Effects of Inappropriate Empirical Antimicrobial Therapy on the Outcomes of Patients with Community-Onset Bacteremia: Age Matters


 Background: Studies have reported the effects of delayed administration of appropriate empirical antimicrobial (AEA) on the prognosis of patients with bloodstream infections; however, whether there is an age-related difference in these effects remains debated. Methods: In this 4-year multicenter cohort study, patients with community-onset bacteremia were retrospectively included and categorized into “middle aged” (45–64 years), “old” (65–74 years), and “very old” (≥75 years) groups. To determine the timing of AEA administration for each patient, all causative microorganisms were prospectively obtained. For each age group, the effects of delayed AEA administration on 30-day mortality were investigated after adjustment for the independent predictors of 30-day mortality determined using a logistic regression model.Results: Significant differences were observed in the distribution of comorbidity types, comorbidity severity, bacteremia sources, bacteremia severity, and causative microorganisms among 968 (33.2%) middle-aged, 683 (23.4%) old, and 1,265 (43.4%) very old patients. Although significant effects (adjusted odds ratio [AOR], 1.002; P = 0.07) of delayed AEA administration on prognosis were not observed in middle-aged patients, each hour of AEA delay resulted in an average increase in the 30-day crude mortality of 0.036% (AOR, 1.0036; P < 0.001) and 0.38% (AOR, 1.0038; P < 0.001) in old and very old patients, respectively. Practically in critically ill patients, each hour of delayed AEA administration resulted in an average increase of 0.03% (AOR, 1.003; P = 0.04), 0.4% (AOR, 1.004; P < 0.001), and 0.5% (AOR, 1.005; P = 0.001) in 30-day crude mortality in middle-aged, old, and very old patients, respectively. Conclusions: Regardless of bacteremia severity, the adverse effects of delayed AEA administration on the survival of patients with community-onset bacteremia increased with patients’ age. To achieve favorable outcomes, rapid AEA administration is recommended in older patients.


Introduction
Community-onset bacteremia is a common problem encountered by emergency department (ED) physicians. The annual incidence of community-onset bacteremia was 0.82% in a population-based study [1], and its high morbidity and mortality of up to 40% can result in considerable healthcare costs [2,3]. The hemodynamic support and early administration of appropriate antimicrobial therapy can offer survival bene ts for patients with bacteremia, particularly those with critical illness at the onset of bacteremia [4][5][6]. However, because of an increased proportion of causative microorganisms harboring antimicrobial resistance genes [7], rapid administration of appropriate empirical antimicrobials (AEAs) remains challenging for ED physicians.
Appropriate management has always been di cult in older patients with bloodstream infections because bacteremia incidence and mortality markedly increase with age [8] and older patients may present with nonspeci c functional decline and atypical presentation [9]. Moreover, a recent study reported that agerelated differences in bacteremia severity and antimicrobial-resistant isolates cause an increased delay in Page 3/25 AEA administration [10]. Because the severity of bacteremia increases with patients' age, we hypothesized that the prognostic effects of delayed AEA administration would be enhanced with age. Therefore, in this study, we investigated and compared the effects of delayed AEA administration on short-term outcomes of adults aged ≥ 45 years with community-onset bacteremia among different age groups.

Study design and sites
Between January 2015 and December 2018, we conducted a 4-year multicenter retrospective cohort study at the EDs of three hospitals, namely a university-a liated medical center (1,300 beds) and two teaching hospitals (one with 460 beds and one with 380 beds) in southern Taiwan. The cohort consisted of patients aged ≥ 45 years who had developed community-onset bacteremia in the ED. This study was approved by the institutional review board of the study hospital, and the requirement of obtaining informed consent was waived. Clinical information was obtained and reported according to the Strengthening the Reporting of Observational Studies in Epidemiology.

Patient selection
From a database search, we identi ed adult patients who had undergone blood culture sampling at the ED during the study period. We retrieved the medical information of patients with bacterial growth in their blood cultures. We excluded patients aged < 45 years and those with contaminated blood cultures, fungemia, mycobacteremia, hospital-onset bacteremia or bacteremia prior to arrival at the ED, an uncertain mortality date, and incomplete chart records. Among patients who had multiple episodes of bloodstream infections during the study period, we examined only the rst episode. The primary study endpoint was crude 30-day mortality after arrival at the ED (i.e., onset of bacteremia).

Data collection
We retrospectively collected the demographic and clinical data of eligible patients by using a predetermined form. Clinical data were jointly collected by a board-certi ed ED physician and an infectious disease physician who were trained in reviewing medical charts and blinded to the aim and hypothesis of this study. Any discrepancy in the recording of clinical data was resolved through discussion between the two physicians. The following patient information was collected: age, vital signs, laboratory data, comorbidity types and severity (according to the Charlson comorbidity index [CCI]), severity of bacteremia (according to the Pitt bacteremia score [PBS]), duration and types of antimicrobial agents administered, and length of ED stay. Moreover, data regarding the length of hospital stay, durations and types of antimicrobial agents administered, imaging study ndings, timing and types of surgical and radiological interventions, causative microorganisms, bacteremia diagnosis, and 30-day mortality were collected.

Microbiological methods
Blood cultures were incubated in a Bactec 9240 instrument (Becton Dickinson Diagnostic Systems, Sparks, MD, USA), and the causative microorganism was identi ed using a semiautomated system (Vitek 2 system, bioMe'rieux, Durham, NC, USA). After the causative microorganism had been identi ed, it was stored for further susceptibility testing, which was performed using the disk diffusion method for aerobes and the agar dilution method for anaerobes in accordance with the guidelines of the Clinical and Laboratory Standards Institute (CLSI) [11]. The susceptibility of Gram-negative aerobes to the following antibiotics was tested: ampicillin/sulbactam, piperacillin/tazobactam, levo oxacin, moxi oxacin, cefazolin, cefuroxime, cefotaxime, ceftazidime, cefepime, ertapenem, and imipenem. For streptococci and enterococci, the tested antibiotics were penicillin and ampicillin, respectively. Furthermore, the susceptibility of anaerobes to ampicillin/sulbactam, piperacillin/tazobactam, metronidazole, and moxi oxacin was tested. To determine the timing of AEA administration for each patient, provided that empirical antibiotics were not included in the initial susceptibility panel, the susceptibility to the indicated antibiotic was tested.

De nitions
Eligible patients were categorized as "middle aged" (45-64 years), "old" (65-74 years), and "very old" (≥ 75 years) according to the cut-off age used in a previous study [12]. Community-onset bacteremia refers to bacteremic episodes with onset in the community and includes healthcare facility-acquired and community-acquired bacteremia [4,13]. According to a previously described criterion [14], contaminated blood cultures are de ned as those exhibiting the growth of potential contaminating pathogens.
According to previous studies [4,13], antimicrobial therapy is appropriate when the following criteria are ful lled: (i) the route and dosage of antimicrobial agents administered adhere to the recommendations in the Sanford Guide [15] and (ii) all bacteremia-causing pathogens in one episode of bacteremia are susceptible to the antimicrobials administered in vitro according to the CLSI criteria [11]. Time to appropriate antibiotic was de ned as the period between the onset of bacteremia (arrival at the ED) and the administration of the rst dose of appropriate antimicrobials [4,13].
The severity of bacteremia was determined according to the PBS, which was calculated using a previously validated scoring system based on the following variables assessed during the rst 24 hours after ED arrival: vital signs, usage of vasopressor agents, mental status, receipt of mechanical ventilation, and cardiac arrest [4,13]. Patients with a PBS of ≥ 4 were categorized as critically ill. Comorbidities were de ned as described previously [16], and malignancies included hematological malignancies and solid tumors. The severity of comorbidities was examined using the CCI [17], and patients with a CCI of ≥ 5 were regarded as having severe comorbidities. Crude mortality was de ned as death from all causes.

Statistical analyses
Statistical analyses were performed using the Statistical Package for Social Science for Windows (version 23.0; Chicago, IL, USA). Categorical and continuous clinical variables were compared using the Pearson chi-square test or Fisher's exact test, if the expected value was < 5, and the independent t-test, respectively. Spearman's correlation coe cients were calculated to analyze age-related trends in clinical covariates and outcomes. To investigate the effects of delayed AEA administration on 30-day mortality, the signi cant predictors of 30-day mortality recognized by the univariate analysis and the continuous covariate of "time to appropriate antibiotic" were jointly processing under the model of hierarchical logistic regression. The E-value was calculated to examine the potential effect of unmeasured confounders in our study [18]. A two-sided P value of < 0.05 was considered signi cant.

Results
Demographic characteristics of the entire cohort Initial manifestations and clinical outcomes in different age groups Table 1 presents a comparison of clinical manifestations and outcomes between the middle-aged, old, and very old groups. Signi cant differences were noted in the proportion of male patients, the proportion of nursing-home residents, major comorbidities, bacteremia severity (PBS), major bacteremia sources, types of empirical antimicrobials administered, and presence of polymicrobial bacteremia among the three groups. In addition, we observed a positive age-related trend (all γ = 1.00, P < 0.01) in bacteremia severity; the proportion of nursing-home residents; comorbidities of hypertension, heart failure, or coronary artery diseases; presence of polymicrobial bacteremia and bacteremic pneumonia; and 30-day crude mortality rates.

Predictors of 30-day mortality in overall patients
In the univariate analysis (Table 2), the following variables were found to be positively associated with 30-day mortality: male sex; residence at a nursing home; polymicrobial bacteremia; critical illness (PBS ≥ 5) at onset; inadequate source control; bacteremic pneumonia; S. aureus, Klebsiella species, or Pseudomonas species as the causative microorganisms; severe comorbidities (CCI ≥ 5); and comorbidities of malignancies, liver cirrhosis, and neurological diseases. Other predictors of 30-day mortality included bacteremia due to liver abscesses, urinary or biliary tract infections, E. coli as the causative microorganism, and comorbidities of hypertension and diabetes mellitus.   (Table 4), the adverse effect of the AEA delay (every hour) on 30-day mortality was signi cant (AOR, 1.003; P = 0.04) after adjustment for the following predictors of 30-day mortality recognized in the univariate analysis: polymicrobial bacteremia; bacteremic pneumonia; bacteremia due to urinary tract infections and liver abscesses; severe comorbidities (CCI ≥ 5); and comorbidities of malignancies, diabetes mellitus, and liver cirrhosis. Among the 144 and 297 critically ill patients in the old and very old groups, the adverse effect of AEA delay (every hour) on 30-day mortality remained signi cant (AOR, 1.004; P < 0.001 and AOR, 1.005; P < 0.001, respectively) after respective adjustment for the predictors of 30-day mortality recognized in the univariate analysis (Table 4).

Discussion
Similar to the ndings of previous studies [4][5][6], we observed a signi cant effect of delayed AEA administration on patient survival in the overall cohort. The results of the subgroup analysis revealed that the adverse effects of delayed AEA administration on the prognosis of middle-aged patients were not signi cant; however, we found that each hour of delay in AEA administration resulted in an average increase of 0.36% and 0.38% in the 30-day crude mortality of old and very old patients, respectively.
Moreover, in critically ill patients, each hour of delayed AEA administration contributed to an average increase of 0.3%, 0.4%, and 0.5% in 30-day crude mortality in middle-aged, old, and very old patients, respectively. In sum, regardless of bacteremia severity, the effects of delayed AEA administration on short-term mortality increased with patients' age.
Studies have reported age-related differences in the distribution of comorbidities, bacteremia sources, and causative microorganisms [8,10,19,20]. Consistent with these ndings, a positive age-related trend in bacteremia severity, comorbidity severity, and short-term mortality was observed in the present study.
Accordingly, physicians adopted different empirical antimicrobials for patients in different age groups, leading to similar durations of the time to appropriate antibiotic among the different age groups included in this study.
Numerous studies have reported a trend of delay in antimicrobial administration affecting the short-term prognosis of patients with bacteremia in various settings, such as those with community-onset bacteremia [4,13], severe illness [21], nosocomial bacteremia [22], neutropenia [23], and speci c causative microorganisms (i.e., Pseudomonas aeruginosa [24], S. aureus [25], and extended-spectrum beta-lactamase-producing Enterobacteriaceae [26]). Notably, the ndings of the present study indicate that regardless of whether a patient was critically ill, the adverse effects of delay in AEA administration on patient survival increased with patients' age. Accordingly, in older patients with bacteremia episodes, faster AEA administration is recommended. To administer AEAs, epidemiological surveillance of causative microorganisms and susceptibilities or application of the biomolecular technique of rapid pathogen identi cation is essential for older patients.
Similar to previous studies [4,27,28] that have emphasized that severe bacteremia requires earlier AEA administration to achieve favorable outcomes and therapeutic e cacy, we found that each hour of AEA delay resulted in a higher increase in 30-day crude mortality in critically ill patients than in overall patients; that is, middle-aged (0.3% vs. nonsigni cant), old (0.4% vs. 0.36%), and very old (0.5% vs. 0.38%) groups. Therefore, regardless of patients' age, rapid AEA administration is crucial for preventing more severe episodes of bacteremia. Early identi cation of critically ill patients with bloodstream infections among ED patients through the use of a scoring algorithm, such as the Mortality Emergency Department Sepsis score [29] or quick Sequential Organ Failure Assessment [30], is crucial. These tools can assist in accurately assessing bacteremia severity at onset and identifying patients with a high mortality risk.
This study has several limitations. First, consistent with the ndings of previous studies regarding the appropriateness of empirical antimicrobial administration [5,13,31,32], no detailed components of the Surviving Sepsis Campaign were evaluated as covariates. However, this study investigated the adverse effects of inadequate source control for complicated bacteremia that are superior to the ndings of reports regarding this campaign. Second, because previous studies have reported the neglected difference in therapeutic e cacies between narrow-and broad-spectrum antimicrobials administered as the appropriate empirical agent [33,34], the differential e cacy between various AEAs was not considered herein owing to the study design being based on previous ndings [5,13,31,32]. Third, to investigate the effects of AEA timing on patient prognosis, patients with undetectable antimicrobial administration, incomplete clinical information, or uncertain date of death were excluded. Selection bias may have occurred as a result of the small sample size in this study. Fourth, the retrospective nature of this study made it prone to recall bias during data collection. To reduce this bias, all clinical information was randomly retrieved by two physicians who were blinded to the study hypothesis, and they inspected medical records together to resolve any discrepancies. Fifth, a low E-value was obtained for the prognostic effects of delayed AEA administration on various age groups; thus, unmeasured confounders in our cohort should be trivial. Finally, because study hospitals were located in southern Taiwan, the results of this study should be validated in other communities with variable bacteremia sources, causative microorganisms, or antibiotic-resistant isolates.

Conclusions
Regardless of bacteremia severity, the adverse effects of delayed AEA administration on patient mortality increased with patients' age. Furthermore, each hour of AEA delay resulted in a higher increase in 30-day crude mortality in critically ill patients than in overall patients in various age groups. Therefore, to achieve favorable outcomes in older patients with bacteremia, particularly those who are critically ill, rapid pathogen identi cation is necessary for prompt AEA administration.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.