Blood Cultures and Broad-Spectrum Antimicrobials are Essential for Patients with Out-of-Hospital Cardiac Arrest

Background: Etiologies of out-of-hospital cardiac arrest (OHCA) have been majorly focused on cardiac origins. Little is known regarding the role of bloodstream infections (BSIs) in OHCA episodes. Our aim was to disclose clinical features and incidences of BSIs and the survival benet of prompt administration of appropriate antimicrobial therapy (AAT) in adults with OHCA. Methods: In the 10-year multicenter cohort, clinical information was retrospectively collected and causative microorganisms were prospectively stored for species identication and susceptibilities. The effect of delayed AAT administration on 30-day mortality was examined after adjustment for independent predictors of mortality, recognized by a multivariate regression analysis. Results: Of 1,021,177 emergency department encounters, OHCA visits had a higher BSI incidence than non-OHCA visits (413/3,429, 12.0% vs. 7,429/242,302, 3.1%; P<0.001). Compared with the matched non-OHCA (2,478) patients, OHCA (413) patients experienced more bacteremic episodes due to lower respiratory tract infections, fewer urosepsis events, fewer Escherichia coli bacteremia, and more streptococcal and anaerobes bacteremia. More antimicrobial-resistant organisms, such as methicillin-resistant Staphylococcus aureus and extended-spectrum beta-lactamase-producing Enterobacteriaceae, were evident in OHCA patients. Notably, each hour delay in AAT administration was associated with an average increase of 0.8% in crude 30-day mortality rates (adjusted odds ratio [AOR], 1.008; P = 0.04) in OHCA patients, 7% (AOR, 1.007; P < 0.001) in critically ill patients without OHCA, and 3% (AOR, 1.003; P < 0.001) in less critically ill patients. Conclusions: BSIs should be considered in patients experiencing OHCA; bacteremia sources, causative microorganisms, and antimicrobial susceptibilities differed between the OHCA and non-OHCA patients. The incorporation of blood culture samplings and rapid initiation of broad-spectrum antimicrobial therapy as the rst-aids is essential for OHCA patients.


Introduction
Bloodstream infections (BSIs) are associated with substantial morbidity and mortality that cause a signi cant burden of healthcare costs [1]. Community-onset bacteremia was reported to have an annual incidence of 0.14%-0.82% [2,3] and the short-term fatality rate of up to 41.5% [4]. In addition to hemodynamic support, prompt administration of appropriate antimicrobials has been evidenced to provide survival bene ts for patients with BSIs, particularly those who are critically ill [4][5][6][7].
Despite globe variation in the incidence and survival rate, out-of-hospital cardiac arrest (OHCA) heralded a universally miserable outcome and its main etiology was originally considered to be of cardiac origins [8].
Although many studies have reported non-cardiac causes of OHCA [9][10][11], little is known regarding the prevalence and characteristic of BSIs in patients with OHCA [12]. Therefore, emergency department (ED) physicians do not include blood cultures to be a routine test for patients with OHCA and do not administer antimicrobial therapy as the rst aids to such patients. Accordingly, we aimed to elucidate the etiological role of BSIs and the potential bene t of early administration of appropriate antimicrobials in adults experiencing OHCA.

Study design and sites
The multicenter retrospective study was conducted from January 2012 to December 2017 at the EDs of three hospitals with 1,200, 460, and 380 beds, respectively, in southern Taiwan. The targeted population was OHCA adults with community-onset bacteremia at the ED. All resuscitations were directed by boardcerti ed ED physicians and nurses following the contemporary guidelines of advanced cardiac life support (ACLS). For patient experiencing OHCA, their information prior to ED arrival was captured from an Utstein-style population database, a prospective registry database, in southern Taiwan, and partial information in this database has been published [13]. The study was approved by the institutional review boards of three study hospital and the requirement of informed consent was waived.

Patient Population
First, medical records of all ED visits during the study period were retrospectively reviewed in a computer database for the OHCA presentation. Patients with OHCA who had blood cultures sampled during or prior to resuscitation and those without OHCA patients who had blood cultures sampled during the ED stay were included. The medical records of adults with documented bacterial growth in blood cultures were carefully reviewed. Patients with contaminated blood cultures, those with hospital-onset bacteremia, and those transferred from other hospitals were excluded from this study. Second, to investigate the initial manifestation of bacteremia in patients with and without OHCA, adults with OHCA with records of community-onset bacteremia were included as the OHCA group. Six patients without OHCA who had community-onset bacteremia and temporally near ED arrival of the enrolled OHCA visit were matched as the non-OHCA comparators. Finally, to study the impact of delayed administration of appropriate antimicrobials on short-term prognoses, regardless of whether the patient experienced OHCA, patients with uncertain fatality or incomplete clinical information within 30 days of bacteremia onset were excluded.

Data Collection
For OHCA and non-OHCA patients, clinical information after ED arrival was collected by reviewing medical records by one board-certi ed ED physician and another infectious-disease clinician, and the discrepancy was solved by the discussion between the authors. Using a predetermined form, the capturing data included demographic information, vital signs, comorbidities, comorbidity severity (McCabe-Johnson classi cation), bacteremia severity (a Pitt bacteremia score) at onset, bacteremia sources, durations and types of antimicrobial therapy, and causative microorganisms. Moreover, frequencies of ED visits within six months before the onset of bacteremia, subsequent discharge through the ED or hospitalization, and short-term outcomes were obtained. The study primary outcome was crude 30-day mortality after bacteremia onset (i.e., ED arrival).
Focusing on OHCA patients, prehospital data included the witness status, the presence of bystander cardiopulmonary resuscitation (CPR) and witnessed cardiac arrest, the response time (de ned as the period between the ambulance departure from the re station and the arrival at the scene) of the emergency medical service (EMS), the EMS scene time (de ned as the gap between the ambulance arrival at the scene and the ambulance departure from the scene), and the EMS transport time (de ned as the period between the ambulance departure from the scene and ED arrival).

Microbiological methods
After sampling, blood cultures were incubated in a BACTEC 9240 instrument (Becton Dickinson Diagnostic Systems, Sparks, MD, USA) for 5 days at 35 °C. The causative microorganisms were identi ed by a semiautomated system (Vitek 2 system, bioMe'rieux, Durham, NC), and were stored for further susceptibility testing using disk diffusion method for aerobes or agar dilution method for anaerobes, if the susceptibility results of administrated antimicrobials were provided by the study hospital.
To further investigate the incidence of antimicrobial-resistant organisms (AROs), antimicrobial susceptibility of ampicillin was tested for enterococci, cefoxitin for Staphylococcus aureus (to re ect methicillin susceptibility), penicillin for streptococci, and ampicillin/sulbactam for anaerobes. For the isolates of Escherichia coli, Klebsiella species, and Proteus mirabilis (EKP), susceptibilities to cefepime and levo oxacin were tested and the production of extended-spectrum beta-lactamases (ESBLs) was determined by the phenotypic con rmatory test of cephalosporin-clavulanate combination disks [14]. All susceptibilities were interpreted by the contemporary CLSI recommendation [15].

De nitions
Cardiac arrest was diagnosed based on the updated de nition issued by the American Heart Association and the International Consensus Conference on Cardiopulmonary Resuscitation [16]. The term 'bacteremia' is generally understood to mean the bacterial growth of blood cultures obtained from central or peripheral venipuncture, after the exclusion of contaminant sampling. In accordance with previous criteria [17], blood cultures with the growth of coagulase-negative staphylococci, Bacillus spp., Micrococcus spp., Propionibacterium spp., or Gram-positive bacilli, were considered as contaminant samplings. Community-onset bacteremia indicates that the place of the bacteremia onset is the community, and includes health care-associated and community-acquired bacteremia [5,7].
Polymicrobial bacteremia was de ned as the isolation of more than one microbial species from a single bacteremic episode, whereas the other was regarded as monomicrobial bacteremia.
As previous descriptions [5,7], antimicrobial therapy was regarded as appropriate, if the following two criteria were ful lled: (i) the route and dosage of antimicrobials was administered as recommended in the 2020 Sanford Guide [18]; (ii) antimicrobials administered were in vitro active against all causative microorganisms of bacteremia, based on the contemporary CLSI breakpoints [15]. The time-to-appropriate antibiotic (TAA) was de ned as the time gap between the rst dose of appropriate antibiotic administration and bacteremia onset [5,7].
Comorbidities were de ned as described previously [19], and the comorbidity severity was graded as the classi cation system proposed by McCabe and Johnson [20]. BSI severity was graded in accordance with a Pitt bacteremia score using a previously validated scoring system during the rst 24 hours after ED arrival [5,7]. Patients having a high Pitt bacteremia score (≥ 4) was categorized as the critical illness, whereas those with a Pitt bacteremia score of < 4 as the less critical illness. Bacteremia sources were clinically identi ed according to one of the following criteria: the presence of an active infection site coincident with bloodstream infections; and the isolation of a microorganism from other clinical specimens before or on the same date as that of bacteremia onset [21]. For complicated bacteremia, the removal of infected hardware, drainage of infected uid collections, or resolution of obstruction for biliary or urinary sources was referred as appropriate control of bacteremia source [22]. Crude mortality was equated with the death from all causes [7].

Statistical analyses
The Statistical Package for the Social Science for Windows (Version 23.0; Chicago, IL, USA) was used for statistical analyses. Categorical variables were compared using the Fisher exact or Pearson chi-square test. Continuous variables were presented as medians (interquartile ranges, IQRs) and were compared using an independent t test or Mann-Whitney U test.
To investigate the independent effect of the TAA (each hour) on 30-day mortality, the variables of 30-day crude mortality, recognized through the univariate analysis with a P value of ≤ 0.05, or 0.1 for a small patient population, and the TAA were together included in a stepwise and backward multivariable logistic regression model. A two-sided P value of ≤ 0.05 was considered statistically signi cant.

Clinical characteristics and outcomes of patients with and without OHCA
The comparisons of clinical variables between patients with and without OHCA, in terms of demographics, bacteremia severity, types and severity of comorbidities, timing of appropriate antimicrobial administration, and patient outcomes, were examined by the univariate analyses (Table 1). OHCA patients were more likely to be the elderly, nursing home residents, bed-ridden status, and to have a critical illness (Pitt bacteremia score ≥ 4), polymicrobial bacteremia, bacteremia due to lower respiratory tract infections, high leucocytes or c-reactive protein, comorbidities of diabetes mellitus, chronic obstructive pulmonary diseases, neurological, psychological, urological, or coronary artery diseases. In addition, OHCA patients had more prior ED visits. However, bacteremia due to urinary tract, intraabdominal, biliary tracts, or skin and soft-tissue infections, and comorbid liver cirrhosis were less common in OHCA patients. The longer TAA and more frequencies of ED visits within prior six months was observed in OHCA patients. Not surprisingly, the crude 7-day, 15-day, or 30-day mortality rate in OHCA patients was higher than that in non-OHCA patients.

Microorganisms and susceptibilities in OHCA and non-OHCA Patients
Because of the presence of 90 and 233 polymicrobial episodes in the OHCA and non-OHCA patients, 526 and 2,715 causative microorganisms were collected, respectively (Fig. 2) (30), respectively, of 164 EKP isolates. These gures were signi cantly higher than those in corresponding pathogens among non-OHCA patients (Fig. 3B).
Impacts of the time-to-appropriate antibiotic on mortality of patients with varied bacteremia severity After the exclusion of patients with uncertain outcome, a positive trend of the TAA in 30-day mortality rate among OHCA patients (γ = 0.966, P = 0.002), non-OHCA patients with the critical illness (γ = 0.995, P < 0.001), and the less critically ill patients (γ = 0.960, P = 0.002) was evidenced in Fig. 4  0.03 AOR = adjusted odds ratio; CI = con dence interval; COPD = Chronic obstructive pulmonary diseases; NS = not signi cant (after processing the backward multivariate regression); OR = odds ratio.
* The time-to-appropriate antibiotic, a continuous variable, was included in the multivariable logistic regression model; boldface indicates statistical signi cance with a P value of ≤ 0.01 under the multivariate regression model.
For 427 non-OHCA patients with a critical illness, the TAA remained as an independent determinant (AOR, 1.007; P < 0.001) of 30-day mortality, after adjustment of four independent predictors (i.e., inadequate source control, fatal comorbidities, bacteremia due to urinary or lower respiratory tract infections) recognized by the multivariate regression model (Table 2).
For 2,023 with a less critical illness, each hour delay of the TAA was associated with an average increase in 30-day mortality of 0.3% (AOR, 1.003; P < 0.001), after adjustment of independent predictors (nursing home residents, polymicrobial bacteremia, bacteremia due to urinary or lower respiratory tract infections, fatal comorbidities, and comorbidities of liver cirrhosis, malignancy, or chronic obstructive pulmonary diseases).

Discussion
The present study demonstrated a higher contamination rate of blood cultures and bacteremia incidence in patients with OHCA, compared with those without OHCA at the ED. Notably, signi cant differences in bacteremia sources, causative microorganisms, and susceptibilities between two patient groups were evident. Compared with the matched non-OHCA patient with community-onset bacteremia, the OHCA patient exhibited a greater likelihood of having bacteremia due to lower respiratory tract infections and a lower likelihood of having urosepsis. Furthermore, more bacteremic events due to anaerobes and streptococci, but fewer events due to E. coli were observed in OHCA patients, and AROs more often existed among such patients. More importantly, the survival bene t of early administration of appropriate antimicrobials in OHCA patients was rstly recognized. Accordingly, the clinical practice of blood culture samplings should be encouraged and the incorporation of broad-spectrum antimicrobials as empirical therapy into the antibiotic stewardship program should be considered as the rst aids for OHCA patients.
Generally, the etiologies of cardiac arrest are presumed to be of cardiac origins, regardless of the presenting cardiac rhythm. However, the fact that non-cardiac origins account for up to 30% of cardiac arrest population should not be ignored [23]. Recent investigations dealing with non-cardiac etiologies have indicated that asphyxia, respiratory failure, and comorbid respiratory diseases are the leading etiologies [9,10,24], but information available on infectious etiologies is limited in the literature. In our OHCA cohort, lower respiratory tract infections were the most common etiology of bacteremia. Moreover, risk factors linked to the development of lower respiratory tract infections, such as the elderly, comorbid neurological diseases, and bed-ridden status, were frequently observed in OHCA patients. Accordingly, we reasonably infer that lower respiratory tract infection is the leading etiology in the frail elderly presenting with OHCA.
Although optimization in the "chain of survival" has been evidenced to improve survival and neurological outcomes in patients with OHCA [24,25], the appropriateness of empirical antimicrobial therapy has not been address during the formulation of strategies to reduce morbidity and mortality. Generally, a delay in the administration of appropriate antimicrobials increases the risk of unfavorable prognoses in adults with community-onset bacteremia, particularly in those who are critically ill [5,7]. In the latter presenting with cardiac arrest, the prognostic bene t of prompt administration of appropriate antimicrobials was recognized herein. Because of a higher extent of antimicrobial resistance among the etiological pathogens, a longer elapse of the TAA in patients with OHCA who had community-onset bacteremia was not surprised.
Studies have identi ed several prehospital predictors of survival from OHCA, such as bystander CPR [24,25] , , witnessed cardiac arrest [26], EMS transport time [27], ambulance response time [28], and EMS scene time [29]. Although two predictors, namely bystander CPR and witnessed cardiac arrest, were demonstrated to be independent determinants of survival outcomes, the time-related EMS variables, in terms of the response time, transport time, and scene time, were not associated with patient fatality in our cohort. The cohorts in previously established studies differ from those in our study because we included only patients with documented BSIs. Notably, our work is the rst to together investigate the effects of pre-hospital factors and antimicrobial therapy on the prognosis of patients experiencing OHCA. We believe that impacts of any time-related EMS variable on patient prognoses might be negligible, because the elapsed TAA was found to be considerably greater than any of the values for time-related EMS variables. Accordingly, efforts to provide prompt treatment for patients with OHCA through the administration of appropriate antimicrobial therapy should be emphasized.
Contaminated blood cultures was widely known to adversely affect health care and medical expenditures, causing unnecessary hospitalizations, antimicrobial administration, and microbiological studies [30].
Furthermore, ED overcrowding is a growing problem worldwide and is associated with blood culture contamination [31], delayed administration of appropriate antibiotics [32,33], and unfavorable patient outcomes [33]. Patients with cardiac arrest are emergently resuscitated and comprehensively evaluated, which substantially increases the utilization of emergency care resources and the degree of ED overcrowding. Therefore, a high contamination rate of blood culture samples in patients experiencing OHCA is not unexpected.
As the study controls, our non-OHCA patients with BSI were selected at a ratio of 6:1, according to the arrival timing of each patient with OHCA. Such ED encounters were su ciently representative of the comparators of community-onset bacteremia for the following reasons. First, the main causative microorganisms, namely E. coli, S. aureus, Streptococcus species, and Klebsiella species, as well as common bacteremia sources, such as urosepsis, intra-abdominal infections, and soft-tissue infections, in the matched non-OHCA patients, were compatible with those in previously published cohorts with community-onset bacteremia [2,5,7,34]. Second, the proportion of critically ill patients at bacteremia onset (approximate 20%) and the crude mortality rate (approximate 15%) in the present cohort were also similar to the corresponding data reported in other studies [2,5,7,34].
Of AROs in the community, ESBL-producers have well known adverse effects on the prognoses of bacteremic patients in the literature [35]. Clinical predictors of ESBL-producers in community-onset bacteremia, including healthcare facility residents, urinary catheter use, previous antimicrobial therapy, frequent ED visits [36], and previous hospitalization have been reported [35]. Because of the retrospective nature of the present study, some variables were di cult to assess. However, studies assessing the frequency of ED visits have suggested that it may be correlated with increased healthcare use, serious ill health, and socioeconomic distress [37,38]. It was rightfully understandable that etiologic microorganisms in bacteremic patients experencing OHCA were more likely to be antimicrobial-resistant because of frequent utilization of medical care.
Our ndings should be interpreted with cautions for several limitations inherent to the study design. First, the recall basis could not be completely abolished because of the retrospective and observational nature.
Therefore, two authors recorded information to ensure data accuracy and minimize record discrepancies. Second, the present study was conducted at three hospitals located in southern Taiwan, and our ndings may not be applicable to other areas or hospitals. However, we included approximately 3,000 adults that could be regarded as representative of community-onset bacteremia in southern Taiwan. Third, although bacterial translocation raised from intestinal ischemia during the periarrest period might result in secondary bloodstream infections [39], OHCA patients simultaneously experiencing systemic infections were proven for the high level of C-reactive protein herein. Finally, our cohort represents 56% of all patients experiencing OHCA sampled for blood cultures at EDs. Selection bias may have exerted a certain in uence on the results and re ects the inadequate alertness of sepsis as one of the causes of OHCA, as observed in the literature [9,10]. We believe a multicenter, prospective study to evaluate the incidence and signi cance of BSIs are warranted in the future.

Conclusions
Sampling of blood cultures can be routinely performed for the identi cation of infectious etiologies of OHCA. Among OHCA patients with community-onset bacteremia, the survival advantage of early administration of appropriate antimicrobials and the prevalence of AROs indicates the incorporation of broad-spectrum antimicrobials with anti-anaerobic coverage as the empirical therapy is warranted. Declarations Figure 1 The owchart of patient selections. ED = emergency department; OHCA= out-of-hospital cardiac arrest.
*Six non-OHCA patients temporally near the ED arrival of OHCA visits were matched. between two groups. # The denominators and numerators respectively indicate the isolate numbers of all microorganisms and antimicrobial-resistant pathogens.