Methicillin-resistant Staphylococcus aureus bloodstream infection: risk factors and clinical outcome in non-intensive-care units

Introduction: Methicillin-resistant Staphylococcus aureus (MRSA) is spread out in hospitals across different regions of the world and is regarded as the major agent of nosocomial infections, causing infections such as skin and soft tissue pneumonia and sepsis. The aim of this study was to identify risk factors for methicillin-resistance in Staphylococcus aureus bloodstream infection (BSI) and the predictive factors for death. Methods: A retrospective cohort of fifty-one patients presenting bacteraemia due to S. aureus between September 2006 and September 2008 was analysed. Staphylococcu aureus samples were obtained from blood cultures performed by clinical hospital microbiology laboratory from the Uberlândia Federal University. Methicillinresistance was determined by growth on oxacillin screen agar and antimicrobial susceptibility by means of the disk diffusion method. Results: We found similar numbers of MRSA (56.8%) and methicillin-susceptible Staphylococcus aureus (MSSA) (43.2%) infections, and the overall hospital mortality ratio was 47%, predominantly in MRSA group (70.8% vs. 29.2%) (p=0.05). Age (p=0.02) was significantly higher in MRSA patients as also was the use of central venous catheter (p=0.02). The use of two or more antimicrobial agents (p=0.03) and the length of hospital stay prior to bacteraemia superior to seven days (p=0.006) were associated with mortality. High odds ratio value was observed in cardiopathy as comorbidity. Conclusions: Despite several risk factors associated with MRSA and MSSA infection, the use of two or more antimicrobial agents was the unique independent variable associated with mortality.

A broad variety of infections, ranging from minor infections of the skin to severe infections as bloodstream infections, can be caused by Staphylococcus aureus. Methicillin-resistant Staphylococcus aureus (MRSA) is an endemic nosocomial pathogen, but its geographic spread in the community has been documented 1 . Some previous studies of patients with MRSA bloodstream infection (BSI) have reported higher mortality rates, increased morbidity, and longer hospital length of stay than those with methicillin-susceptible Staphylococcus aureus BSI 2,3 . In the UK, MRSA infection is cited with increasing frequency as causing or contributing to death 4,5 . Risk factors for methicillin-resistance in these infections have been extensively described, but the studies vary among institutions 6,7 .
Methicillin-resistance among S. aureus remains an important problem in Latin America hospitals 8 , but rates vary significantly from hospital to hospital. In Brazil, the Antimicrobial Surveillance Program (SENTRY) 9  The objectives of this study were to identify institution-specific risk factors for methicillinresistance in S. aureus BSI to determine the predictive factors for death and assess the impact of methicillinresistance on mortality in patients in non-intensivecare units of a Brazilian University Hospital. The study was conducted in a 500-bed teaching hospital that provides tertiary care for Uberlândia and surroundings. The investigation was designed as a retrospective cohort of patients presenting bacteraemia due to S. aureus. Clinical and epidemiological data were collected from the hospital patient records.

Naves KSC et al -MRSA bloodstream infection
Bloodstream infections were classified as primary when they were not related to any other focus of infection. BSIs were considered to be secondary when they were clinically related to infection in another site than the vascular system 12 . Previous antimicrobial therapy was defined as the use of any antibiotic within 30 days prior to bacteraemia 13 . Antimicrobial therapy was considered to be adequate if the drug used within the first 48h after blood culture collection had in vitro activity against the isolated S. aureus strain 13 . Bacteraemia-Between September 2007 and September 2008, 134 S. aureus BSIs were identified. Methicillin-resistance rate was 59.7% (80/134). Among these cases, 83 were excluded from the study because of incomplete or missing associated mortality was characterized by the patient's death during the bacteremic episode and/or 30 days after hospitalization period without any other explanation 14 .
Hemocultures were performed by inoculating 5-10mL of blood into a flask of the automatic commercial system Bactec/ Alert® (Vitek system). Positive cultures were further sub-cultured in Müeller-Hinton Agar (Isofar Ltda, Brazil) supplemented with 5% of human blood and incubated for 24-48h at 35 ± 2°C in the hospital microbiology laboratory. Methicillin-resistance was determined by growth on oxacillin screen agar and antimicrobial suscetibility by means of the disk diffusion method with disks containing antimicrobial agents.
Univariate analysis was used to compare variables for the outcome groups of interest. Continuous variables were compared using Student's t test for normally distributed variables. The χ 2 statistic or Fisher´s exact test was used to compare categorical variables. Multivariate analysis was performed using multiple logistic regression with stepwise approach for entering new terms into the BioEstat5.0 model with 0.2 as the limit for their acceptance or removal. All p values lower than 0.2 were considered significant.

Ethical considerations
The present study was approved by the Ethics Committee of Federal University of Uberlândia (UFU) (014/06). data. A total of 51 episodes in 39 patients were then reviewed. Ten patients presented multiple episodes of MRSA BSI, and two presented multiple episodes of MSSA BSI. Thirty-one infections occurred in male patients, and twenty occurred in female patients (Table 1). Table 1 shows clinical and demographic characteristics of patients with MRSA and MSSA BSI. Methicillin-resistant Staphylococcus aureus bloodstream infection occurred in older patients, after a longer time following admission when compared with MSSA BSI. All Methicillin-resistant Staphylococcus aureus bloodstream infection were classified as nosocomial. The use of more than two antimicrobial agents, comorbidity, and length of hospital stay prior to bacteraemia longer than seven days were more frequent in MRSA group. Presence of intravascular device was significantly higher in MRSA group than in MSSA group. The sources of S. aureus BSI are also listed in Table 1, and in both, MRSA and MSSA, primary BSIs were the most common sources than secondary ones, and 22 (75.9%) of whom with MRSA episodes were associated with central venous catheter.  The overall mortality rate was 47%. Univariate analysis for potential prognostic death factors associated with S. aureus bacteraemia is presented in Table 2. The presence of MRSA (p=0.057), clinical status, represented by patient´s unit (p=0.046), and length of hospital stay prior to bacteraemia higher than seven days (p=0.006) were significantly associated with death. No difference was found regarding sex, age, severity of underlying disease, nosocomial origin of the infection, or presence of intravascular device among the patients who died compared with the patients who survived.
In multivariate analysis (Table 3), the use of more than two antibiotics was independently associated with mortality by S. aureus (OR=8.65; 95% CI=1.92 to 39.07; p=0.05). ORs observed for long duration of hospital stay (OR=4.7) and cardiopathy (OR=5.85) were both high in death patients group.
Hospital-acquired bloodstream infections caused by S. aureus, mainly those due to methicillin-resistant S. aureus (MRSA), are associated with significant mortality and morbidity 15 , adding considerable costs to hospital care 16 . In Brazil, S. aureus was the most common cause of bloodstream infection (20.2%), and resistance to methicillin was observed in 31% of S. aureus isolates in the Brazilian hospital participating in the SENTRY Antimicrobial Surveillance Programme 9 . However, MRSA rates may vary greatly among hospitals (55.9-73%) 8,13 . In our hospital, bacteraemia rates caused by MRSA were both high in both critical (63.7%) and non-critical units (62.5%) 11 . In our study, we observed a similar rate (56.8%) of BSI due to MRSA in clinical, burned, and oncology wards, and almost all BSIs (96%) had nosocomial origin.
The emergence of MRSA is largely due to the dissemination of clonal strains between patients favored for poor infection control politics and antimicrobial pressure 17 , but this correlation has been difficult to establish due to the high number of variables involved 18 . Risk factors for methicillin resistance reported in literature also vary among institutions and patients. The major independent risk factors include: advanced age 19,20 , residence in a nursing home 7,21 , long duration of hospitalization 4 , prior antibiotic exposure 6,22 , insulin-requiring diabetes 23 , intravascular devices 3,24 , presence of decubitus ulcers or pneumonia as source of BSI 6,15,24 , inadequacy of antimicrobial therapy, and severity of clinical status 13 .
In the present study, based on univariate analysis, age and presence of central venous catheter (CVC) were risk factors for MRSA bacteraemia as also reported previously 10,13 . Most (72.4%) of MRSA bacteraemia were classified as primary, and CVC was probably the foci of infection as discussed by Das et al. 24 . A high proportion of these infections were detected in clinical ward (34.5%), while most studies reported higher MRSA infections in critical-care units 24,25 .
Other analyzed variables such as length of hospital stay prior to bacteraemia (OR=3.19), cardiopathy as comorbidity (OR=2.85), and more than two antimicrobial drugs usage (OR=2.74), despite having no significance, presented high OR values when compared with MRSA and MSSA infections. Several studies also have demonstrated that longer hospitalization before the onset of bacteraemia 25 and antimicrobial use in greater frequency 10 were risk factors for MRSA infection. None of these factors -age, central vascular catheter presence or even the use of more than two antimicrobial drugs, cardiopathy, or longer hospital stay prior to bacteraemia -were independently prognostic factors for MRSA infection in our study.
Comparisons of mortality between patients with MRSA and MSSA have been contradictory in the literature 15 . Several studies including a meta-analysis have demonstrated an increase in mortality among patients with MRSA bacteraemia versus MSSA bacteraemia 3,4,26,27 . In this study, a higher crude mortality of 58.6% vs 24.1% (OR=3.036; p=0.05) was observed in the MRSA group than the MSSA group, respectively, a higher rate than those reported in United Kingdom and United States hospitals 15,27 .
Traditional risk factors for mortality include older age 10,15,28 , inadequate treatment 13,29 , and severity of comorbidity 25 . Despite the fact that these variables were not significant in our study, we attribute an association between these risk factors and mortality due to the high OR values of 2.80, 2.04, and 3.85, respectively. Intensive-care unit admission before bacteraemia is also a predictor for MRSA-BSI mortality 15,24 . Most of the deaths due to staphylococcal BSI occurred in the clinical ward (45%; p=0.046), with only 4% of patients carried out in the ICU.
Length of hospitalization is a measure of morbidity 15 , as well as invasive device use such as CVC 30 . In the univariate analysis, death was associated with an increase in length of hospital stay for more than seven days (p<0.05) and the use of more than two antimicrobial drugs (p=0.034), but only the former was independently significant (OR=8.65; p=0.005) in multiple regression analysis. In multivariable analysis for the length of hospital stay prior to bacteraemia superior to seven days (OR=4.7) and cardiopathy as comorbidity (OR=5.8), we observed high OR values. These findings coincide with previous reports demonstrating that factors for mortality include: longer of hospital stay before bacteraemia 25 and comorbidities as cardiopathy 31 . Neverthless, in our study, patients with bacteraemia were usually confined in non-critical units due to lack of sufficient beds in the ICU where 33.3% of patients received inadequate empirical antimicrobial therapy that resulted in higher rate (41.6%) of hospital mortality. Eighty percent of the patients who died using an inadequate treatment were from MRSA BSI group.
One of the limitations of the study was that it was done retrospectively and performed at a single hospital.
In addition, the small sample size of the study limited the detection of statistically significant differences. We also did not use the Apache II or other scores to assess severity of illness prognostic indicator on the hospital mortality.
In conclusion, high hospital mortality was observed even in cases of patients who were admitted in non-critical-care units. We found several risk factors associated to MRSA and MSSA bacteraemia and high mortality level of S. aureus bacteraemia. However, the use of two or more antimicrobial agents was the unique independent variable for death. This can be partially justified by high frequency of inadequate empirical therapy observed in our study. In spite of the limitations, mainly the small number of patients, the findings suggest an imperative for hospitals to review their antimicrobial policies.