This prospective cohort study was conducted between August 2008 and June 2011 at the Asan Medical Center, a-2,700-bed hospital in Seoul that provides primary and tertiary care for patients from throughout Korea. By reviewing daily computerized reports on blood cultures, all patients aged ≥ 18 years with a MRSA-positive blood culture were identified and enrolled in the study. Patients were excluded if clinical isolates of MRSA were not available for microbiologic tests, or if they had polymicrobial bacteremia or clinically insignificant MRSAB. Insignificant bacteremia was defined as satisfying all of the following conditions: isolation from only one blood culture, no clinical finding consistent with bacteremia, and no anti-staphylococcal treatment. The study consisted of two sets of analyses: early mortality versus non-early mortality, and severe sepsis or septic shock versus non-severe sepsis or septic shock (Fig. 1). Early mortality was defined as death within 2 days of blood culture, since it takes about 48 hours to react to target concentrations of vancomycin as the drug of choice [12,14]. Late mortality was defined as death within 28 days of blood culture. The other patients were classified into the survival group. Severe sepsis or septic shock was defined according to International Sepsis Definitions Conference 2001 criteria [15]. The study was approved by the Institutional Review Board of Asan Medical Center (IRB No. 2008-0274).

The following were recorded: demographic characteristics of the patients, days spent in the hospital before detection of MRSAB, underlying diseases or conditions, severity of illness, presence or absence of foreign material such as an intravenous catheter or prosthetic device, site of infection (catheter-related bloodstream infection, arteriovenous fistula infection, pneumonia, skin and soft tissue infection, bone and joint infection, urinary tract infection, and/or infective endocarditis), susceptibility to antibiotics, antibiotics used to treat MRSA (vancomycin, teicoplanin, tigecycline, and linezolid), and inhospital mortality. If the site of infection could not be determined, it was considered to be unknown. Hospitalacquired bacteremia was defined as a positive blood culture obtained from a patient who had been hospitalized for 48 hours or longer. Community-acquired bacteremia and healthcare-associated bacteremia were defined as described by Friedman et al. [16]. To estimate the severity of the illness, the sepsis grade [17] and Pitt bacteremia score [18] were obtained. To estimate the severity of comorbidities, the Charlson comorbidity score (CCS) was calculated, as previously described [19]. When blood culture was performed, disease severity was classified according to the McCabe and Jackson classification system [20]. Eradicable foci included surgically removable infections, drainable abscesses, and indwelling foreign bodies such as peripheral and central intravenous catheters [21]. Early removal of eradicable foci was defined as removal within 2 days of blood culture. Patients with non-eradicable foci and whose eradicable foci were not removed early were categorized as having non-eradicated foci. If vancomycin, teicoplanin, tigecycline, or linezolid was prescribed, the department of infectious diseases was automatically consulted and the appropriateness of the treatment was reviewed. Vancomycin dose was adjusted according to the report published by the American Society of Health-System Pharmacists [22].

All isolates were confirmed as MRSA by polymerase chain reaction (PCR) for the mecA gene, and tested for antimicrobial susceptibility by standard techniques according to Clinical and Laboratory Standards Institute guidelines [23]. Staphylococcal cassette chromosome (SCC) mec type MRSA was identified [24]. The presence of bacterial virulence factors, including adhesins and toxins, was examined by multiplex PCR [25,26], and multi-locus sequence typing was also performed [27]. To examine agr dysfunction, the extent of δ-hemolysin production was measured by streaking each MRSA isolate next to a β-hemolysin disk (Remel, Lenexa, KS, USA) [28]. The vancomycin minimum inhibitory concentrations (MICs) were determined using the vancomycin E-test (AB Biodisk, Piscataway, NJ, USA) on Mueller-Hinton agar. Heteroresistant vancomycin-intermediate S. aureus was identified by a modified population analysis profile-area under curve ratio method [29].

Statistical analyses were performed using SPSS version 14.0 (SPSS Inc., Chicago, IL, USA). Continuous variables were compared using the Mann-Whitney U test or Student t test. Categorical variables were compared using the Pearson chi-square test or Fisher exact test. Binary logistic regression was used to identify variables significantly associated with early mortality before detection of MRSAB by blood culture. Owing to the small size of the early-mortality group, the number of variables had to be minimized. Variables found to be statistically significant at the 5% level in the univariate analysis were included in the multivariate analysis. Underlying disease was introduced as a comorbidity variable only, and CCS and McCabe and Jackson classification were not included. Severe sepsis or septic shock was used to define severity of illness. Early removal of eradicable foci was also not included because of its low frequency. Anti-MRSA therapy was also excluded from the multivariate analysis because the frequency of non-treated patients was less than two in the late mortality and survival groups. Variables found to be statistically significant at the 10% level in the univariate analysis were included in the multivariate analysis of risk factors for severe sepsis or septic shock. If no patient fit the criteria for any group, that variable was excluded from the multivariable analysis. There was no multicollinearity among the variables included in the multivariate analysis. All significance tests were two-tailed, and a p < 0.05 was considered significant.

During the study period, 577 patients with MRSAB aged ≥ 18 years were identified. Of these patients, 192 were excluded for the following reasons: clinical isolates were not collected (117 patients), polymicrobial bacteremia was found (66 patients), or the MRSAB was clinically insignificant (nine patients). Of the 385 enrolled patients, 158 (41.0%) died; 25 patients (6.5%) died early and 50 patients (13%) died within 28 days. Severe sepsis or septic shock occurred in 124 patients (32.2%) with MRSA bacteremia. Ultimately, of 385 enrolled patients, 25 (6.5%), 50 (13.0%), and 310 patients (80.5%) were assigned to the early mortality, late mortality, and survival groups (Fig. 1) Also, 124 (32.2%) and 261 patients (67.8%) were assigned to the severe sepsis or septic shock group and non-severe sepsis or septic shock group (Fig. 1).

The clinical and microbiological characteristics of patients in the early mortality, late mortality, and survival groups are shown in Table 1. According to the severity index, the patients with early mortality had more severe disease than did those in the late mortality or survival groups. Early removal of eradicable foci in early mortality patients was less frequent than in the late mortality or survival groups. Of the patients treated with antibiotics, 221 (92.1%), 14 (3.5%), three (0.8%), and one (0.3%) were treated with vancomycin, teicoplanin, linezolid, and tigecycline, respectively. Some of the patients with early mortality did not receive anti-MRSA therapy within 24 hours.

ST5-SCCmecII-agrI (254 patients, 66.0%) was the most common MRSA strain, followed by ST72-SCCmecIV-agrII (87 patients, 22.6%) and ST239-SCCmecII-agrI (16 patients, 4.2%). The ST72-SCCmecIV-agrII strain was less often seen in the late mortality group than in the early mortality and survival groups, although this difference was not statistically significant in the multivariate analysis. The distribution of vancomycin MICs was as follows: in 121 cases (31.4%) it was ≤ 1.0 μg/mL, in 177 (45.9%) it was ≤ 1.5 μg/mL, in 82 (21.3%) it was ≤ 2.0 μg/mL, and in 5 (1.8%) it was > 2 μg/mL. The following virulent genes were not found in any isolates: seb, sed, see, seh, sej, eta, etb, lukM, hlg, and edin. Regardless of survival, the following genetic factors were very rare: Panton-Valentine leukocidin, map/eap, sea, sek, sep, and seq. Virulent genes such as fnbA, clfA, clfB, can, and icaA were found in all isolates, and fnbB, bbp, ebps, sdrD, sdrE, seg, sei, sem, sen, seo, lukDE, lukE, hla, hld, and hlg2 were found in most isolates. The frequencies of virulent genes such as sdrC, sec, sel, and TSST-1 are given in Table 1. Among the virulent genes, sel was less frequent in patients with early mortality than in those with late mortality.

To compare the clinical and microbiologic factors seen in patients with early mortality versus those with late mortality, occurrence of severe sepsis or septic shock, presence of the ST72-SCCmecIV-agrI strain, and sel as a virulent gene were included in the multivariate analysis (Table 2). Compared with patients with late mortality, severe sepsis or septic shock (adjusted odds ratio [aOR], 4.49; 95% confidence interval [CI], 1.28 to 15.75) was the only statistically significant independent risk factor in patients with early mortality. To compare the clinical and microbiologic factors of patients with early mortality and those who survived, the following variables were included in the multivariate analysis: liver cirrhosis, severe sepsis or septic shock, and anti-MRSA therapy on day of blood culture (Table 2). Compared with patients who survived, liver cirrhosis (aOR, 3.79; 95% CI, 1.38 to 23.86) and severe sepsis or septic shock (aOR, 10.98; 95% CI, 3.82 to 31.52) were statistically significant independent risk factors in patients with early mortality. When rapidly or ultimately fatal McCabe and Jackson classification was added to the multivariate analysis and severe sepsis or septic shock was replaced with Pitt bacteremia score ≥ 3, both were also statistically significant (data not shown). Compared with patients who survived, rapidly or ultimately fatal McCabe and Jackson classification (aOR, 4.61; 95% CI, 2.37 to 9.00) and severe sepsis or septic shock (aOR, 2.95; 95% CI, 1.54 to 5.63) were statistically significant independent risk factors in patients with late mortality.

The clinical and microbiologic characteristics of patients in the severe sepsis or septic shock and non-severe sepsis or septic shock groups are shown in Table 3. In the multivariate analysis, rapidly or ultimately fatal McCabe and Jackson classification as severity of comorbidity (aOR, 1.94; 95% CI, 1.25 to 3.02) and pneumonia as site of infection (aOR, 2.04; 95% CI, 1.03 to 4.02) were significantly associated with severe sepsis or septic shock. Fewer patients with isolates with vancomycin MICs ≥ 1.5 μg/mL experienced severe sepsis or septic shock (aOR, 0.53; 95% CI, 0.33 to 0.84).