A retrospective cohort study was conducted to investigate risk factors associated with mortality in S. marcescens bacteremia at Severance Hospital, a 2000-bed, tertiary-care teaching hospital, Seoul, South Korea. Inclusion criteria for this study included patients with 18 years of age or older, identified as having one or more blood cultures positive for S. marcescens between January 2006 and December 2012. For subjects reporting more than one episode of S. marcescens bacteremia, only the first episode was accepted. Demographic and clinical variables were evaluated using microbiological laboratory records and clinical data gained from electronic medical records; these included age, gender, length of hospital stay, underlying diseases, predisposing conditions, portal of entry, appropriateness of antimicrobial agents, appropriateness of definitive therapy, results of antimicrobial susceptibility testing, laboratory data at the time of bacteremia onset, Sequential Organ Failure Assessment score (SOFA),14 and 28-day all-cause mortality. This study was approved by our Institutional Review Board. Informed consent was exempt from our local ethics committee because this study was concerned to cause minimal harm on persons.

Significant S. marcescens bacteremia was defined as S. marcescens isolates cultured from one or more blood samples obtained from a patient, combined with clinical symptoms compatible with systemic inflammatory response syndrome.15 Hospital-acquired bacteremia was defined as a positive blood culture taken from a patient no sooner than 48 h after hospital admission, whereas healthcare-associated bacteremia was defined as a positive blood culture taken from a patient receiving home and/or ambulatory intravenous therapy, hemodialysis, wound care, chemotherapy, or nursing care, or who had attended a hospital clinic within the last 30 days; patients hospitalized in an acute care hospital for ≥2 days within the last 90 days; or those living in a nursing home or long-term care facility.16 The primary site of infection was presumed to be the source of bacteremia if S. marcescens was identified from any culture specimens at the time of bacteremia onset; if S. marcescens was not identified from any culture other than the blood, the source was presumed to be primary bacteremia. Polymicrobial bacteremia was defined as bacteremia where more than one organism were isolated from the same blood culture specimen. Septic shock was defined as sepsis-induced organ hypo-perfusion, combined with either a systolic blood pressure <90 mm Hg or <40 mm Hg less than baseline, or a mean arterial pressure <65 mm Hg after a fluid resuscitation, eventually leading to require the vasopressor use.17 Underlying chronic diseases included hemato-oncological disease, chronic renal disease, chronic liver disease, chronic lung disease, cardiovascular disease, and cerebrovascular disease, as defined by the International Classification of Disease, 10th Revision.18 Prior use of antimicrobial agent was defined as receipt for at least 48 h within 1 month prior to the bacteremic episode. Appropriateness of initial empirical antimicrobial agents was defined as the use of at least one in vitro susceptible antimicrobial agent within 24 h of positive blood culture before the susceptibility was known.19 Definitive therapy was defined as antibiotic therapy given properly according to the results of final blood culture.20 Hypoalbuminemia was defined as a serum albumin of less than 3.0 g/dL at the time of bacteremia.21,22 Twenty-eight-day all-cause mortality was investigated to confirm the primary outcome.

Clinical isolates were evaluated using either conventional techniques or the ATB 32 GN system (bioMérieux, Marcy l' Etoile, France). Antimicrobial susceptibility testing was performed by microbiology laboratory staff using the disk-diffusion method or a VITEK-2 N131 card (bioMérieux, Hazelwood, MO, USA). Results were interpreted using the guidelines set forth by the Clinical and Laboratory Standards Institute.23

Student's t-test was used to compare continuous variables; and categorical variables were analyzed using either a χ² or Fisher's exact test as appropriate. Nonparametric variables were analyzed using the Mann-Whitney U test. Univariate and multivariate analyses to evaluate independent risk factors for all-cause mortality in S. marcescens bacteremia were performed through the logistic regression models. Statistical analyses were performed using the SPSS software, version 20 (SPSS Inc., Chicago, IL, USA). p-values <0.05 were considered to indicate statistical significance; all values reported are for two-tailed analyses.

A total of 98 episodes of S. marcescens bacteremia were identified between January 2006 and December 2012. The annual distribution of S. marcescens bacteremia is shown in Fig. 1. To confirm the outbreak, the trend was investigated and stratified by ward and period, but there were no outbreaks.

Table 1 shows the demographics, clinical characteristics, and underlying conditions of all patients with S. marcescens bacteremia. Sixty patients (61.2%) were male, and the ages of patients ranged from 26 to 92 years (median age, 63.5 years). Of 98 bacteremic episodes, hospital-acquired bacteremia accounted for 73.5% of the episodes (72 patients), whereas healthcare-associated bacteremia accounted for 18.4% of the episodes (18 patients). The remaining 8 episodes (8.1%) were defined as community-acquired bacteremia. The most common underlying condition was malignancy (46.9%), which included conditions such as solid organ and hematologic malignancy; and diabetes mellitus (30.6%) were also common. The most frequent portal of entry was the lower respiratory tract (50%). Other portals of entry included the urinary tract (20.4%), abdomen (11.2%), skin and soft tissue (10.2%), and intra-venous catheter (5.1%). Half of the patients had central venous catheters (CVCs), including chemo-ports, and urinary catheters. 56.1% of the patients showed bacteremia during the intensive care unit (ICU) stay. Among them, the fatal group showed a greater incidence of bacteremia during the ICU stay (77.3%) than the non-fatal group (50%) (p=0.020). 46.9% of the episodes (46 patients) showed the 3rd cephalosporin resistance, but there was no statistically significant difference between the fatal and non-fatal group (p=0.744).

The antimicrobial susceptibilities of S. marcescens clinical isolates are characterized in Fig. 2. The majority of isolates were susceptible to ertapenem (100%), meropenem (99%), imipenem (93.4%), cefepime (87.8%), isepacin (81.8%), and ceftazidime (76.5%). A few isolates exhibited susceptibility to amoxicillin/clavulanic acid (1.1%), ampicillin (2.1%), and ampicillin/sulbactam (1.7%).

The 28-day all-cause mortality was 22.4% (22/98). Univariate analysis revealed significant differences in the number of important clinical covariates including the onset of bacteremia during the ICU stay (p=0.020), serum albumin level (p=0.011), serum C-reactive protein level (p=0.041), presence of indwelling urinary catheter (p=0.023), and SOFA score at the onset of bacteremia (p<0.001) between patients in the fatal and non-fatal groups (Table 1). In a multivariate analysis, lower serum albumin level [adjusted odds ratio (OR) 0.206, 95% confidential interval (CI) 0.044-0.960, p=0.040], and elevated SOFA score (adjusted OR 1.474, 95% CI 1.200-1.810, p<0.001) were all found to be independent risk factors for mortality in patients with S. marcescens bacteremia (Table 2).