Time to Appropriate Antibiotic Therapy is an Independent Indicator of Poor Outcome in Children with Nosocomial Klebsiella Pneumoniae Bloodstream Infection

Objectives: To evaluate the effects of time to appropriate therapy (TTAT) on outcomes in children with nosocomial K. pneumoniae bloodstream infection, and to nd an optimal time window for empiric antibiotics administration. Methods: Children with nosocomial K. pneumoniae bloodstream infection hospitalized in Children’s Hospital of Chongqing Medical University from April 2014 to December 2019 were enrolled retrospectively. TTAT cutoff point and risk factors were determined and analyzed by Classication and Regression Tree (CART) analysis and Logistic Regression analysis. Results: Overall, sixty-seven patients were enrolled. The incidence of septic shock and mortality was 17.91% (12/67) and 13.43% (9/67), respectively. The CART-derived TTAT cutoff point was 10.7 hours. The multivariate logistic regression analysis indicated delayed therapy (TTAT ≥ 10.7 h), PRISM III scores ≥ 10, early TTP (TTP ≤ 13 h), and need for invasive mechanical ventilation were independent risk factors of septic shock (OR 9.87, 95% CI 1.46-66.59, P = 0.019; OR 9.69, 95% CI 1.15-81.39, P = 0.036; OR 8.28, 95% CI 1.37-50.10, P = 0.021; OR 6.52, 95% CI 1.08-39.51, P = 0.042; respectively) and in-hospital mortality (OR 22.19, 95% CI 1.25-393.94, P = 0.035; OR 40.06, 95% CI 2.32-691.35, P = 0.011; OR 22.60, 95% CI 1.78-287.27, P = 0.016; OR 12.21, 95% CI 1.06-140.67, P = 0.045; respectively). Conclusions: TTAT is an independent predictor of poor outcome in children with nosocomial K. pneumoniae bloodstream infection. Initial appropriate antibiotic therapy should begin within 10.7 hours from the onset of bloodstream infection.


Introduction
Klebsiella pneumoniae (K. pneumoniae) is one of leading gram-negative pathogens of bloodstream infection in hospitalized children [1], also a major worldwide source and shuttle for antibiotic resistance [2], and with high morbidity and mortality. Antibiotic therapy plays a crucial role in the treatment of bloodstream infection, and the time of initiating appropriate antibiotic has signi cant effects on the prognosis [3][4][5][6][7][8][9]. The Surviving Sepsis Campaign in 2020 [10] recommends that the antibiotic should be administrated within 1 hour after the recognition of septic shock, and within 3 hours after the recognition of sepsis-associated organ dysfunction without shock. The 1-hour and 3-hour goals are strongly recommended, while with low quality of evidence and remains controversial [10,11]. Meanwhile, our previous study showed that the delayed appropriate antibiotic therapy ≥ 13.6 hours, not ≥ 1 or 3 hours, was associated with the highest sepsis-related mortality in children with Streptococcus pneumoniae sepsis [7]. Furthermore, the 1-hour and 3-hour goals are sometimes unrealistic to be achieved due to limitations in identi cation and diagnosis of sepsis-associated organ dysfunction and septic shock [11].
Immediate antibiotic treatment is lifesaving for some patients, while, the overdiagnosis of sepsis and aggressive time-to-antibiotic targets may lead to antibiotic overuse and antibiotic-associated harms [8,12]. The Infectious Diseases Society of America states the administration time of antibiotic varies with different pathogens and populations [13]. In adult patients, the optimal appropriate therapy time was 24 hours for K. pneumoniae bloodstream infection [3], 48.1 hours for Enterococci bloodstream infection [4], 52 hours for Pseudomonas aeruginosa bloodstream infection [5] and 44.75 hours for Staphylococcus aureus bacteremia [6]. The appropriate antibiotic time may be different in bacteremia patients with different pathogens and it still remains unknown in pediatric patients with K. pneumoniae bloodstream infection. Therefore, more studies are needed to explore the appropriate antibiotics administration time in different populations. We aimed to evaluate an optimal time window for appropriate antibiotic administration, to determine the effects of time to appropriate therapy (TTAT) on outcomes in children with nosocomial K. pneumoniae bloodstream infection.

Study designs and patients
This was a retrospective, observational cohort study conducted in Children's Hospital of Chongqing Medical University, a 2000-bed tertiary teaching hospital in Chongqing, China, ranked the top two domestic children's hospitals (rank list: http://top100.imicams.ac.cn/home). Patients hospitalized between April 2014 and December 2019 with K. pneumoniae bloodstream infection were enrolled. The inclusion criteria were of the following: (i) inpatients, (ii) aged 1 month to 18 years, (iii) diagnosed with monomicrobial K. pneumoniae bloodstream infection. The exclusion criteria included any of the following: (i) patients diagnosed with community-acquired K. pneumoniae bloodstream infection, (ii) patients with incomplete clinical information and (iii) patients received appropriate antibiotics against K. pneumoniae prior to blood culture. This retrospective study was approved by the Ethics Committee of Children's Hospital of Chongqing Medical University. Informed consent was waived from the parents/guardians owing to the retrospective design of this study.

Data collection and de nitions
The collected data included demographic characteristics (age and gender), underlying conditions, axillary temperature, serum albumin level, sources of infection, severity of illness, antibiotic susceptibility testing, antibiotic therapy during hospitalization, TTAT, time to positivity (TTP) and clinical outcomes (septic shock and mortality).
K.pneumoniae bloodstream infection was de ned as at least 1 blood culture positive for K. pneumoniae associate with related clinical manifestations of infection [14]. Nosocomial bloodstream infection was de ned as positive blood culture obtained > 48 hours after admission, while signs and symptoms of infection were absent at admission [14]. The immunosuppression patients were de ned as patients who received immunosuppressive chemotherapy or high dose steroid therapy more than 2 weeks, or with primary immunode ciency diseases. Hypoalbuminemia was de ned as intravascular albumin level < 2.5 g/dL for children younger than 7 months and < 3.4 g/dL for children 7 months or older [15]. Source of infection was de ned according to the CDC /NHSN surveillance guidelines [16]. The severity of illness was assessed by the Pediatric Risk of Mortality (PRISM) III score [17]. TTP was de ned as the time span between incubation of blood and detection of bacterial growth. Our previous study demonstrated that TTP ≤ 13 hours and a PRISM III score ≥ 10 indicated poor outcomes in children with K. pneumoniae bloodstream infection [18]. Empiric antibiotic treatment was de ned as initial antimicrobial therapy for suspected infection without de nitive microbiologic pathogen identi cation [10]. The appropriate antibiotic therapy was de ned as the patients received at least one intravenous antibiotic documented in vitro susceptibility according to the breakpoint established according to the Clinical and Laboratory Standards Institute (CLSI) guideline [19]. Multi-drug resistant (MDR) was de ned according to the European Centers for Diseases Prevention and Control (ECDC) international expert proposal [20]. The TTAT was de ned as the time duration from onset of bloodstream infection to receive initial appropriate antibiotic therapy [3]. The onset of bloodstream infection was identi ed by no less than two senior infectious disease physicians according to clinical manifestations (e. g. fever, chill and so on) and biomarkers (e. g. C-reactive protein, procalcitonin and so on), and approved by the subsequent positive blood culture result. Sepsis was de ned as infection complicated by one or more organ dysfunctions [21]. Organ system dysfunctions are assessed with an increase in the pediatric Sequential Organ Failure Assessment (pSOFA) score by 2 or more points [22]. Septic shock was de ned as patients with sepsis and hypotension requiring vasopressor therapy and lactate greater than 2 mmol/L despite adequate uid resuscitation [21]. Hypotension was diagnosed according to cutoffs of the age-adapted mean arterial blood pressure in pSOFA score system [22].

Clinical outcomes
The primary outcome was in-hospital mortality, the second outcome was incidence of septic shock.

Statistical analysis
Classi cation and regression tree (CART) analysis [23], which included optimal tree selection based on pruning and 10-fold cross-validation, was used to nd the optimal cutoff point of TTAT, and the patients at highest risk for in-hospital mortality. The CART-derived TTAT cutoff point was also assessed by receiver operating characteristic (ROC) curve analysis [24]. Hazard curves were generated by the Kaplan-Meier method, and differences in survival were compared using the log-rank test. The corresponding inhospital mortality of different cutoff points of TTAT were assessed by the χ 2 test for a linear trend.
Categorical variables were compared by χ 2 test or Fisher's exact test, and continuous variables were compared by Student's t test or Mann-Whitney U test. Univariate and multivariate logistic regression test were constructed to explore independent risk factors of septic shock and in-hospital mortality. Variables with P-level < 0.10 in univariate analysis were further included in multivariate models, with forward likelihood ratio selection. Odds ratio (OR) and the corresponding 95% con dence interval (CI) were calculated. All statistical analyses performed using SPSS software for Windows, v.23 (SPSS Inc., Chicago, IL, USA). The level of signi cance was set at P-value < 0.05 (two-sided).

Study population
One hundred and thirty-two patients were retrospectively enrolled at the beginning. Sixty-ve of them were excluded: sixty cases were classi ed as community-acquired infection, three cases with incomplete clinical information, and two cases received effective antibiotic against K. pneumoniae isolates prior to blood culture. Finally, sixty-seven cases were enrolled in this study (Fig. 1).
Clinical characteristic of K. pneumoniae bloodstream infection in children respectively. There were 28 (41.79%) patients receiving antibiotic therapy prior to blood culture. While, thirty-eight (56.72%) patients were treated with carbapenem empirically before the susceptibility tests.

TTAT of K. pneumoniae bloodstream infection in children
The TTAT cutoff point derived by CART to delineate the risk of in-hospital mortality was 10.7 hours. Patients were classi ed into early (TTAT < 10.7 h) and delayed therapy group (TTAT ≥ 10.7 h) according to TTAT cutoff point. Twenty-three (34.33%) patients received delayed therapy. The in-hospital mortality in delayed therapy group was signi cantly higher than that in early therapy group (29.17% vs 4.65%, P = 0.028). In the subgroup of patients with early therapy, the in-hospital mortality was signi cantly higher in patients with PRISM III scores ≥ 10 than those with PRISM III scores < 10 (33.33% vs 2.50%, P = 0.008). In the subgroup of patients with early therapy and PRISM III scores < 10, patients with TTP ≤ 13 h had remarkably higher in-hospital mortality than those with TTP > 13 h (10.00% vs 0.00%, P = 0.002) (Fig. 2).
In ROC curve analysis, the CART-derived TTAT cutoff point had the best predict value of in-hospital mortality (AUC [95% con dence interval (CI)], 0.721 [0.564-0.879], 77.78% sensitivity and 70.69% speci city), with moderate predictive e cacy [25]. The Kaplan-Meier survival curve of these patients is shown in Fig. 3. In χ 2 test for a linear trend, patients in TTAT ≥ 10.7 h group had the highest in-hospital mortality when compared to those in TTAT < 3 h and 3 h ≤ TTAT < 10.7 h periods groups. (P = 0.008) (Fig. 4).

Comparisons of clinical characteristics between the early and delayed therapy groups
Characteristics of two TTAT groups were shown in Table 2. When compared with the delay therapy group (TTAT ≥ 10.7 h), more patients in early therapy (TTAT < 10.7 h) group had hematologic malignancy (84.09% vs 30.43%, P < 0.001) and immunosuppression (72.73% vs 39.13%, P = 0.007). There were prominently more early therapy patients administrated with carbapenem empirically before the susceptibility tests than delayed therapy patients (68.18% vs 34.78%, P = 0.009). Meanwhile, patients received delayed therapy may attribute to the notably higher proportion of empirical third-generation cephalosporin therapy (26.09% vs 4.55%, P = 0.029) and cephalosporin resistant isolates (39.13% vs 13.64%, P = 0.017) than those received early therapy. Accordingly, the delayed therapy patients had signi cantly higher incidence of secondary hypoalbuminemia (56.52% vs 20.45%, P = 0.002) and septic shock (39.13% vs 6.82%, P = 0.003), higher proportion of requiring invasive mechanical ventilation (34.78% vs 6.82%, P = 0.010), higher in-hospital mortality (30.43% vs 4.55%, P = 0.010) than those early therapy patients. While, the PRISM III scores, the length of stay before the onset of bloodstream infection and length of the whole hospitalization stay were with no differences between the two groups.

Comparisons of clinical characteristics between the survival and non-survival groups
The clinical characteristics of the survival and non-survival patients were compared in the Table 3. The non-survival patients had signi cantly higher proportion of cephalosporin resistant and extended spectrum beta-lactamase (ESBL) positive isolates, higher proportion of PRISM III scores ≥ 10, TTP ≤ 13 h and TTAT ≥ 10.7 h, higher incidence of requiring invasive mechanical ventilation and septic shock when compared to those in survival group. (P < 0.05). The length of stay before the onset of bloodstream infection and length of the whole hospitalization stay were with no signi cant differences between two groups.

Risk factors of in-hospital mortality
Univariate and multivariate analyses were conducted to nd independent risk factors of in-hospital mortality, and the results were presented in

Risk factors of septic shock
The univariate and multivariate logistic regression analysis of risk factors of septic shock were shown in

Discussion
In this study, we demonstrated that patients with PRISM III scores ≥ 10, early TTP (TTP ≤ 13 h), requiring for invasive mechanical ventilation were independent factors associated with poor outcomes, which were in accordance with our previous study [18,26]. Furthermore, we also showed that delayed therapy (TTAT ≥ 10.7 h) was risk factor of septic shock and in-hospital mortality, which was consistent with the results of previous studies indicating delayed appropriate antibiotic therapy was associated with poor outcomes [6,5,4,3,[27][28][29]. Falcone et al. [3] indicated that appropriate antibiotic therapy should begin within 24 h from the collection of blood culture in adult carbapenemase-producing K. pneumoniae bloodstream infection patients.  [3] didn't explore the optimal TTAT cutoff point. Third, although we both enrolled patients with K. pneumoniae bloodstream infection, the patients enrolled in our study were children rather than adult. Two studies [8,9] stated that TTAT > 3 h was related to higher mortality, which was much shorter than that in our study. The explanations may as the following. First, patients with septic shock should administrate appropriate antibiotic more aggressively than those with sepsis-associated organ dysfunction but without shock [10]. There were 17.91% (12/67) patients with septic shock in our study.
While, there were 78.13% (125/160) and 79.23% (103/130) patients with septic shock in Han's [8] study and Weiss's [9] study, respectively. The lower proportion of septic shock patients in our study may explain the longer TTAT cutoff point. Second, the methods of de ning TTAT cutoff points were different. We used the CART analysis while the other two studies used multivariate analysis.
We found that the secondary hypoalbuminemia during hospitalization may be associated with delayed appropriate antibiotic therapy. The delayed antibiotic therapy may lead to persistent bloodstream infection, which resulted in increased capillary permeability, escape of serum albumin, and shorten the half-time of albumin [30]. Low albumin levels may indicate severe condition and poor outcomes [31]. Moreover, our results indicated that patients in delayed therapy group had signi cantly higher proportion of empiric third-generation cephalosporin administration prior to blood culture than that in early therapy groups. The explanation may as the following. The third-generation cephalosporin is one of the most recommended empiric broad-spectrum antibiotic therapies for patients with nosocomial infection [32].
However, with increased of third-generation resistant K. pneumoniae isolates [2], empirical thirdgeneration cephalosporin administration may result in delaying appropriate antibiotic therapy. K. pneumoniae is a major worldwide source and shuttle for antibiotic resistance [2], and the nosocomial gram-negative bacteria bloodstream infection patients had higher proportion of inappropriate antibiotic therapy [33]. Therefore, it is very important for clinicians to evaluate whether the empiric antibiotic therapy is appropriate or not. More than half (38/67, 56.72%) of patients in our study had been empirically treated with carbapenem. And the prevalence of carbapenem-resistant K. pneumoniae in this study (7/67, 10.45%) was higher than that in many European countries according to the data from the European Centre for Disease Prevention and Control (website: http://atlas.ecdc.europa.eu/public/index.aspx?Instance=GeneralAtlas). We consumed that frequently using carbapenem may contribute to carbapenem-resistant K. pneumoniae isolate.
Appropriate antibiotic therapy can improve the clinical outcomes in children with severe bloodstream infection. However, to avoid the overuse or misuse of antibiotic, it is very important for clinician to recognize the bloodstream infection and identify the correct pathogen. In high-income countries, some rapid diagnostic testing technologies can help the clinician to identify K. pneumoniae quickly. However, in some low-income countries, the clinical experiences and education level of recognizing K. pneumoniae bloodstream infection may be more important. Furthermore, building susceptibility databases of K. pneumoniae isolates may help guiding clinicians to choose more appropriate and timely empiric antibiotic therapy.
This study has some limitations. Firstly, this is a single-center, retrospective study with relatively small sample size, and multi-center, larger sample size, prospective study is expected to strength the results of this study. Secondly, we only enrolled patients with nosocomial K. pneumoniae bloodstream infection, and this may in uence the extrapolation of our data to other populations. Thirdly, when applied our results to clinical practice, we should pay attention to the difference of de nitions of the start point of TTAT between us and other studies.

Conclusions
Our study demonstrated that TTAT may serve as an independent risk factor of septic shock and inhospital mortality in children with nosocomial K. pneumoniae bloodstream infection. The clinicians should initiate appropriate antibiotic within 10.7 hours of the onset of the K. pneumoniae bloodstream infection.

Declarations
Funding: This study was supported by the Science and Technology Department of Chongqing (cstc2018jscx-msybX0021).