Epidemiology, risk factors and outcomes of bloodstream infection caused by ESKAPEEc pathogens among hospitalized children

Background Bloodstream infection (BSI) resulting from ESKAPEEc pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Escherichia coli, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp) is relevant to high mortality and economic cost. Data concerning the impact of BSI due to ESKAPEEc in pediatric population was virtually scant. Our purpose was to summarize the epidemiology, risk factors and outcomes of ESKAPEEc BSI among hospitalized children. Methods Inpatients diagnosed with BSI with definite etiology between January 2016 and December 2018 were enrolled retrospectively at the West China Second University Hospital. Data were systematically reviewed on patients’ clinical characteristics and laboratory findings to ascertain independent predictors, clinical features and outcomes. Results Of the 228 patients with BSI, 174 (76.3%) were caused by ESKAPEEc (124 MDR-ESKAPEEc). Multivariate analysis demonstrated that premature and/ or low birth weight (odds ratio [OR] = 2.981, P = 0.036), previous surgery and/or trauma (OR = 5.71, P = 0.029) and source of urinary tract infection (OR = 10.60, P = 0.004) were independently associated with ESKAPEEc BSI. The independent risk factor for MRD-ESKAPEEc BSI was nosocomial infection (OR = 3.314, P = 0.037). The overall mortality rate in patients with ESKAPEEc BSI was 14.4% (25/174), and no significant difference was ascertained in mortality between MRD-ESKAPEEc and non-MRD ESKAPEEc BSI groups (13.7% vs. 11.4%, P = 0.692). In addition, previous surgery and/or trauma, thrombocytopenia, and mechanical ventilation were significant risk factors for mortality caused by ESKAPEEc BSI. Conclusions More than two-thirds of BSI among hospitalized children were caused by ESKAPEEc. Previous surgery and/or trauma, thrombocytopenia and mechanical ventilation increased the risk rate for mortality in ESKAPEEc BSI. The risk factors ascertained could assist physicians to early suspect ESKAPEEc BSI and MDR ESKAPEEc BSI.


Background
The notorious group of pathogens, Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Escherichia coli, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp., named "ESKAPEEc" owing to their high resistances to multiple antimicrobial agents, have recently aroused global concern [1,2]. Currently, the incidence of bloodstream infection (BSI) caused by ESKAPEEc has increased rapidly [3], and ESKAPEEc BSI brought about worse outcomes [3], longer hospital stays, higher economic costs [4][5][6], and increased mortality [4][5][6]. In addition, effective antimicrobial agents against ESKAPEEc strains were limited due to the growth of resistance to multiple antibiotics in these bacterial species. Inappropriate and delayed empirical antimicrobial agents' treatment for patients with BSI was connected with high risk of mortality [7][8][9][10]. Therefore, a wide understanding of the main clinical characteristics of ESKAPEEc BSI among hospitalized children was crucial for physicians to early recognition and select proper empirical therapy.
To our knowledge, the existing studies have given attention to the epidemiology and antimicrobial resistance trends of ESKAPEEc in patients with BSI [11][12][13], and clinical data about ESKAPEEc BSI were mainly available in adult populations with cancer or solid organ transplantation [14][15][16][17]. So far, there was no data concerning the BSI caused by ESKAPEEc in pediatric populations. Thus, we sought to investigate the epidemiology, clinical characteristics among hospitalized children with ESKAPEEc BSI. Simultaneously, we also assessed risk factors, and clinical outcomes of them.

Study population
Hospitalized children who were diagnosed with BSI based on a positive blood culture between 2016 and 2018 at the West China Second University Hospital of Sichuan University and aged under 14 years old were enrolled retrospectively after obtaining ethics approval. The exclusion criteria were listed next ( Fig. 1): (1) polymicrobial infections;(2) diagnosed with fungal BSI; (3) incomplete clinical data.

Data collection
This study obtained the following data from electronic medical records, including age, gender, previous hospitalization (within 1 month), history of surgery and/ or trauma (within 3 months), records of previous antibiotic use (within 1 month), underlying diseases, nosocomial infection or not, symptoms, microbiology data (microorganisms and resistance to antimicrobial agents), likely source of infection, blood products transfusion, pediatric intensive care unit (PICU) admission, invasive operation (indwelling gastric tube, central venous catheter, urinary catheter, mechanical ventilation), empirical antibiotics therapy, length of hospital stay and patients' clinical outcomes. Meanwhile, the additional laboratory results within the first 24 h of admission were collected: the blood routine, C-reactive protein, alanine aminotransferase (ALT) and aspartate aminotransferase (AST). This study was analyzed through three parts. First, all enrolled patients were classified into ESKAPEEc and non-ESKAPEEc groups to ascertain the risk factors of ESKAPEEc BSI. Second, risk factors of patients with BSI resulting from MRD ESKAPEEc were investigated by comparing the MRD ESKAPEEc and non-MRD ESKA PEEc groups. Finally, the 174 patients with ESKAPEEc BSI according to patients' final condition at discharge were divided into survivor and non-survivor groups to investigate the risk factors closely connected with mortality.

Definitions
BSI was defined as the causative bacteria isolated in the blood cultures and clinical manifestations in accordance with sepsis syndrome [18]. The diagnosis criteria were determined by the definitions of National Healthcare Safety Network and CDCP [19]. ESKAPEEc were previously defined elsewhere [1,2,4,12]. The definition of MRD was based on the international expert proposal for interim standard [20]. Polymicrobial infection referred to ≥2 causative organisms isolated from a single blood culture [21]. Empirical antibiotic treatment was regarded as being "appropriate" if the isolated causative organisms were susceptible to ≥1 of the antimicrobial agents administered in vitro. Otherwise, the treatment was considered to be "inappropriate". Overall mortality referred to death by any cause during hospitalization. Some children who had multiple organ failure, demanded inotropes support or mechanical ventilation were discharged from hospital voluntarily and gave up any further treatment. These children were considered nonsurvivor groups within this period.

Microbiological Methods
The blood cultures were fulfilled strictly in accordance with the standard operating procedures. The causative bacteria identification and antibiotic resistance testing were conducted using the VITEK 2 (BioMérieux, Marcyl'Étoile, France) system. The interpretation criteria of antibiotic resistance tests was determined by the definitions of Clinical and Laboratory Standards Institute [22].

Statistical analysis
We reported the median (interquartile range [IQR]) for measurement dates and analyzed them using Mann-Whitney U test appropriately. We calculated constituent ratios for count dates and analyzed them utilizing the χ 2 test or Fisher's exact test. Variables in the univariate analysis that exhibited statistical differences (P-value < 0.05) were placed into the binary logistic regression analysis to investigate factors potentially associated with ESKAPEEc BSI, MRD-ESKAPEEc BSI and mortality. All dates were implemented with the SPSS version 22.0. Pvalues < 0.05 were deemed significant.

Comparison of MDR ESKAPEEc and non-MDR ESKAPEEc BSI
The 168 patients with resistance data were divided into MDR ESKAPEEc and non-MDR ESKAPEEc BSI groups.
The differences of the main characteristics between the 2 groups were showed in Table 3. MDR ESKAPEEc patients with BSI had more nosocomial infections (41.1% vs 11.4%, P < 0.001) and the presence of underlying disease (72.6% vs 43.2%, P < 0.001) was also higher, whereas the age and sex between the 2 groups were not statistically different. Compared with non-MDR ESKAPEEc BSI, patients with MDR-ESKAPEEc BSI were less likely to have a history of antibiotic use within 1 month (21.8% vs 43.2%, P = 0.006), source of skin or soft tissue infection (6.5% vs 27.3%, P < 0.001). Furthermore, the median level of platelet count was significantly lower in MDR-ESKAPEEc group with BSI than that in non-  (Table 3). Whereas, no significant differences were ascertained in MODS, septic shock, mechanical ventilation, PICU admission, mortality between MDR and non-MDR ESKAPEEc BSI groups (all p > 0.05).

Predictors for mortality among hospitalized children with ESKAPEEc BSI
A total of 174 hospitalized children with ESKAPEEc BSI were identified in our study. The overall mortality rate of these patients was 14.4% (25/174), and no significant difference was ascertained regarding mortality between MRD-ESKAPEEc and non-MRD ESKAPEEc BSI groups    The interpretation for the abbreviations were listed in Table 2 [12] and Rome (61.7%) [13], but higher than that in the US (27.2%) [11]. This discrepancy may be associated with geographical regions and study population. Previous studies frequently reported MDR bacteria were Escherichia coli and Klebsiella pneumoniae, most of which trended to be ESBLproducing organisms [23][24][25]. Furthermore, an upward trend in the proportion of ESBL-producing Enterobacteriaceae ESKAPEEc BSI was found [12]. Similarly, a high rate of MRD-ESKAPEEc BSI resulting from an ESBLproducing strain among children was also observed in this study. One common reason in many developing countries which may result in the high incidence of antibiotic resistant BSI was the overuse and misuse of antibiotics. Meanwhile, our study revealed that the most common carbapenem non-susceptibility strains were Klebsiella pneumoniae (6.8%,15/222), which was consistent with data reported by CHINET (5.5%) [26], but differed from the result from southwest China (12.8%) [12]. Studies have shown that carbapenem-resistant Klebsiella pneumoniae (CRKP) was relevant to high mortality [27,28]. Yet the treatment regimens for CRKP BSI were limited in clinic. CRKP infections have posed an extreme antibiotic-resistant threat [28]. Thus, additional studies should concentrate on the molecular and epidemical mechanism of CRKP.
To our knowledge, there was no study using multivariate analyses to investigate risk factors for ESKAPEEc BSI. In the present study, we found that premature and/ or low birth weight, previous surgery and/or trauma and source of urinary tract infection were independent risk factors for ESKAPEEc BSI. Moreover, there was only two studies among adult population focusing on the risk factors for drug-resistant ESKAPEEc BSI in multivariate analyses [14,15]. Gudiol C, et al. in 2014 have found that drug-resistant ESKAPEEc bacteremia were significantly associated with underlying diseases, previous antibiotic use, and source of urinary tract infection [14]. Marta B, et al. in 2013 have revealed that solid-organ transplant patients who had previous antibiotic use and septic shock were more likely to develop drug-resistant ESKAPEEc bacteremia [15]. Based on the existing information, it was likely that underlying diseases, previous invasive operation, previous antibiotic use and source of infection were the key factors for MRD-ESKAPEEc BSI. We further investigated the independent risk factor for MRD ESKAPEEc patients with BSI was nosocomial infection (OR = 3.314, P = 0.037). This result was not in line with those risk factors ascertained in the previous studies [14,15]. The discrepancies may be owing to the diverse study population, diverse definition of MRD [14][15][16][17], diverse inclusion criteria [14][15][16][17]. Various studies have showed that ESKAPEEc pathogens, especially MRD-ESKAPEEc, have emerged as the predominant opportunistic organisms responsible for nosocomial infections [29][30][31], resulting in severe infection. Accompanying by nosocomial infection, there was urgently need for empiric treatment to cover MRD-ESKA PEEc pathogens among children with BSI. The association between bacterial resistance development and previous antibiotic use remained inconsistent. Some studies revealed that previous antibiotic exposure (especially carbapenem) was an independent risk factor for MRD pathogens infection [14,15,28,32]. Conversely, other study has demonstrated no association was identified between MRD bacterial infections and previous antibiotic use [33], in accordance with our data. Thus, further association between bacterial resistance development and antibiotic use previous to infection need to be monitored closely.
To explore the possible influence of MRD pathogens for the outcome of patients with ESKAPEEc BSI, patients' characteristics were systematically evaluated. The overall mortality rate among children with ESKAPEEc BSI was 14.4% (25/174), and no significant difference was ascertained in mortality between MRD-ESKAPEEc and non-MRD ESKAPEEc BSI patients (13.7% vs. 11.4%, P = 0.692), which was in accordance with previous studies [16,17]. This finding reflected the fact that MRD-ESKAPEEc was not a risk factor associated with worse outcome, and although resistant strains were not easily to cure, might be less virulent. Moreover, previous studies have reported various predictors of poor outcome in patients with ESKAPEEc BSI, including female sex [16], lymphocyte counts < 300/mm 3 [16], corticosteroid therapy [14] and ß-lactam monotherapy [14], septic shock [16,17],ICU admission [14]. In our study, we found that mechanical ventilation (OR = 7.997, P = 0.004) was the strongest risk factor associated with mortality. Further, previous surgery and/or trauma and thrombocytopenia increased the risk rates for mortality among children with ESKAPEEc BSI. Thrombocytopenia was a well-known indicator of serious infections development [34,35], which can be considered an early screening tool for poor outcome in children with ESKA PEEc BSI. Low platelet count resulting from bacterial infection was recorded closely and trended to indicate impaired production caused by bone marrow suppression and/or increased destruction owing to endothelial cell injury and platelet aggregation [36][37][38].
Though it is the first time to summarize the main characteristics of ESKAPEEc BSI among hospitalized children, several limitations exist. First, our study is a single-center study representing most of the pediatric population in the Southwest China, some findings may not be generalizable to other settings. Second, selection and recall bias occurred inevitably due to the retrospective analysis and these variations may partly affect the findings of the study.

Conclusions
The present study revealed that more than two-thirds of BSI among hospitalized children were caused by ESKA PEEc. It also demonstrated that previous surgery and/or trauma, thrombocytopenia and mechanical ventilation increased the risk rate for mortality in ESKAPEEc BSI. The risk factors ascertained could assist physicians to early suspect ESKAPEEc BSI and MDR ESKAPEEc BSI.