Healthcare-associated Invasive Klebsiella Pneumoniae Infections in Chinese Pediatric Patients: Risk Factors, Outcomes and Antimicrobial Resistance


 Background:High prevalence of carbapenem-resistant Klebsiella pneumoniae (CRKP) has become a serious clinical concern in China. This study first described the risk factors, outcomes and antimicrobial resistance pattern of healthcare-associated KP infections in Chinese pediatric patients.Methods:This retrospective study was conducted in the nine tertiary hospitals during 2016-2018. The 324 pediatric inpatients who had KP isolated from blood and cerebrospinal fluid and had complete medical records reviewed were included. We analyzed the risk factors, outcomes and antimicrobial resistance pattern of KP-infected patients based on comparison between healthcare-associated KP infections (HAI) and community-acquired infections.Result:Of the 324 enrolled patients, 275 (84.9%) were clinically defined as HAI, including 175 (63.6%) neonates and 100 (36.4%) aged >28 days. The overall prevalence of CRKP was 35.8% (40.4% in HAI verse 10.2% in CAI, P <0.05). Prematurity (odds ratio (OR):37.07, 95% CI:8.29-165.84), hematologic malignancies (OR:15.52, 95% CI:1.89-127.14) and invasive mechanical ventilation (OR:13.09, 95% CI: 1.66-103.56) were independent risk factors for HAI. Patients from rural area (OR: 1.78, 95% CI:1.06-2.99), invasive mechanical ventilation (OR:1.79, 95% CI: 1.01-3.19), antibiotic therapy prior to admission (OR: 1.88, 95% CI:1.07-3.29) and prior hospital stay in the past 30 days (OR: 2.17, 95% CI:1.26-3.73) were associated with healthcare-associated CRKP infections. Organ dysfunction was independently correlated with poor outcomes (OR:2.92, 95% CI: 1.23-6.95)Conclusions:Pediatric invasive KP infections and high prevalence of CRKP infections largely occurred in health-care settings in China. The adequate and intensified infection control measures should be focused on high-risk hematologic patients, neonatal patients and intubated patients.


Introduction
Klebsiella pneumoniae (KP) has been implicated in severe infections and one of the most important causes of nosocomial infections (1). According to the global surveillance data, KP is the third leading bloodstream pathogen in children and the frequency of antibiotic resistant KP has been increasing over years (2)(3)(4). In recent years. Carbapenem-resistant Klebsiella pneumoniae (CRKP) represents a 'critical concern' of the World Health Organization and poses great challenges for clinical treatment (5,6). Of particular concern, the prevalence of CRKP has been sharply rising from 2.9%-3% in 2005 to 25%-26.3% in Page 4/21 Currently, KP infections become an important clinical issue in pediatrics, especially in neonates since very limited effective antibiotics are available for the treatment choice (8,9). The existing studies from other countries showed KP as a key opportunistic pathogen causing severe infections in hospitalized children (10)(11)(12). So far, surveillance study and clinical data on healthcare-associated KP infections are lacking in China although CRKP is now a pronounced problem. According to some regional surveillance studies, most of Carbapenem-resistant organisms (CRO) are healthcare-associated (10,13). Therefore, we carried out this retrospective multicenter study to understand the current situations of invasive healthcareassociated KP infections in Chinese pediatric patients, and analyzed risk factors, outcomes and antimicrobial resistance of healthcare-associated KP and CRKP infections, aiming to identify the main problem of invasive KP infections and help nd the effective solution to reduce and prevent healthcareassociated KP and CRKP infections in pediatric patients.

Method Study design and subjects
This is a retrospective multicenter study conducted between January 2016 and December 2018 at the 9 tertiary hospitals in China, including 4 provincial Children's Hospitals, 2 General Hospitals and 3 municipal Women and Children's Hospitals. These hospitals locate in different administrative regions of China: 5 hospitals in the Southeast, 3 in the South and 1 in Central, respectively. The inclusion criteria of eligible patients were: 1) hospitalized pediatric inpatients aged <18 years, 2) inpatients who had KP isolated from blood and or cerebrospinal uid (CSF) specimens. The patients were excluded if their clinical data collection were incomplete or their medical records were unavailable for review by the end of 2019.

De nitions
Healthcare-associated infections (HAI) were de ned as: 1) KP strain recovered from sterile specimens collected at least 48h after hospital admission in a patient; 2) KP culture positive within the rst 48 hours of admission in a patient with prior hospital stay for a few of consecutive days in the past 30 day. Community-acquired infections (CAI) were de ned as KP culture positive within 48 hours of admission in a patient without prior hospitalization within the past 30 days. Poor outcomes were considered if a patient died in hospital or was hopelessly discharged from hospital under the request of parents based on the medical records.

Data collection
The following data were collected, including: patient's demographics and important underlying diseases, previous history of antibiotic use and hospitalization within the 30 days prior to the admission date, ward unit during hospitalization, clinical diagnosis on admission and discharge, length of hospitalization, interval between the occurrence of KP infections (the collection date of the rst KP-positive sample ) and the date of admission, intravascular catheter use, mechanical ventilation, imaging ndings, clinical outcomes and microbiological data.

Microbiological study
The rst invasive KP isolate recovered from a patient was included to this study for microbiological analysis. Species identi cation and antimicrobial susceptibility testing were performed at local hospital laboratories by automated systems such as Vitek or Phoenix, according to the 2018 CLSI performance guideline (14). Antimicrobial susceptibility testing was performed for clinical isolates using minimum inhibitory concentrations (MICs) or Kirby-Bauer disk diffusion method. The original results were classi ed according to the breakpoints of the 2018 CLSI standards. For polymyxin B, the criteria referred to the epidemiological cutoff value of colistin in the CLSI le (MIC ≤2 μg/mL for wild strains; MIC ≥4 μg/mL for non-wild strains) (7).

Statistical analysis
Data was entered and analyzed into Excel version 2016 (Microsoft, Redmond, Washington) and were analyzed using SPSS (IBM Statistic 20.0). Results were described as absolute numbers and percentage or as median and interquartile ranges. Categorical variables were compared using the Chi-square or Fisher's exact tests. Continuous variables were compared by Student's t test or Mann-Whitney U test according to their distribution. The risk factors were analyzed by Binary regression analysis and Multinomial regression analysis, and the results were presented as odds ratios (ORs), 95% con dence intervals (CIs) and P values. Signi cant variables with P value of <0.2 were then selected for multiple regression analysis to evaluate risk factors for clinical poor outcomes. A difference with P <0.05 was considered to be statistically signi cant.

Result
The basic characteristics of KP-infected patients A total of 425 pediatric inpatients met the diagnosis criteria of invasive KP infections based on the hospital microbiology laboratory records, of whom, 248 (58.4%) neonates ≤ 28 days and 177 (41.6%) were children aged > 28 days on admission. Finally, 324 (76.2%) patients met the enrollment criteria and were included in this study and 101 (23.8%) were excluded due to unavailability of medical records (99 patients) and incomplete clinical data collection (2 patients).
Of the 324 enrolled patients, 299 (92.3%) had KP isolated from blood, 11 (3.4%) had KP isolated from CSF and 14 (4.3%) had KP isolated from both blood and CSF. On the clinical grounds and case de nition, 275 (84.9%) patients were considered as HAI and 49 (15.1%) patients were considered as CAI.
Clinical features and risk factors of healthcare-associated KP infections in pediatric patients.
Risk factors of poor outcomes in pediatric patients with healthcare-associated KP infections As shown in Table 5, univariate analysis showed that invasive mechanical ventilation, any organ dysfunction and septic shock were the signi cant risk factors for poor outcomes. Multiple regression analysis showed that organ dysfunction was an independent risk factor for the poor outcomes (OR:2.92, [95% CI: 1.23-6.95], P < 0.05).
Antimicrobial susceptibility patterns of healthcare-associated KP isolates As shown in Table 6, healthcare-associated KP strains showed high frequency of resistance to carbapenem as well as other clinical important antibiotics usually recommended for the treatment of Enterobacteriaceae infections, such as third-generation cephalosporins, cefepime and piperacillin/tazobactam. Also, healthcare-associated KP strains showed signi cantly higher frequency of resistance to clinical important antibiotics than community-acquired KP stains. Besides, healthcareassociated KP strains showed relatively higher resistance percentage to tigecycline, amikacin, cipro oxacin and levo oxacin (12.5%-23.7%). Healthcare-associated KP strains almost remained sensitive to polymyxin B.

Discussion
This multicenter retrospective study rst revealed that most episodes of invasive KP infections were healthcare-associated in pediatric patients in China. We further demonstrated that the high percentage of CRKP infections usually occurred in healthcare setting (40.4%) rather than in community setting (10.2%). Of particularly concern, 58.9% of pediatric invasive healthcare-associated KP infections occurred in neonates. KP infections impose a health threat to hospitalized neonates and children.
According the surveillance in Chinese adults, KP was the second common pathogen in hospitalassociated bloodstream infections (15). Our recent study revealed that KP was the second common pathogen of bloodstream bacterial infections in Chinese pediatric patients (4). Our present study further showed that 84.9% of invasive KP infections were hospital-associated. Thus, it is feasible to prevent and reduce invasive KP infections through a bundle of effective infection prevention and control (IPC) measures. Identifying some key risk factors is essential to formulate the targeted measures. In this study, hematologic malignancies, prematurity, and invasive mechanical ventilation were identi ed as the independent risk factors for HAI. The available evidences have demonstrated that IPC practices and procedures can effectively prevent the occurrence and the spread of HAI in healthcare facilities (16, 17). Thus, reinforcement of IPC at high-risk pediatric units such as hematological units, pediatric and neonatal intensive care units is key to prevent invasive healthcare-associated KP infections. Healthcare-associated KP infections in preterm neonates is of particular concern because 58.9% of invasive healthcareassociated KP occurred in neonate and the odds ratio for prematurity is highest. Prematurity is a common risk factor of any nosocomial infections in neonatal units (18,19). Our results further support this nding and emphasize prematurity as athe most key risk factor of severe invasive nosocomial KP infections. Therefore, hospitalized preterm neonates should be placed on the highest priority group for whom IPC practice should be strengthened and improved. Besides, we noticed that 43.3% of children with hospital-associated KP infections had accompanying pneumonia and 34.9% of patients received mechanical ventilation therapy. Thus, part episodes of invasive healthcare-associated KP infections were likely to be secondary to ventilation-associated pneumonia.
We noticed that the high prevalence of CRKP in hospital setting but low prevalence of CRKP in community setting in pediatric patients. CRKP indeed represents a clinical and economic impact on pediatric patients and tertiary pediatric hospital in China. In this study, previous antibiotic therapy prior to hospitalization, previous mechanical ventilation, and prior hospital stay within the past 30 days were independent risk factors for acquisition of healthcare-associated CRKP infections, which is consistent with most of previous studies conducted in children and adults (20)(21)(22). In addition, we found that children from rural area was an independent risk factor for invasive healthcare-associated CRKP infections in Chinese children. We reasoned that rural children are likely to develop severe diseases and usually at higher risk of acquiring nosocomial infections. Unlike most previous single-center or speci cunit studies (22,23), we didn't nd that patients with hematologic malignancies were an independent risk factor of invasive CRKP infections. This difference is possibly owing to the selected bias of study hospitals and enrolled patients.
Although a few of studies suggested HAI was a risk factor for mortality of KP infections based on univariate analyses (24-26), we found no statistical difference in clinical outcomes of invasive KP infections between HAI and CAI patients. This discrepancy is probably owing to loss to follow-up in discharged patients of this cohort. However, HAI patients were more likely to develop organ dysfunction, which was the only independent risk factor of poor outcomes. Most episodes (79.3%) of invasive healthcare-associated KP infections are accompanied by focal organ involvement, of which, pneumonia (43.3%) and meningitis (23.3%) were the common manifestations, furthermore, KP-associated meningitis almost occurred in HAI patients. These ndings add our understanding of clinical presentation of invasive healthcare-associated KP infections, which will be helpful to predict the possible source and secondary complications of invasive KP infections and select appropriate antibiotic regimen. In Africa countries, the mortality rate of bloodstream KP infections in pediatric patients was reported to reach as high as 21.6%-56.5% (8, 11, 27), signi cantly higher than our rough estimation (13.6%) in Chinese children. This could re ect the regional difference in accessibility of tertiary pediatric medical resource and capacity of pediatric critical intensive care for pediatric patients. On the other hand, virulence factors of endemic KP isolates are also a contributing factor of the fatal outcomes (28). We need to further characterize the virulence pro le of KP strains in the next step to understand KP pathogenicity well.
Of particular concern is that healthcare-associated KP strains displayed higher prevalence of resistance to the tested antibiotics than community-acquired KP isolates, especially to extended-spectrum cephalosporin and carbapenem. The frequency of ceftriaxone-resistant healthcare-associated KP and community-acquired KP was 70.2% and 13.0%, respectively, and the frequency of meropenem-resistant healthcare-associated KP and meropenem-resistant community-acquired KP was 45.4% and 9.4%, respectively. The high prevalence of KP resistance in hospital setting makes it di cult to empirically and de nitely select antibiotic therapy for invasive healthcare-associated KP infections in pediatric patients because the prescription of polymyxin B and tigecycline ,which are almost sensitive to KP, are strictly restricted to use in children due to safety consideration, especially in neonates (29).

Our study rst described important pro les of healthcare-associated KP infections in Chinese pediatric patients and add some knowledge and understanding of invasive KP and CRKP infections in children.
Although this study is a retrospective study, the enrolled patients from all pediatric units of the 9 tertiary children's hospitals are representative and all episodes of invasive KP infections are culture-proven. However, our study has two major limitations. Firstly, retrospective data collection probably resulted in clinical information missing, such as prior healthcare history, prior previous antibiotic therapy and PICC utilization. Second, there is a potential estimation bias of clinical outcomes because KP-associated death post discharge was not captured. Further prospective studies are needed to trace the potential transmission source of KP in hospital setting and explain the disease severity based on integration clinical data with molecular mechanisms of carbapenemases and virulence factors of KP strains. The efforts will help to take intensi ed and effective IPC measures to reduce healthcare-associated KP infections and CRKP infections in pediatric patients.

Declarations
Acknowledgment: We thank Prof. Hong Zhao, the secretary of China Society of Infectious Diseases, for her help with the program initiation. We thank the members of the collaborative working group of pediatric subgroups of China Society of Infectious Diseases for their collecting the data for this study.

Funding Source:
This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors.

Con ict of Interest:
All authors declared no potential con icts of interest to disclose.

Ethical approval:
This study was reviewed and approved by the Ethics Committee of Fudan Children's Hospital. Informed consent from patients was not required by the Ethics Committee, because there was no contact with patients and all data were deidenti ed Authors' contribution: Yue Qiu designed the study and data collection instruments, collected data, carried out the initial analyses, drafted the initial manuscript, and revised the manuscript. Yi Xu; Daojiong Lin; Fang Wang; Yibing Cheng; Qingxiong Zhu; Chunhui Zhu; Chaomin Wan; Yu Zhu; Jianning Tong; Rui Li; Qionghua Zhou; Minxia Chen; Qingwen Shan; Zhiqiang Zhuo; Caihong Wang; Shiyong Zhao; Wen Song participated in study design nalization, coordinated and supervised data collection, and reviewed manuscript. Yi Xu and Daojiong Lin also participated in data analysis and summary. Mei Zeng conceptualized and designed the study and data collection instruments, reviewed and revised the manuscript for important intellectual content.
All authors have read and agreed to the nal draft before submission.