Background
Bloodstream infections (BSI) are a common cause of mortality despite advances in antimicrobial agents and sepsis management [
1‐
3]. Although short-term mortality of BSI has been estimated from 12 to 34%, the range of mortality could be highly variable according to the acquisition site (whether community or hospital-acquired BSI), a specific population, pathogens, drug resistance, and geographical region [
1‐
8]. For example, hospital-acquired (HA)-BSI usually show higher crude mortality than community-acquired (CA)-BSI [
2]. Although
Escherichia coli is the most common pathogen causing BSIs worldwide, higher mortality has been reported with bacteremia due to other Gram-negative bacteremia other than
E. coli [
3,
9,
10]. Therefore, the epidemiology of pathogens contributing to mortality may differ from the actual bacteremia incidence of each specific pathogen.
In the Republic of Korea,
E. coli,
Staphylococcus aureus, and
Klebsiella pneumoniae were the three most common pathogens of true bloodstream infection; In reports after the 2000s, drug resistance of Gram-negative bacilli (GNB) has been a more predominant problem than that of Gram-positive cocci (GPC) [
10,
11]. Carbapenem resistance has been frequently observed with
Acinetobacter baumannii, and reported to be more than 90% in a surveillance study in 2017 [
12]. Although there are data on the proportion of each pathogen isolated from BSIs and their drug resistance, data on pathogens causing mortality and the extent of their contribution are rarely reported.
We believed that knowing the epidemiology of BSIs contributing to mortality and their susceptibility patterns could help better understand BSIs and set priority for infection control and institutional or national antibiotic policy. Therefore, this retrospective study was conducted to investigate pathogens related to mortality and to compare species and their characteristics according to where the BSI was acquired.
Methods
Study design and definition
Adult patients (≥ 18 years old) admitted to one of two teaching hospital between January 2019 and December 2019 were enrolled. The participating institutions were Samsung Changwon Hospital and Inje University Busan Paik Hospital located in Changwon and Busan, Korea, respectively. Electronic medical records were reviewed to collect data of laboratory results and clinical characteristics of patients.
We defined BSIs contributing mortality as a BSI within the 2-weeks before death, as prior studies have used a 14-day mortality to represent short-term mortality from BSI [
8]. The results of blood cultures performed on the study population during the 2-weeks before death were collected. When bacteria or fungi were isolated from more than two separate blood samples or from a single blood sample in patients with explainable clinical symptoms and focuses other than skin commensals, they were considered as true BSIs [
13]. We only considered isolated pathogens from true BSI as the pathogens contributing to mortality.
Patients and their isolated pathogens were classified by their acquisition sites: Community-acquired (CA)-, healthcare-associated (HCA)- and HA-BSI. HCA infection was defined as infected patients with at least one of the four elements: (1) Parenteral treatment within 30 days, (2) outpatient chemotherapy or hemodialysis within 30 days, (3) hospitalization for ≥ 2 days in the preceding 90 days and (4) nursing home residence [
14]. To compare baseline characteristics of patients and pathogens in CA-, HCA-, and HA-BSI groups, the following data were collected: age, sex, underlying disease, prior major surgery (within 4 weeks), prior antibiotics use (within 4 weeks), infection focus of BSI, and appropriateness of empirical antimicrobial treatment. Appropriate empirical antibiotic use was defined when agents to which the microbes were susceptible were administered within 48 h after obtaining the blood culture sample. In addition, the antimicrobial susceptibility of each pathogen was classified into four major resistance patterns: cefotaxime-resistant
Enterobacteriaceae (not including carbapenem-resistant), carbapenem-resistant GNB (CRGNB), methicillin-resistant
S. aureus (MRSA), and vancomycin-resistant
Enterococcus (VRE).
Stenotrophomonas maltophilia, which has intrinsic resistance to carbapenem, was also classified as CRGNB [
15]. In addition, difficult-to-treat resistance (DTR), defined as nonsusceptibility to all β-lactams and fluoroquinolones tested, was identified among CRGNB [
16,
17].
Blood culture and drug sensitivity test
Blood was taken via a peripheral vein and/or a central line. Blood was drawn for two sets of blood culture bottles (aerobic and anaerobic, 8-10 cc blood for each bottle), and culture bottles were incubated in the Bactec-9240 system (Becton Dickinson, Sparks, MD) or BacT/Alert 3D system (bioMérieux Inc., Marcy l’Etoile, France).). All samples were cultured in blood agar and MacConkey agar plates in a 35 °C incubator for 24 h and identified using the Vitek MS system (BioMérieux, Hazelwood, MI, USA). All antimicrobial susceptibility tests (ASTs) were performed using A Vitek II automated system (bioMérieux Inc.) according to the Clinical and Laboratory Standards Institute 2018 guidelines [
18]. All procedures were performed according to the manufacturer’s instructions.
Statistical analysis
All statistical analyses were performed using SPSS 23.0 for Window (IBM Corp., 2015, Chicago, IL, USA). To compare characteristics between the CA-, HCA-, and HA-BSI groups, a Student’s t-test or Mann–Whitney test was used to compare continuous variables of two groups, and a one-way analysis of variance (ANOVA) or Kruskal–Wallis test was used to compare continuous variables of multiple groups. Categorical variables were compared by using the chi-square test or Fisher’s exact test. For post hoc analysis for multiple comparison, the Bonferroni method was used. All P values were two-tailed, and P values < 0.05 were considered statistically significant.
Discussion
Our study showed common isolates from BSIs that contributed to mortality and patient clinical characteristics. Pathogens and characteristics of HA-BSIs related to mortality showed very different epidemiology from that of CA-BSIs in terms of species, portal of entry, antimicrobial resistance pattern, and appropriateness of empirical antimicrobial therapy. HA-BSIs contribute considerably to mortality in hospital, and the combination of carbapenem and glycopeptides—which might be the most preferred empirical antibiotics regimens of clinicians—did not prevent HA-BSIs nor did it meet the criterion of appropriate antimicrobial therapy.
GNBs more contributed to mortality than GPCs. Increasing antimicrobial resistance of GNB has been continuously reported and, along with the lack of new licensed antibiotics since the 1990s, has become a significant problem [
10,
11,
19]. In this study, cefotaxime-resistant
Enterobacteriaceae demonstrated the most predominant resistant pattern in the study population, especially in the CA-BSI group. Although ESBL-producing
Enterobacteriaceae are usually associated with HA infection, its acquisition in the community setting has increased since the mid-2000s [
20]. Surveillance data of antimicrobial resistance in South Korea in 2011 showed that only 18.3% of
E. coli were resistant to cefotaxime; however, resistance rates increased to 34.7% by 2017 [
10,
11]. Our study supported the real-world relationship between increasing ESBL-producing
Enterobacteriaceae and mortality.
CRGNBs or DTR-GNBs has been evolving as a challenging health problem in hospital acquired infection [
11,
16]. Numerically, 8.4% (156/1849) of patients who died in hospital had HA-BSI, and 25% (40/156) of these patients died because of acquisition of CRGNB bacteremia, which was most frequently caused by
A. baumannii. Indeed, the high carbapenem resistance of
A. baumannii has been reported worldwide [
12,
21]. Although colistin is usually the only treatment option for carbapenem-resistant
Acinetobacter baumannii (CRAB), the efficacy of colistin for CRAB bacteremia is not very reliable [
22]. One multicenter study from South Korea reported the 28-day mortality of patients with CRAB bacteremia treated with colistin at 61.4% [
23]. However, even for patients with CRAB bacteremia who did not receive appropriate antibiotics, the mortality rate was 69.8% [
24]. High mortality, difficulty in administration of appropriate empirical antibiotics, and lack of reliable antimicrobial therapeutic options might be related to
A. baumannii being the most predominant species from HA-BSI contributing to mortality in this study. Before effective treatment options against CRGNB to introduce, efforts are needed to comply with infection prevention management and to audit antibiotic prescriptions [
25,
26].
In our study,
Candida species were reported as the second most common isolates for HA-BSIs causing patient mortality. Appropriate empirical treatment of candidemia is another challenging problem similar to bacteremia caused by multidrug resistant pathogens. Although early appropriate treatment is associated with favorable outcomes [
27], only a few patients received appropriate empirical anti-fungal agents due to difficulty in early diagnosis [
28]. In our study, only 28.6% of patients who died with candidemia received appropriate antifungal agents. To reduce mortality resulting from candidemia, a more rapid and reliable diagnostic tool than blood culture is needed. The new T2Candida molecular test using magnetic resonance with molecular diagnostics has been introduced and is expected to be a rapid, promising tool for the detection of candidemia [
29]. However, due to high initial and maintenance costs, there is a difficulty in wide application of T2 Candida in the clinical setting [
30]. More accessible tools with high accuracy for candidemia diagnosis should be developed.
There were some limitations in our study. Even though our data were collected from two teaching hospitals, the epidemiology of BSI depends both on patient demographics and the geographical characteristics of each hospital. For example, only a small number of transplant recipients were included in our study; medical centers specializing in transplantation might show different HA-BSI epidemiology. In addition, in our study, we only considered short-term mortality of BSI. From this study, we cannot infer long-term mortality or changes in disability caused by BSI. Further studies are needed.
Conclusions
In conclusion, GNBs comprised a larger portion of microbes related to mortality than GPCs. K. pneumoniae and E. coli were the predominate pathogens associated with mortality in CA- and HCA-BSIs, while A. baumannii and Candida species were for HA-BSI. Along with increasing drug resistance in GNB, inappropriate empirical therapy was primarily observed in the HA-BSI group because of difficulties in predicting drug resistance and identifying fungal pathogens. Our data showed that, in the hospital setting, a significant number of patients die from acquired BSI, which the infection control team should make every effort to prevent. A combination of carbapenem and glycopeptide led to other pathogens arising with higher resistance or to fungal infection. The importance of infection prevention and antibiotics stewardship program should be emphasized, especially until more reliable treatment options for multidrug resistant GNB and rapid diagnostic tools are introduced.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.