Prevalence and Antimicrobial Resistance of Klebsiella pneumoniae Infections in Children: A Ten-year Retrospective Study at a Tertiary Hospital in Guangzhou, China

Abstract

Purpose

To investigate the clinical distribution and resistance trends of Klebsiella pneumoniae (KP) among pediatric patients at a tertiary pediatric hospital in Guangzhou, China (2015–2024), with a focus on extended-spectrum β-lactamase-producing (ESBL-KP) and carbapenem-resistant (CRKP) strains.

Patients and Methods

A retrospective analysis was performed on non-duplicate KP isolates obtained from pediatric patients (aged ≤14 years) at Guangdong Women and Children Hospital (2015–2024). Time trends of isolated rates and resistance rates for the KP strains were analyzed using Joinpoint regression, with statistical analysis performed in SPSS and Python.

Results

A total of 2907 non-duplicate KP isolates from 2557 patients were analyzed. The proportion of KP among all clinical isolates decreased significantly from 18.4% (492/2670) in 2015 to 6.5% (203/3135) in 2024 (P < 0.05), with the highest proportion (49.0%,1254/2258) observed in infants aged ≤1 month. The KP strains were most often cultured from lower respiratory tract specimens (70.8%, 2059/2907), followed by blood (10.0%, 292/2907) and urine specimens (7.8%, 227/2907). The strains showed two peaks of third- generation cephalosporin resistance in 2018 (64.3% [231/359] for ceftriaxone, 52.1% [215/498] for ceftazidime) and 2023 (42.0% [55/131] for ceftriaxone, 34.1% [45/132] for ceftazidime), and the highest resistance rates of carbapenems were 20.8% (97/466) in 2016. The proportion of ESBL-KP among all KP isolates decreased from 54.3% (266/490) to 29.1% (59/203) (P < 0.05). ESBL-KP isolates were found more frequently in blood and urine. The proportion of CRKP remained stable (7.5% [37/492] to 7.9% [16/203]; P > 0.05). CRKP was mainly found in ICU wards. CRKP rates dropped from 67.6% (25/37) to 0% (0/16) in the neonatal intensive care unit (NICU) but increased from 10.3% (4/39) to 50.0% (8/16) in the pediatric intensive care unit (PICU). Patients infected with ESBL-KP or CRKP were associated with significantly longer hospital stays compared to those caused by non-multidrug-resistant strains (P < 0.05).

Conclusion

Despite a decreasing overall prevalence of KP, the persistent high prevalence of ESBL-KP and the rapidly increasing trend of CRKP in the PICU suggested that recalibrated, unit-specific antimicrobial strategies were necessary. This study provides supporting information for the rational use of antimicrobial agents in clinical practice based on the comprehensive investigation of antimicrobial resistance change.

Introduction

Klebsiella pneumoniae (KP) is a formidable opportunistic pathogen responsible for a broad spectrum of healthcare-associated and community-acquired infections.^1–3 A major driver of its clinical impact is its remarkable capacity to enhance virulence through gain- or loss-of-function mutations, establishing it as a significant public health concern.^4–6 This threat is particularly acute in children, who face an elevated risk of severe outcomes, especially neonates and infants.^7,8 Recent systematic reviews have reported pooled mortality rates of 22.9% (95% CI 13.0%-32.9%) in carbapenem-resistant KP (CRKP) infections among neonates, significantly exceeding those of carbapenem-susceptible strains.⁹ Their vulnerability is compounded by a narrow selection of safe antibiotics and age-specific pharmacokinetics that collectively constrain therapeutic options, often leading to suboptimal outcomes, prolonged hospitalization, and substantial healthcare burdens.¹⁰

The clinical management of KP infections faces growing challenges due to rapidly expanding antimicrobial resistance.¹¹ CRKP and third-generation cephalosporin-resistant KP have been listed as Critical Priority pathogens by the World Health Organization in 2024.¹² Surveillance data from China further illustrate the escalating challenge. According to the 2024 China Antimicrobial Resistance Surveillance System (CARSS) report, KP was the second most frequently identified bacterial pathogen nationwide, underscoring its substantial clinical burden (https://www.carss.cn/). Data from CHINET (2015–2024) outline a rapidly evolving resistance landscape. The detection rate of CRKP increased sharply from 14.4% to 23.4%, whereas extended-spectrum β-lactamase-producing Klebsiella pneumoniae (ESBL-KP) was detected in 40.9–47.2% of all KP isolates during the same period (www.chinets.com). Particularly alarming are the findings in the pediatric population, with CHINET data (2015–2021) showing CRKP and ESBL-KP detection rates of 22.6% and 66.7%, respectively.¹³ This troubling national profile is further exacerbated in Guangdong Province. Our study indicates that the CRKP detection rate rose from 6.1% in 2019 to 11.5% in 2023, exceeding the national average, while resistance to third-generation cephalosporins also remained considerably higher than the national level (https://www.carss.cn/). It is noteworthy that resistance to third-generation cephalosporins in KP may also arise from non-ESBL mechanisms, including hyperproduction of AmpC β-lactamases, porin loss (OmpK35/36), or efflux pump overexpression, which are clinically significant yet often underreported in routine surveillance.^14,15

Although the epidemiology and antimicrobial resistance profiles of KP have been extensively investigated in adult populations, data specifically focusing on pediatric patients, particularly neonates and infants, remain limited. Existing pediatric studies have primarily focused on short-term periods and specific patient populations, such as ICU cohorts or those with particular infection sites.^16–18 However, long-term epidemiological surveillance data in the general pediatric population are limited. Therefore, we performed a decade-long (2015–2024), single-center analysis of 2907 KP isolates from a major pediatric hospital in Guangzhou. This study details the clinical distribution and antimicrobial resistance profiles of ESBL-KP and CRKP, with a focus on their dynamics in the neonatal intensive care unit (NICU) and the pediatric intensive care unit (PICU), to provide a reference for hospital infection control and the rational use of antibiotics.

Materials and Methods

Bacterial Isolates and Identification

This retrospective study was conducted at Guangdong Women and Children Hospital. A total of 2907 non-duplicate KP isolates were collected from 2557 pediatric patients (age ≤14 years) between 2015 and 2024, with 350 patients (13.7%) yielding multiple isolates from distinct specimen sites. The selection of KP strains for analysis was restricted to those exhibiting clinical significance. Isolates from sterile sites were considered infectious. For isolates originating from non-sterile sites, infection was defined by the presence of compatible clinical symptoms, supportive laboratory or radiographic evidence, and clinical response to antibiotic therapy.¹⁹ Isolates classified as colonization were excluded from the study. To avoid duplication and phenotypic changes, only the first isolate from each distinct specimen site per patient was retained for analysis. The lower respiratory tract specimens comprised sputum, bronchoalveolar lavage fluid and endotracheal aspirate. Bacterial identification was performed using MALDI-TOF MS (Bruker).

Antimicrobial Susceptibility Test

Bacterial identification was performed using the VITEK 2 system (bioMérieux, France) prior to 2018 and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS, Bruker Daltonics, Germany) thereafter. Antimicrobial susceptibility testing (AST) was primarily performed using the VITEK 2 AST-GN card, with the Kirby–Bauer disk diffusion method (Oxoid, UK) employed for confirmation or supplementary testing when indicated. The antimicrobial susceptibility of KP isolates was determined according to the guidelines of the Clinical and Laboratory Standards Institute (CLSI).^20–22 ESBL production was also determined using the VITEK 2 AST-GN card following CLSI.^20–22

Definition

ESBL represents a group of beta-lactamases that could hydrolyze various types of beta-lactam antibiotics, including expanded-spectrum (or third-generation) cephalosporins, commonly cefotaxime or ceftriaxone. CRKP was defined as resistance to imipenem, meropenem, or ertapenem.

Data Process and Statistical Analysis

The subsequent data analysis was performed using Python (CRNI, USA), following the exclusion of duplicate isolates from the same patient and anatomical site. Statistical significance was assessed using SPSS 26.0 (SPSS Inc., USA) with the Chi-square test or Fisher’s exact test. A two-sided P <0.05 was considered to indicate statistical significance. Time trends of isolated rates and resistance rates for the KP strains were analysed using Joinpoint software (NCI, USA), which allows for the identification of up to two joinpoints. The quantification of trends was conducted through the utilisation of annual percent change (APC), with the designation of “increase” or “decrease” attributed to APC slopes that exhibited a significant deviation from zero at a 0.05 level of significance. It is acknowledged that the statistical power of the analyses may have been compromised by the limited sample size in some subgroups. Consequently, findings derived from small-sample comparisons should be interpreted with caution, acknowledging the possibility of Type II error (failure to detect a true difference). The GraphPad Prism 9.4.0 software (GraphPad Software, La Jolla, CA, USA) was used for mapping.

Results

Epidemiological and Clinical Characteristics of KP Isolates

In this 10-year retrospective study (2015–2024), we analyzed 2907 non-duplicate KP isolates obtained from 2557 pediatric patients. The proportion of KP among all clinically significant bacterial isolates declined significantly, from 18.4% (492/2670) in 2015 to 6.5% (203/3135) in 2024 (APC = –13.02, P < 0.05) (Table 1).

Table 1.

Proportion of KP Among All Clinical Isolates from Pediatric Patients, 2015–2024

Collecting Year	No. of Isolates	No. of KP	Proportion of KP (%)
2015	2670	492	18.4
2016	2763	466	16.9
2017	3162	498	15.8
2018	3080	359	11.7
2019	2807	309	11.0
2020	1620	150	9.3
2021	2174	151	7.0
2022	2171	147	6.8
2023	2370	132	5.6
2024	3135	203	6.5
Total	25952	2907	11.2

As detailed in Table 2, the patient cohort exhibited a marked male predominance (male-to-female ratio approximately 2:1) and a striking concentration in early infancy, with infants aged ≤1 month accounting for 49.0% (1254/2557) of all cases. The NICU and the Pediatrics were the predominant departments for positive cultures, accounting for 36.1% (922/2557) and 33.9% (868/2557) of cases, respectively.

Table 2.

Epidemiological and Clinical Characteristics of Pediatric Patients Infected with KP (n=2557)

Characteristics	No.	Proportion (%)
Age groups
≤1m	1254	49.0
1–6m	905	35.4
6–12m	267	10.4
>1	131	5.1
Gender
Male	1681	65.7
Female	876	34.3
Department
Pediatrics	868	33.9
Pediatric Surgery	166	6.5
NICU	922	36.1
Neonatal Surgery	172	6.7
PICU	156	6.1
CICU	111	4.3
Others	162	6.3

Abbreviations: No, Number; NICU, neonatal intensive care unit; PICU, pediatric intensive care unit; CICU, cardiac intensive care unit.

Regarding specimen distribution, the vast majority of isolates (70.8%; 2059/2907) originated from lower respiratory tract specimens, followed by blood (10.0%; 292/2907) and urine (7.8%; 227/2907) (Figure 1A). A longitudinal analysis of specimen sources revealed notable shifts over the decade (Figure 1B). The proportion of isolates from lower respiratory tract specimens demonstrated a pronounced and consistent decreasing trend, from 80.1% (394/492) in 2015 to 48.8% (99/203) in 2024. The proportions from both blood and urine specimens showed an increase during the same period.

Figure 1.

Distribution and Temporal Trends of Specimen Sources for KP among Pediatric Patients (2015–2024). (A) Distribution of clinical specimen sources for KP isolates (n=2907). (B) Temporal trends in specimen source distribution for KP isolates (n=2907).

Antimicrobial Susceptibility

The Dynamics of the Resistance Rates of KP to Different Antimicrobial Agents in Pediatric Patients

KP showed considerable heterogeneity in antibiotic resistance rates across antimicrobial agents, ranging from 0.4% to 64.3%. Among these, resistance to imipenem, levofloxacin, and amikacin was below 10%, and to ertapenem and piperacillin/tazobactam below 20%, while ceftriaxone exceeded 50% (Table 3). From 2015 to 2024, the levofloxacin resistance rate increased (0.8% [4/492] in 2015 to 15.8% [32/202] in 2024; APC = 41.99; P < 0.05). The resistance rate to ceftriaxone decreased significantly (62.2% [306/492] in 2015 to 40.4% [82/203] in 2024; APC = −5.72, P < 0.05). In addition, trimethoprim/sulfamethoxazole resistance showed a biphasic pattern, with an initial increase from 32.4% (159/491) in 2015 to 50.6% (252/498) in 2017 (APC = 29.30, P < 0.05), followed by a decrease to 27.1% (55/203) in 2024 (APC = −7.81, P < 0.05) (Table 3 and Figure 2).

Table 3.

Susceptibility and Resistance Rates (%) of KP to Antimicrobial Drugs in Pediatric Patients from 2015 to 2024

Antimicrobial Drugs	2015		2016		2017		2018		2019		2020		2021		2022		2023		2024		Total
	n=492		n=466		n=498		n=359		n=309		n=150		n=151		n=147		n=132		n=203		n=2907
	S	R	S	R	S	R	S	R	S	R	S	R	S	R	S	R	S	R	S	R	S	R
IPM	91.1	7.5	79.0	20.8	93.8	5.0	85.0	13.6	92.2	5.2	88.7	9.3	92.7	3.3	91.2	5.4	87.9	11.4	91.1	7.9	88.8	9.7
ETP	91.3	8.7	77.5	22.5	93.8	6.0	86.4	13.4	90.0	7.8	90.0	9.3	96.0	4.0	92.5	7.5	88.5	11.5	92.0	8.0	88.9	10.7
CAZ	57.9	38.6	60.5	38.0	52.2	43.2	42.6	52.1	49.7	42.2	59.3	36.0	80.8	16.6	78.2	19.7	64.4	34.1	68.0	29.1	57.9	38.2
CRO	36.4	62.2	45.1	54.7	38.6	59.6	32.6	64.3	40.7	59.3	54.4	45.6	59.6	40.4	62.6	37.4	58.0	42.0	59.6	40.4	44.2	54.8
FEP	55.1	39.0	60.7	34.5	55.6	38.6	53.8	44.8	55.2	41.6	65.8	32.9	83.4	15.2	75.5	18.4	74.0	23.7	72.9	26.6	61.1	35.1
TZP	90.2	8.9	73.1	24.3	82.1	10.0	66.9	22.6	67.5	21.1	81.3	17.3	96.0	4.0	87.8	9.5	78.6	17.6	75.4	20.7	79.0	16.0
SXT	67.6	32.4	68.9	31.1	49.4	50.6	46.1	53.1	59.1	39.3	61.7	37.6	69.5	30.5	63.9	36.1	67.4	29.5	70.9	27.1	61.0	38.4
AMK	99.6	0.4	97.9	2.1	99.4	0.6	98.1	1.9	97.1	2.6	92.0	8.0	99.3	0.7	98.0	2.0	93.9	6.1	95.5	4.5	97.8	2.2
LVX	98.6	0.8	96.8	1.3	97.4	2.0	94.7	2.8	78.9	7.5	52.7	12.0	51.0	13.2	57.8	12.2	50.0	14.4	56.4	15.8	83.5	5.5

Abbreviations: n, number; R, drug resistance rate (%); S, drug sensitive rate (%); CAZ, ceftazidime; CRO, ceftriaxone; FEP, cefepime; SXT, trimethoprim/sulfamethoxazole; IPM, imipenem; ETP, ertapenem; TZP, piperacillin/tazobactam; LVX, levofloxacin; AMK, amikacin.

Figure 2.

Antimicrobial resistance rates of KP isolates in Pediatric Patients from 2015 to 2024 (n=2907).

Antimicrobial Resistance Rates of KP from Different Clinical Specimens

To investigate potential differences in drug resistance across different sample sources, we analyzed KP isolated from lower respiratory tract specimens, urine, and blood. As shown in Table 4, significant differences in resistance rates to almost all the antimicrobial agents were observed between the sample sources. KP isolates from urine and blood samples showed greater rates of resistance. Of these, ESBL-KP isolates were found more frequently from blood (58.5%, 168/287) and urine (52.0%,117/225), KP isolates from blood specimens exhibited the highest resistance rates to ceftriaxone (68.6%, 197/287) and trimethoprim/sulfamethoxazole (51.2%, 147/287), and the detection rate of ESBL-KP isolated from blood specimens was found to be the highest (58.5%, 168/287). KP isolates from urine specimens exhibited the highest resistance rates to ceftazidime (58.0%,131/226), cefepime (47.3%, 107/226), levofloxacin (17.3%, 39/226), and piperacillin/tazobactam (26.1%, 59/226). No statistically significant differences were observed in resistance rates to imipenem, ertapenem, or amikacin among KP isolates from different specimen sources (P > 0.05).

Table 4.

Comparison of Antimicrobial Resistance Rates of KP Isolates from Different Clinical Specimen Sources (%)

Antimicrobial Drugs	Lower Respiratory Tract Specimens	Blood	Urine	P-value
	n=2059	n=287	n=225
ESBL	44.1	58.5	52.0	<0.001*
CRO	54.7	68.6	67.3	<0.001*
CAZ	37.5	48.1	58.0	<0.001*
FEP	34.7	44.6	47.3	<0.001*
IPM	10.2	10.1	10.2	1.000
ETP	11.5	10.1	13.5	0.517
AMK	2.2	1.7	2.7	0.783
LVX	4.6	7.3	17.3	<0.001*
SXT	37.6	51.2	47.8	<0.001*
TZP	15.4	18.5	26.1	<0.001*

Notes: The resistance rates were statistically analyzed, * represents P< 0.05.

Detection Rates and Departmental Distribution of ESBL-KP and CRKP between 2015 and 2024

While the overall hospital detection rate of ESBL-KP declined from 52.9% (162/306) in 2019 to 29.1% (59/203) in 2024 (APC = –6.74, P < 0.05). In contrast, the detection rate of CRKP fluctuated without a consistent monotonic trend during the study period (APC = –4.17, P > 0.05). The rate peaked at 20.8% (97/466) in 2016, reached its lowest point at 3.3% (5/151) in 2021, and was 7.9% (16/203) in 2024 (Figure 3A). The distribution of these resistant strains varied considerably across departments, the detection rates of ESBL-KP and CRKP were 54.9% (876/1596) and 13.7% (218/1596) in the ICU wards, and they were both higher than in non-ICU wards (31.1% [408/1311] and 7.3% [96/1311]) (Figures 3B and C). In the NICU, the proportion of ESBL-KP among all KP isolates decreased significantly from 53.8% (143/266) to 37.3% (22/59) (APC = –8.83, P < 0.05), and the proportion of CRKP fell markedly from 67.6% (25/37) to 0% (0/16) (APC = –9.90, P < 0.05). Within the Pediatric Surgery, the proportion of ESBL-KP increased from 1.5% (4/266) to 15.3% (9/59) (APC = 33.60, P < 0.05), while the proportion of CRKP rose sharply but transiently, peaking at 60.0% (3/5) in 2021. A concerning and sustained increase in the proportion of CRKP was observed in the PICU from 2018 onward, rising from 10.3% (4/39) to 50.0% (8/16) (APC = 26.42, P < 0.05). Detailed data supporting these trends are provided in Supplementary Tables 1 and 2.

Figure 3.

Temporal trends in detection rates and clinical distribution of ESBL-KP and CRKP among pediatric patients (2015–2024). (A) Temporal trends in detection rates of ESBL-KP (n=1284) and CRKP (n=282). (B) Distribution of ESBL-KP isolates across clinical departments (n=1284). (C) Distribution of CRKP isolates across clinical departments (n=282).

The Antimicrobial Drug Resistance Rates of ESBL-KP and CRKP

In this study, 98.4% (1263/1283) of the included ESBL-KP isolates were resistant to ceftriaxone. Resistance rates to ceftazidime, cefepime, and trimethoprim-sulfamethoxazole were similar, ranging from 58.6% (752/1283) to 59.5% (763/1283). In contrast, resistance to β-lactam/β-lactamase inhibitor combinations was low, with piperacillin/tazobactam resistance at 12.1% (155/1283). All evaluated carbapenems, fluoroquinolones, and aminoglycosides demonstrated resistance rates below 10% (Figure 4A). The resistance rate of CRKP to common antimicrobial agents is remarkably high (Figure 4B), including piperacillin/tazobactam, cephalosporins, and carbapenems, which ranged from 93.3% (251/269) to 100% (269/269). The resistance rate of CRKP to trimethoprim/sulfamethoxazole was 42.8% (115/269). The resistance rates of both fluoroquinolones and aminoglycosides were below 20%.

Figure 4.

Resistance rates of ESBL-KP and CRKP to antimicrobial drugs. (A) Resistance rate of ESBL-KP to antimicrobial drugs (n=1284). (B) Resistance rate of CRKP to antimicrobial drugs (n=282).

Association of ESBL-Production and Carbapenem Resistance with Clinical Outcomes

Clinical outcomes differed significantly among the ESBL-KP, CRKP, and non-MDR groups (Table 5). The median hospital length of stay was substantially longer for patients infected with ESBL-KP or CRKP (median [IQR] 25.76 [12.61–52.85] and 26.56 [13.04–51.39] days, respectively) compared to those with non-multidrug-resistant (non-MDR) KP infections (10.94 [6.90–22.95] days) (P < 0.001), with no significant difference observed between the two MDR KP groups.

Table 5.

Comparison of Clinical Outcomes Among Pediatric Patients with Non-MDR, ESBL-KP, and CRKP Infections

Clinical Outcome	Non-MDR KP (n=1045)	ESBL-KP (n=1033)	CRKP (n=250)	P- value	Pairwise Comparisons
Hospital length of stay
Median, days (IQR)	10.94 (6.90–22.95)	25.76 (12.61–52.85)	26.56 (13.04–51.39)	<0.001	S vs E: <0.001
					S vs C: <0.001
					E vs C: 1.000
Primary discharge outcome, n (%)				0.02
Clinical stability	854 (81.7%)	798 (77.3%)	191 (76.5%)
Composite Adverse Discharge Outcome	191 (18.3%)	235 (22.7%)	59 (23.6%)

Note: S represents Non-MDR KP; E represents ESBL-KP; C represents CRKP; Composite Adverse Discharge Outcome: A composite outcome including patients discharged against medical advice, transferred to another healthcare facility for continued treatment, or who died during hospitalization.

Regarding the primary discharge outcome, the rate of the composite adverse outcome was significantly higher in the ESBL-KP group (22.7% [235/1033]) and in the CRKP group (23.6% [59/250]) than in the non-MDR KP group (18.3% [191/1045]) (P = 0.02).

The specific components constituting composite adverse outcome are detailed in Supplementary Table 3.

Discussion

KP is recognized as one of the leading pathogens causing community-onset or hospital-acquired infections, which become invasive due to impaired host defenses or interrupted resident microbiota and persisting colonization.¹ The isolation of KP exhibited a gradual decline at Guangdong Women and Children Hospital since 2015 from our data. The majority of KP isolates in this study sourced from lower respiratory tract, bloodstream and urinary tract, and the three types of samples also dominated because of the common infection sites or tracts.^23–25 Regarding antimicrobial resistance, ESBL-KP and CRKP strains were prevalent, comprising 44.4% (1284/2889) and 9.7% (282/2907) of isolates, respectively. Furthermore, infections involving ESBL-KP and CRKP resulted in significantly longer hospital stays than those caused by non-MDR KP, highlighting an urgent need for healthcare resources and attention on KP-infections with antimicrobial resistance.

Pediatric infections caused by KP also present a great challenge for newborn healthcare based on the higher incidence reported in infants than in adults.^26,27 In consistence with our previous studies,^18,28 infants aged ≤1 year comprised the majority of cases in this study. Approximately half (49.0%, 1254/2557) of all pediatric KP infections involved neonates within 1 month of age, most of whom (36.1%, 922/2557) were managed in the NICU. It is notable that Guangdong Women and Children Hospital has served as the provincial center for neonatal critical care of Guangdong Province, and a significant number of neonatal patients, especially serious patients have come from different cities, which may be responsible for the high proportion of specimens originating from the NICU in our study. The elevated infection risk observed in these youngest patients may be attributed to the following factors: 1) Neonates are inherently vulnerable due to their immature immune systems and the lack of fully developed, protective commensal microbiome.²⁹ 2) This inherent vulnerability is likely to worsen under widespread exposure to broad-spectrum antimicrobials and invasive procedures, which may simultaneously disrupt microbial homeostasis and breaches anatomical integrity.^30–32 Therefore, rational use of antimicrobial agents as well as reasonable invasive procedures are of vital significance for preventing neonatal infections.

The findings of this study indicated that the antimicrobial resistance of KP isolates from our institution has shifted over the past decade. The resistance of KP to cephalosporins has shown a downward trend over the past decade, with transient peaks observed in 2018 and 2023. This finding is consistent with national surveillance data, although resistance rates at our centre were higher than the corresponding national averages. Conversely, the prevalence of ESBL-KP exhibited a significant decline, from 54.3% (266/490) in 2015 to 29.1% (59/203) in 2024, which aligns with the reports from CHINET (https://www.chinets.com). It is noteworthy that although 41.4% (531/1283) of ESBL-KP isolates showed susceptibility to ceftazidime in vitro, this susceptibility rate was lower than the national figure, indicating a higher level of local resistance. The resistance to ceftazidime ranged from 16.6% (25/151) to 52.1% (187/359), consistently lower than the resistance to ceftriaxone, which ranged from 37.4% (55/147) to 64.3% (231/359), which is consistent with the CHINET data.^13,16 This phenomenon may be suggested that CTX-M-15 extended-spectrum β-lactamase has a higher prevalence in this field, as this enzyme variant typically confers greater resistance to ceftriaxone than to ceftazidime.^13,33 Among the ESBL-KP isolates, resistance to ceftriaxone and ceftazidime were 98.4% (1263/1283) and 58.6% (752/1283), respectively; whereas resistance to piperacillin-tazobactam was 12.1%; carbapenem resistance remained low (<5%).In light of the consistent clinical evidence demonstrating inferior efficacy in comparison to carbapenems, particularly in the context of bacteremia, piperacillin-tazobactam is not recommended as a monotherapy for serious ESBL-KP infections, regardless of in vitro susceptibility.^34,35

Conversely, resistance rates to carbapenems were consistently low. However, an exception was observed in 2016, where a transient spike in imipenem resistance was recorded, reaching 20.8%. However, at the same period, carbapenem resistance rates at our center were similar to or lower than the national pediatric averages (12.7%–27.0%) reported by CHINET from 2015 to 2021.¹³ A multidisciplinary intervention was initiated in 2016 by the Hospital Infection Control Department, in collaboration with the Pharmacy and Clinical Microbiology Laboratory, to combat multidrug-resistant organisms and optimise the use of antimicrobial agents. Core strategies included active screening for multidrug-resistant organisms (MDROs) when patients admission, reinforced the management of personnel entering and exiting the ICU, performed strict and frequent hand hygiene management, and implemented stringent antimicrobial stewardship. Following the implementation of the protocol, a significant decrease in carbapenem resistance among KP isolates was observed; the rates were sustained at around 10%.

In this study, CRKP isolates demonstrated a relatively low level of resistance to amikacin (19.0%, 51/269) and levofloxacin (14.1%, 38/269). In contrast, resistance to β-lactams and β-lactamase inhibitor combinations exceeded 90%, with resistance to trimethoprim-sulfamethoxazole at 42.8% (115/269). This profile is consistent with reports that NDM (metallo-β-lactamases) produced KP are predominant among pediatric CRKP strains in China, which typically do not confer high-level resistance to aminoglycosides and quinolone antibiotics.^36,37 In order to verify the predominant carbapenemase types and thereby support the development of precise interventions, it is recommended that routine carbapenemase genotyping be implemented in this setting. It provides essential data not only for refining infection prevention measures but also for tracking local resistance patterns. Specifically, monitoring facilitates the assessment of the preserved susceptibility to agents. This finding is of mechanistic and surveillance importance despite its limited use in paediatrics, and informs the rational use of newer antimicrobial agents.

KP isolates from bloodstream and urinary tract infections exhibited significantly higher antimicrobial resistance rates compared to respiratory isolates, aligning with a previous study.³⁸ This may stem from distinct pathophysiological contexts. Bacteremia frequently represents a secondary event, wherein pathogens may have undergone prior antimicrobial selection and acquired resistance determinants at a primary site before hematogenous dissemination.^39,40 And the urinary tract, due to its relatively open anatomical architecture and stagnant fluid dynamics, presents a susceptible environment for bacterial colonization and persistent infection.^41,42 Furthermore, the urothelial milieu facilitates biofilm formation, which is a key virulence determinant that enhances microbial persistence, impairs antibiotic penetration, and promotes the emergence of resistant subpopulations.⁴³

The distribution of ESBL-KP and CRKP was not uniform across wards. While detected in multiple departments, these resistant strains were concentrated in critical care units, especially in the NICU. Reported by previous study, the key risk factors for CRKP infection, include immunosuppressive therapy, ICU admission, exposure to a broad range of antimicrobial drugs, surgical interventions, mechanical ventilation, central venous catheterization, and the presence of indwelling devices such as nasogastric tubes.⁴⁴ Over the past decade, the distribution of resistant KP strains across hospital departments has followed distinct trends. This evolving epidemiology likely reflects the ongoing clinical specialization within our institution, where focused subspecialty development may alter unit-specific patient populations. This finding highlights the importance of integrating both hospital-wide and unit-specific resistance patterns to inform empirical therapy for KP infections.

A longer length of hospital stays was observed in the pediatric patients infected with ESBL-KP or CRKP than those with non-multidrug-resistant (non-MDR) KP infections, this observation aligns with previous reports.^28,45 Benner et al reported from a PICU setting that ESBL-KP infections were linked to prolonged hospitalization, with a longer pre-infection hospital stay as a key risk factor.⁴⁵ As a regional referral centre for critically ill children, our hospital routinely admits patients with prolonged hospital stays and complex comorbidities, which are well-established risk factors for healthcare-associated multidrug-resistant infections. Moreover, infections with ESBL-KP or CRKP were associated with lower clinical cure rates at discharge compared to non-MDR KP infections in our study, while no significant difference was observed between the ESBL-KP and CRKP infected patients. The therapeutic management of pediatric infections is already constrained by the limited number of antimicrobials with established efficacy and safety profiles in children, and multidrug resistance further narrows the range of effective therapeutic options.¹⁰ Although new antibiotics,such as ceftazidime-avibactam, have been developed, the use of these new antibiotics in pediatrics continues was limited by insufficient pharmacokinetic, pharmacodynamic, and safety data specific to this population, Therefore, there is an urgent need for more comprehensive report data for pediatric medication use.⁴⁶

This study has several important limitations. First, as a single-center retrospective study, a degree of selection bias was likely present, which limits the generalizability of the findings to other settings. Second, the lack of integrated antimicrobial consumption data precludes an analysis of how resistance trends correlate with local prescribing patterns. Finally, the absence of carbapenemase genotyping and molecular epidemiological data restricts our understanding of the transmission dynamics and clonal relationships among the resistant strains. Despite these limitations, this analysis provides a valuable longitudinal perspective on KP resistance within a pediatric cohort at a major tertiary center, elucidating key temporal trends in resistance profiles and their clinical implications. Future, more multicenter studies that implement standardized methodologies and incorporate molecular surveillance are warranted. Multidisciplinary collaborations are essential to generate the robust evidence required to guide optimal clinical management and to inform effective prevention strategies against pediatric KP infections.

Conclusion

This retrospective study of data in the last decade showed that the proportion of KP among all clinically significant bacterial isolates declined significantly. KP resistance in our hospital varied significantly by specimen source and hospital department, with isolates from blood and urine showing higher resistance than those from the lower respiratory tract. Although ESBL-KP detection rates declined overall, their prevalence remained high in ICU wards, while CRKP increased sharply in the PICU. In the face of this challenge, targeted, unit-specific infection control interventions are needed. Resistance rates of CRKP to conventional antimicrobial agents were generally high, suggesting the implementation of routine carbapenemase genotyping to guide containment strategies. Children infected with resistant strains had longer hospital stays and lower cure rates, indicating the clinical burden of antimicrobial resistance. Future studies integrating antibiotic consumption data and molecular surveillance are needed to further inform local prevention efforts.

Funding Statement

This work was supported by Guangdong Medical Science Foundation (B2025291, to Yasha Luo), Guangdong Medical Science Foundation (B2025090, to Junfei Guo), Guangdong Provincial Administration of Traditional Chinese Medicine (20251044, to Weiming Lai).

Declaration Regarding AI Use

During the preparation of this work, the authors used DeepSeek-V3.2 (created by DeepSeek Company) for language polishing and grammar checking. The authors critically reviewed and edited all AI-assisted content. The authors take full responsibility for the entire content of this publication.

Data Sharing Statement

The data that support the findings are available from the corresponding author upon non-commercial request.

Ethics Approval

This study was approved by the Ethics Committee of Guangdong Women and Children Hospital (Ethical Approval Number: 20251008). Informed consent was not required as the information was retrospectively retrieved from medical records. All analyzed bacterial strains were obtained through routine hospital laboratory procedures, and this study did not involve any human genetic resources. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki.

Consent to Participants

The informed consent was waived by the Medical Ethics Committee of Guangdong Women and Children Hospital because of the retrospective nature of the study and all data used in this study were strictly anonymous.

Disclosure

The authors declare that they have no competing interests in publishing this paper.