, a few investigators pointed out the important distinction between bloodstream infections in patients requiring frequent contact with health care settings and those whose infections are truly community-acquired [6, 29, 30]. Graham and colleagues [29] even suggested the term nosohusial to describe infections occurring in patients who receive intravenous treatment at home. Applying the terminology of Graham et al. to the present study, 45.8% of the patients had hospitalacquired IE, 27.1% of patients had nosohusial IE, and only 27.1% had true community-acquired IE. A final important but unexpected finding from this study is that TTE identification of vegetations in patients with S. aureus IE may be prognostically important. Patients whose vegetations CID 1999;28 (January) CID 1999;28 (January) S. aureus Infective Endocarditis overall rate of complications in our patients with S. aureus IE may have been inflated as a result of selection bias. Finally, the observational nature of this study limited our ability to study the impact of different diagnostic and therapeutic approaches on the clinical outcome for patients with S. aureus IE. Despite these potential confounding factors, the mortality and complication rates that we observed were similar to those reported in a recent large report [3], thus suggesting that the diagnosis of IE in our study was accurate even for patients whose vegetations were detected only by TEE. Our results document that changes in the epidemiology of S. aureus bacteremia have occurred and illustrate the fact that intravascular devices are an increasing cause of S. aureus IE. As outpatient therapy replaces inpatient therapy for complicated medical diseases and intravenous therapy is more frequently provided in the outpatient setting, more patients will be at risk of developing community-acquired catheterassociated S. aureus bacteremia. Advances such as TEE and the Duke criteria should be frequently employed to assist in the early and accurate identification of S. aureus IE and therefore improve survival. The authors thank Dr. J. M. Mylotte for his helpful review of the manuscript. References 1. Espersen F, Frimodt-Moller N. Staphylococcus aureus endocarditis. A review of 119 cases. Arch Intern Med 1986; 146:1118 – 21. 2. Sanabria TJ, Alpert JS, Goldberg R, Pape LA, Cheeseman SH. Increasing frequency of staphylococcal infective endocarditis. Experience at a university hospital, 1981 through 1988. Arch Intern Med 1990; 150: 1305 – 9. 3. Watanakunakorn C. Staphylococcus aureus endocarditis at a community teaching hospital, 1980 to 1991. An analysis of 106 cases. Arch Intern Med 1994; 154:2330 – 5. 4. Banerjee SN, Emori TG, Culver DH, et al. Secular trends in nosocomial primary bloodstream infections in the United States, 1980 – 1989. National Nosocomial Infections Surveillance System. Am J Med 1991; 91: 86S – 9S. 5. Mylotte JM, White D, McDermott C, Hodan C. Nosocomial bloodstream infection at a veterans hospital: 1979 to 1987. Infect Control Hosp Epidemiol 1989; 10:455 – 64. 6. Steinberg JP, Clark CC, Hackman BO. Nosocomial and communityacquired Staphylococcus aureus bacteremias from 1980 to 1993: impact of intravascular devices and methicillin resistance. Clin Infect Dis 1996; 23:255 – 9. 7. Watanakunakorn C, Tan JS, Phair JP. Some salient features of Staphylococcus aureus endocarditis. Am J Med 1973; 54:473 – 81. 8. Watanakunakorn C, Tan JS. Diagnostic difficulties of staphylococcal endocarditis in geriatric patients. Geriatrics 1973; 28:168 – 73. 9. Watanakunakorn C, Burkert T. Infective endocarditis at a large community teaching hospital, 1980 – 1990. A review of 210 episodes. Medicine (Baltimore) 1993; 72:90 – 102. 10. Bayer AS, Ward JI, Ginzton LE, Shapiro SM. Evaluation of new clinical criteria for the diagnosis of infective endocarditis. Am J Med 1994; 96: 211 – 9. / 9c5e$$ja11 12-30-98 16:01:10 11. Durack DT, Lukes AS, Bright DK. New criteria for diagnosis of infective endocarditis: utilization of specific echocardiographic findings. Am J Med 1994; 96:200 – 9. 12. Daniel WG, Mugge A. Transesophageal echocardiography. N Engl J Med 1995; 332:1268 – 79. 13. Shapiro SM, Young E, De Guzman S, et al. Transesophageal echocardiography in diagnosis of infective endocarditis. Chest 1994; 105:377 – 82. 14. Fowler VG Jr, Li J, Corey GR, et al. Role of echocardiography in the evaluation of patients with Staphylococcus aureus bacteremia: experience in 103 patients. J Am Coll Cardiol 1997; 30:1072 – 8. 15. Libman H, Arbeit RD. Complications associated with Staphylococcus aureus bacteremia. Arch Intern Med 1984; 144:541 – 5. 16. Kong LK, Fowler VG Jr, Corey GR, Gottlieb G, McClelland RS, Harrell L. Recurrent Staphylococcus aureus bacteremia: pulse field gel electrophoresis findings in 28 consecutive patients [abstract no LM-11]. In: Program and abstracts of the 37th Interscience Conference on Antimicrobial Agents and Chemotherapy (Toronto). Washington, DC: American Society for Microbiology, 1997:366. 17. Tenover FC, Arbeit R, Goering RV, et al. Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: criteria for bacterial strain typing. J Clin Microbiol 1995; 33:2233 – 9. 18. Terpenning MS, Buggy BP, Kauffman CA. Hospital-acquired infective endocarditis. Arch Intern Med 1988; 148:1601 – 3. 19. Frimodt-Moller N, Esperson F, Skinhoj P, Rosdahl VT. Changing epidemiology of Staphylococcus aureus bacteremia during 1960 – 1990 in Denmark [abstract no 1441]. In: Program and abstracts of the 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy (New Orleans). Washington, DC: American Society for Microbiology, 1993: 383. 20. Fernández-Guerrero ML, Verdejo C, Azofra J, de Górgolas M. Hospitalacquired infectious endocarditis not associated with cardiac surgery: an emerging problem. Clin Infect Dis 1995; 20:16 – 23. 21. Saiman L, Prince A, Gersony WM. Pediatric infective endocarditis in the modern era. J Pediatr 1993; 122:847 – 53. 22. Fang G, Keys TF, Gentry LO, et al. Prosthetic valve endocarditis resulting from nosocomial bacteremia. A prospective, multicenter study. Ann Intern Med 1993; 119:560 – 7. 23. Watanakunakorn C, Baird IM. Staphylococcus aureus bacteremia and endocarditis associated with a removable infected intravenous device. Am J Med 1977; 63:253 – 6. 24. Watanakunakorn C. Infective endocarditis as a result of medical progress. Am J Med 1978; 64:917 – 9. 25. Chen SC, Dwyer DE, Sorrell TC. A comparison of hospital and community-acquired infective endocarditis. Am J Cardiol 1992; 70:1449 – 52. 26. Musher DM, Lamm N, Darouiche RO, Young EJ, Hamill RJ, Landon GC. The current spectrum of Staphylococcus aureus infection in a tertiary care hospital. Medicine (Baltimore) 1994; 73:186 – 208. 27. Mylotte JM, McDermott C, Spooner JA. Prospective study of 114 consecutive episodes of Staphylococcus aureus bacteremia. Rev Infect Dis 1987; 9:891 – 907. 28. Nolan CM, Beaty HN. Staphylococcus aureus bacteremia. Current clinical patterns. Am J Med 1976; 60:495 – 500. 29. Graham DR, Keldermans MM, Klemm LW, Semenza NJ, Shafer ML. Infectious complications among patients receiving home intravenous therapy with peripheral, central, or peripherally placed central venous catheters. Am J Med 1991; 91:95S – 100S. 30. Steinberg JP, Clark CC, Hackman BO. Epidemiology of community acquired intravascular device-related bloodstream infections [abstract]. Clin Infect Dis 1997; 25:479. 31. De Castro S, Magni G, Beni S, et al. Role of transthoracic and transesophageal echocardiography in predicting embolic events in patients with active infective endocarditis involving native cardiac valves. Am J Cardiol 1997; 80:1030 – 4. 32. Lutas EM, Roberts RB, Devereux RB, Prieto LM. Relation between the presence of echocardiographic vegetations and the complication rate in infective endocarditis. Am Heart J 1986; 112:107 – 13. cida UC: CID Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 Acknowledgment 113 114 Fowler et al. 33. Jaffe WM, Morgan DE, Pearlman AS, Otto CM. Infective endocarditis, 1983 – 1988: echocardiographic findings and factors influencing morbidity and mortality. J Am Coll Cardiol 1990; 15:1227 – 33. 34. Mugge A, Daniel WG, Frank G, Lichtlen PR. Echocardiography in infective endocarditis: reassessment of the prognostic implications of vegetation size determined by the transthoracic and transesophageal approach. J Am Coll Cardiol 1989; 14:631 – 8. 35. Mugge A. Echocardiographic detection of cardiac valve vegetations and prognostic implications. Infect Dis Clin North Am 1993; 7:877 – 98. 36. Sanfilippo AJ, Picard MH, Newell JB, et al. Echocardiographic assessment of patients with infectious endocarditis: prediction of risk for complications. J Am Coll Cardiol 1991; 18:1191 – 9. 37. Tischler MD, Vaitkus PT. The ability of vegetation size on echocardiography to predict clinical complications: a meta-analysis. J Am Soc Echocardiogr 1997; 10:562 – 8. 38. Pujol M, Pena C, Pallares R, Ayats J, Ariza J, Gudiol F. Risk factors for nosocomial bacteremia due to methicillin-resistant Staphylococcus aureus. Eur J Clin Microbiol Infect Dis 1994; 13:96 – 102. CID 1999;28 (January) 39. Romero-Vivas J, Rubio M, Fernandez C, Picazo JJ. Mortality associated with nosocomial bacteremia due to methicillin-resistant Staphylococcus aureus. Clin Infect Dis 1995; 21:1417 – 23. 40. Harbarth S, Rutschmann O, Sudre P, Pittet D. Impact of methicillin resistance on the outcome of patients with bacteremia caused by Staphylococcus aureus. Arch Intern Med 1998; 158:182 – 9. 41. Hartstein AI, Mulligan ME, Morthland VH, Kwok RY. Recurrent Staphylococcus aureus bacteremia. J Clin Microbiol 1992; 30: 670 – 4. 42. Levine DP, Fromm BS, Reddy BR. Slow response to vancomycin or vancomycin plus rifampin in methicillin-resistant Staphylococcus aureus endocarditis. Ann Intern Med 1991; 115:674 – 80. 43. Mortara LA, Bayer AS. Staphylococcus aureus bacteremia and endocarditis. New diagnostic and therapeutic concepts. Infect Dis Clin North Am 1993; 7:53 – 68. 44. Mylotte JM, Aeschlimann JR, Rotella DL. Staphylococcus aureus bacteremia: factors predicting hospital mortality. Infect Control Hosp Epidemiol 1996; 17:165 – 8. Downloaded from cid.oxfordjournals.org by guest on July 13, 2011 / 9c5e$$ja11 12-30-98 16:01:10 cida UC: CID