Resistance profile and influence factors of carbapenem-resistant Klebsiella pneumoniae (CRKP) causing infections in China: a systematic review and meta-analysis

Abstract

The prevalence of carbapenem-resistant Klebsiella pneumoniae (CRKP) infections has surged in China over the past decade, posing a significant public health concern. However, comprehensive data on CRKP antimicrobial resistance patterns and the impact of the COVID-19 pandemic on these patterns in China remain unclear. We conducted a systematic review of CRKP infections in China, utilizing data from PubMed spanning 2006 to July 2023. We focused on resistance rates of CRKP causing infections, examining variations across time, regions, and age groups, as well as factors contributing to antimicrobial resistance. Our analysis included 68 studies from 19 provinces in China, comprising 1,284 CRKP isolates obtained from 779 patients. The overall mortality rate for CRKP infections in China was 27% (95% CI: 0.14–0.41, I² = 73%, k = 47), with ST11 being the predominant sequence type (Pooled Rate: 80%, 95% CI: 0.67–0.90, I² = 86%, k = 31). Temporal and spatial analyses indicated increased resistance to ciprofloxacin (Random effects model: Q_b = 9.88, df = 1, P < 0.010) and levofloxacin (Random effects model: Q_b = 7.69, df = 1, P < 0.010) during the COVID-19 pandemic. Resistance to chloramphenicol (Random effects model: Q_b = 4.97, df = 1, P = 0.030) and ceftazidime-avibactam (Random effects model: Q_b = 8.58, df = 1, P < 0.010) was lower in southern regions, while tetracycline resistance (Random effects model: Q_b = 9.69, df = 1, P < 0.010) was lower in the north. Higher resistance rates were observed in adults and the elderly. Age and geographic location were key factors associated with antimicrobial resistance. Fourteen out of thirty-five drugs showed a positive correlation with mortality rates, emphasizing their significant impact on CRKP infection mortality. This study underscores the need for targeted interventions to address regional and age-related variations in CRKP resistance and highlights the critical role of antimicrobial resistance in influencing mortality outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12941-025-00827-2.

Introduction

Klebsiella pneumoniae, a significant opportunistic Gram-negative bacillus among the ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, K. pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter species), is closely associated with infections such as urinary tract infections, lower respiratory tract infections, and bacteremia [1]. The rise of hypervirulent and carbapenem-resistant strains has made K. pneumoniae a significant global public health concern [2].

In 2024, the World Health Organization updated and refined the List of Priority Pathogens, emphasizing the urgency of addressing antibiotic-resistant bacterial threats. Among these, carbapenem-resistant Enterobacteriaceae (CRE), including carbapenem-resistant K. pneumoniae (CRKP), has been classified as a critical priority pathogen [3]. CRKP was first identified in the United States in 1996 [4]. Eight years later, bla_KPC-2-positive K. pneumoniae was reported in Zhejiang, China, marking the first case of CRKP in the country [5].

Since then, CRKP has become increasingly prevalent in China. Data from China Antimicrobial Surveillance Network (CHINET) further highlight this trend, with the resistance rate of K. pneumoniae to carbapenems increased from 3% in 2005 to approximately 26% in 2023 [6]. The steady rise in CRKP resistance rates poses challenges to clinical treatment, threatens public health safety, and imposes substantial economic burdens [7].

CRKP has attracted widespread attention due to the high mortality rate it causes. CRKP not only causes hospital-acquired pneumonia and urinary tract infections but is also the leading pathogen of bloodstream infections. A review reported that among 2,462 infected patients worldwide, the pooled mortality rate was as high as 42.14% [8], with higher mortality rates observed in immunocompromised and critically ill patients [9].

In low- and middle-income countries, the mortality rate of neonatal infections caused by CRKP reaches 22.9% [10], posing a substantial threat to neonates by complicating treatment due to resistance and exacerbating economic burdens [11]. Furthermore, the mortality rate of CRKP infections is notably high among the elderly, a population largely impacted by risk factors such as diabetes and malignancies [2].

In China, due to limited treatment options, the incidence of CRKP bloodstream infections (CRKP-BSI) and associated mortality rates increased from 20.69% in 2012 to 37.40% in 2019 [12]. These serious outcomes highlight the urgent need for enhanced surveillance and treatment strategies.

China has a large land area with significant geographical differences. Thus, the distribution of CRKP varies across different regions. It was reported that CRKP strains in different regions of Shanxi province, China varied in sequence types, carbapenemases, serotypes, and virulence genes [13].

Whereas in Jiangxi province, the resistance rate of CRKP strains carrying the bla_KPC gene decreased from south to north, the resistance rate of strains carrying the bla_NDM gene, which spread from the central area to peripheral areas, exhibited a different geographical distribution pattern [14].

During the COVID-19 pandemic, measures such as restricted movement, enhanced hand hygiene, and the use of surgical masks limited the transmission of respiratory bacteria, leading to a reduced prevalence of CRKP [15–17].

The COVID-19 pandemic has potentially influenced antibiotic usage patterns and bacterial resistance. A study in South Korea indicated that the use of carbapenem antibiotics in ICU increased during the pandemic, with a concomitant significant rise in the detection rates of multidrug-resistant bacteria [18].

A retrospective study in Slovakia examined the prevalence of CRKP before and during the pandemic, revealing a fourfold increase in the relative prevalence of CRKP during the COVID-19 period [19]. The high usage of antibacterial agents among COVID-19 patients has likely exacerbated bacterial resistance, contributing to the rapid increase in multidrug-resistant bacteria and potentially accelerating the evolution of CRKP resistance [20].

Current descriptions of CRKP epidemiological trends in China have temporal and regional limitations, and there is a lack of comprehensive analysis of changes in CRKP resistance patterns. Wang et al. found that ST11 is the most dominant sequence type of CRKP in China, which is consistent with our results [21]. The authors analyzed the geographical spread trend of ST11 CRKP in China and speculated that the high prevalence of high-risk ST11 KL64 CRKP subclones is related to genetic factors. However, there is a lack of risk factor analysis for CRKP resistance rates.

To address these gaps, our study aims to conduct a systematic review and meta-analysis of the epidemiological data, mortality rates, antimicrobial resistance profiles, regional distribution, and temporal trends of CRKP infections in China over the past two decades, with a focus on the analysis of the factors influencing these patterns.

Methods

Literature search strategy and study selection

This study was conducted in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines [22], with the protocol registered in PROSPERO (ID: CRD42024512898).

We conducted a comprehensive search of the PubMed database for studies published from 2006 to July 2023 that clearly identified CRKP infections in China. The search strategy included (("Carbapenem-resistant"[tw] OR "Carbapenemase-Producing Enterobacteriaceae"[tw] OR "Carbapenem-Resistant Enterobacteriaceae"[Mesh]) AND ("Klebsiella pneumoniae"[tw] OR "Klebsiella aerogenes" [tw] OR "Bacillus pneumoniae"[tw] OR "Klebsiella rhinoscleromatis"[tw] OR "K. pneumoniae"[tw] OR "Klebsiella pneumoniae"[Mesh]) OR "CRKP"[tw]) AND ("China"[tw] OR "People's Republic of China"[tw] OR "Mainland China"[tw] OR "China"[Mesh]). We limited the results to articles with full-text availability but without restrictions on language.

Studies were included if they provided comprehensive antimicrobial susceptibility data for CRKP infections in China, with no restrictions on patient age or infection type. The included study designs were: randomized controlled trials, case–control studies, surveillance studies, cross-sectional studies, retrospective/prospective cohort studies, and case reports.

Exclusion criteria comprised studies focusing on CRKP colonization rather than infection, as well as those involving CRKP strains from animals or environmental sources, to ensure that our analysis was centered on patient-related infection data. Studies that lacked independent CRKP data were also excluded.

To thoroughly investigate the geographical distribution and variations of CRKP in China, we excluded non-domestic cases and studies that did not clearly specify the geographic location or province. To ensure consistency in results, we used the Clinical and Laboratory Standards Institute (CLSI M100-S33) [23] to interpret the antimicrobial susceptibility data and assess CRKP antimicrobial resistance rates. Studies without specific minimum inhibitory concentration (MIC) values for CRKP or without susceptibility results were excluded. Finally, publication types such as reviews, letters to the editor, and systematic reviews were also excluded.

Study outcomes

The primary outcome of this study was the antimicrobial susceptibility profiles of CRKP in China. Specifically, we analyzed resistance rates for each antibiotic, temporal trends, regional variations, and age-stratified differences. Secondary outcomes included the isolation rate of CRKP (the proportion of CRKP among all K. pneumoniae isolates), mortality rates associated with CRKP infection and the correlation between these mortality rates and antimicrobial resistance.

Data extraction and management

The retrieved literature was initially screened by reviewing titles, abstracts, and keywords to retain studies related to CRKP infections in China. Full texts of the selected studies were obtained and imported into the EndNote reference management software.

A full-text screening was then conducted based on the inclusion and exclusion criteria, followed by data extraction. This task was independently completed by two members from different groups, with any discrepancies resolved by a third reviewer.

For studies that met the inclusion criteria, we used a standardized data extraction form to collect the following information:

• Title and year of publication

• Participants and study design

• Province and sample collection time

• Sample type, patient age and gender

• Infection type and clinical outcomes

• Number/ratio of CRKP strains

• Sequence type and antimicrobial resistance genes (ARGs), including carbapenem resistance genes

• Antimicrobial susceptibility testing methods and susceptibility interpretation standards

• Minimum inhibitory concentration (MIC) values

In cases of missing data, we attempted to contact the study authors or utilized available summary statistics whenever possible.

Quality assessment of included studies

The quality of included case reports, cross-sectional studies, cohort studies, and case-control studies was assessed using the AHRQ tool [24] which is specifically designed for these study types.

The tool consists of nine items, each rated as low risk, moderate risk, or high risk based on the descriptions in the original studies. Two researchers independently performed the quality assessments, with any discrepancies resolved by a third reviewer.

Publication bias was evaluated using funnel plots for antibiotics with at least 10 studies, with asymmetry assessed visually. Due to the limited number of studies available for certain antibiotics, publication bias assessments were not applied, as these methods have low statistical in small-sample meta-analyses.

Instead, sensitivity analyses were conducted by excluding studies that did not clearly identify human-derived isolates, in order to evaluate the robustness of the results.

Statistical analysis

To examine regional differences, we extracted raw data from the included studies and performed a subgroup analysis of CRKP antimicrobial resistance by dividing the regions into northern and southern China based on the Qinling-Huaihe Line. Temporal differences were analyzed by categorizing data into three periods: before December 2019, during December 2019 to December 2022, and after December 2022, capturing variations before, during, and after the COVID-19 pandemic.

All statistical analyses were conducted using R software (version 4.4.0), with meta-analyses performed using the meta and metafor packages. Appropriate effect measures were selected for each outcome.

The pooled estimates for proportional outcomes (all-cause mortality, CRKP isolation rate, and antimicrobial resistance rates) were derived using a random-effects meta-analysis of transformed proportions via the metaprop function. Initially, the distribution of untransformed study-specific proportions was assessed for normality. We then implemented and compared four common transformations designed to stabilize variances and normalize data: logit, log, arcsine, and Freeman-Tukey double arcsine. The specific transformation used for the final meta-analysis of each outcome was chosen based on which method resulted in the best approximation of a normal distribution for the study-level effects. The final pooled proportion and its 95% confidence interval (CI) were obtained by back-transforming the derived results to the original proportion (0–1) scale for interpretation.

Heterogeneity across studies was evaluated using Cochran’s Q test and quantified with the I² statistic. A random-effects model was applied when significant heterogeneity was detected (I² > 50% or P < 0.10); otherwise, a fixed-effects model was used.

Subgroup analysis of resistance rates was performed to explore the potential differences. The differences in pooled resistance rates across time periods, geographical locations, and patient age groups were assessed using Cochran's Q test for subgroup differences, with a P-value of < 0.05 considered statistically significant.

To identify factors associated with antimicrobial resistance, we performed meta-regression analyses using the rma function in the metafor package. Each potential moderator —including study period, geographic region, and patient age— was first examined in a separate meta-regression model with a single covariate. Subsequently, a multivariable meta-regression model was fitted to simultaneously assess the effects of these factors. The statistical significance was set at P < 0.05.

Trends in CRKP resistance rates and geographical distribution across Chinese provinces were visualized using ggplot2.

Results

Results of the literature search

A total of 504 articles reporting on CRKP infections in China between 2006 and July 2023 were retrieved from PubMed. After screening the titles and abstracts to exclude irrelevant studies, 284 articles underwent full-text review.

Of these, 98 studies that reported CRKP related to infection were excluded due to the lack of carbapenem susceptibility or antimicrobial susceptibility data, which were essential for the analysis. Based on the inclusion and exclusion criteria, 68 articles were ultimately included in the study (Fig. 1).

Fig. 1.

Flow diagram of the study selection process. Illustrating the inclusion and exclusion of studies in the systematic review

These articles encompassed data from 19 provinces in China, involving 779 patients infected with CRKP and reporting on 1284 CRKP isolates. The earliest included literature was from 2007 (median study time).

The study included 1 case-control study, 1 cohort study, 27 cross-sectional studies, 1 surveillance study, and 38 case reports. Among the 68 included studies, 12 studies had 1–2 items rated as high risk of bias among the 9 evaluation items, primarily due to selection bias (Fig. 2). All studies were rated as having a low risk of bias rating for four to eight of the nine evaluation items; however, no study receiving a low-risk rating for all items. Overall, the methodological quality of the included studies was generally good. The numbers of published studies (Fig. 3a) and reported CRKP isolates (Fig. 3b) showed progressive increases. Correspondingly, both infection cases (Fig. 3c) and mortality numbers (Fig. 3d) exhibited similar upward trajectories, reflecting CRKP's growing clinical impact.

Fig. 2.

Proportional Distribution of Bias Risk Assessments across included Studies. (L: Low risk of bias; M: Moderate risk of bias; H: High risk of bias)

Fig. 3.

Research Status of CRKP in China, 2006–2023/07. a Annual number of studies reporting CRKP-related infections; b Annual number of CRKP isolates reported in included studies; c Annual number of patients with CRKP infections in included studies; d Annual number of deaths among CRKP-infected patients in included studies. Each point represents a count value corresponding to the y-axis, indicating the quantity of the respective measure (studies, isolates, patients, or deaths). The solid trend line depicts the overall temporal trend. As the PubMed search applied no publication year restriction, the earliest retrieved studies were published in 2006

Pooled CRKP isolation rate, site of infection, and associated mortality

Eight studies reported the CRKP isolation rates among K. pneumoniae strains (Fig. 4), encompassing a total of 3,227 K. pneumoniae isolates from five provinces in China. Among these, 670 were identified as CRKP, yielding a pooled CRKP isolation rate of 20% (95% CI: 0.10–0.31, I² = 98%, k = 8).

Fig. 4.

Geographical distribution of CRKP isolation rates across provinces in China. The map visualizes extracted data on CRKP isolation rates from included studies, with color intensity corresponding to the magnitude of the isolation rate in each province

We categorized all patient infections into three primary types: respiratory infections, urinary system infections, and others. Respiratory infections were the most prevalent, accounting for 77% (95% CI: 0.58–0.93, I² = 63%, k = 22) of cases (Fig. 5).

Fig. 5.

Distribution of infection types among patients with CRKP infections. Infection types are classified into three categories: respiratory infections, urinary tract infections, and other infection-related diseases, each represented by a distinct color. Rectangles indicate the 95% confidence intervals, horizontal lines represent the pooled rates derived from meta-analysis, and the corresponding heterogeneity values are provided

Patient outcomes were reported in 47 studies, which involved 678 hospitalized patients across 18 provinces in China; no community-based cases were included. Among these 678 patients with CRKP infection, 219 deaths were reported, resulting in a pooled all-cause mortality rate of 27% (95% CI: 0.14–0.41, I² = 73%, k = 47).

Compared to northern provinces, southern provinces exhibited higher pooled mortality rates (Fig. 6), including Sichuan (50%, 95% CI: 0.00–1.00), Chongqing (100%, 95% CI: 0.02–1.00), Jiangxi (52%, 95% CI: 0.38–0.66), Anhui (46%, 95% CI: 0.19–0.75), and Zhejiang (51%, 95% CI: 0.27–0.75).

Fig. 6.

Geographical variation in CRKP-associated mortality rates across Chinese provinces. The map illustrates extracted data on mortality among CRKP-infected patients from included studies, with color intensity corresponding to the mortality rate in each province

Clonal background and sample type distribution of CRKP in China

A total of 65 studies reported on 65 sequence types (STs) among 1,089 CRKP isolates from 18 provinces in China (Appendix S11). The most prevalent CRKP-associated STs were ST11 (n = 699, 64.19%) and ST15 (n = 98, 9.00%).

ST11 was detected across 13 provinces, encompassing both northern (Shandong, Jilin, Beijing, Shanxi, Henan) and southern regions (Zhejiang, Shanghai, Anhui, Hubei, Hunan, Jiangxi, Chongqing, Guangdong), which demonstrates the widespread dissemination of the lineage throughout the country.

ST15 (n = 98, 9.00%) was identified in Beijing (northern China) and four southern provinces (Zhejiang, Shanghai, Anhui, and Jiangxi). Other STs, such as ST20 (n = 47, 4.32%), ST76 (n = 20, 1.84%), ST307 (n = 23, 2.11%), and ST37 (n = 15, 1.38%) were distributed across 2–3 provinces, while the remaining STs were reported sporadically.

Regarding sample types, our analysis included 1,047 CRKP isolates from 15 different specimen sources. Sputum (n = 445, 42.5%) and blood (n = 244, 23.30%) were the two most common clinical specimens (Fig. 7).

Fig. 7.

Anatomical distribution of CRKP strain pooled isolation rates across clinical sample types. The chart presents pooled data on CRKP isolation rates from various anatomical sites, as reported in the included studies

Prevalence of carbapenemase-encoding genes in CRKP in China

Among the 68 studies included in this review, 65 provided data on carbapenemase genes from a total of 1,099 CRKP isolates. The major carbapenemase gene types detected were bla_KPC, bla_NDM, bla_IMP, and bla_OXA (Appendix S12).

The bla_KPC-2 variant was identified in 754 isolates (68.61%) across 34 studies, confirming its role as the predominant carbapenemase in China. Additionally, 121 isolates harbored the bla_NDM-1 gene, making it the most prevalent NDM genotype.

Data on co-occurring extended-spectrum β-lactamase (ESBL) genes were also extracted. The CRKP isolates included in this analysis harbored bla_SHV (n = 294), bla_CTX (n = 429), and bla_TEM (n = 323).

Notably, the bla_CTX-M-65 variant was detected in 192 isolates, identifying it as the predominant bla_CTX genotype.

Temporal trends in antimicrobial resistance of CRKP in China

We calculated the pooled resistance rates for 36 antibiotics (Appendix S13) and analyzed their temporal trends (Fig. 8). Sixteen of the 36 agents exhibited pooled resistance rates exceeding 95%. Notably, half of these agents were cephalosporins, which underscores the widespread resistance of CRKP strains to this antibiotic class.

Fig. 8.

Temporal trends in antimicrobial resistance rates of CRKP over the study period. The 36 reported antibiotics were randomly assigned into 9 groups, with 4 antibiotics in each group. a Cefazolin (CZO), Piperacillin-Tazobactam (TZP), Meropenem (MEM), and Imipenem (IMP); b Aztreonam (ATM), Cefepime (FEP), Ceftazidime (CAZ), and Cefuroxime (CXM); c Ceftriaxone (CRO), Colistin (COL), Amikacin (AMK), and Ciprofloxacin (CIP); d Gentamicin (GEN), Ertapenem (ETP), Ampicillin sulbactam (SAM), and Cefoxitin (FOX); e Ceftazidime-Avibatan (CZA), Tigecycline (TGC), Trimethoprim-sulfamethoxazole (SXT), and Levofloxacin (LEV); f Amoxillin-Clavulanic acid (AMC), Ampicillin (AMP), Piperacillin (PIP), and Tobramycin (TOB); g Fosfomycin (FOS), Cefotaxime (CTX), Nitrofurantoin (NIT), and Cefotetan (CTT); h Tetracycline (TCY), Minocycline (MNO), Cefoperazone-Sulbactam (SCF), and Polymyxin B (PMB); i Ticarcillin (TIC), Cefotaxime-Sulbactam (CSL), Chlorampheniclo(CHL), and Doxycycline (DOX).

Polymyxin B —a last-resort agent for managing multidrug-resistant K. pneumoniae infections— demonstrated a relatively low pooled resistance rate of 2% (95% CI: 0.01–0.04, I² = 0%, k = 15), indicating its retained efficacy despite high resistance to other antibiotics.

Our time-stratified subgroup analysis of 14 commonly tested antibiotics revealed significant shifts in resistance rates during the COVID-19 pandemic (Appendix S14). Specifically, resistance rates to ciprofloxacin (Random effects model: Q_b = 9.88, df = 1, P < 0.010) and levofloxacin (Random effects model: Q_b = 7.69, df = 1, P < 0.010) increased significantly.

Levofloxacin resistance rose sharply from 57% (95% CI: 0.34–0.79, I² = 96%, k = 33) in the pre-COVID-19 to 96% (95% CI: 0.79–1.00, I² = 68%, k = 9) during the pandemic.

Geographic variation in antimicrobial resistance rate of CRKP in China

We compared the antimicrobial resistance profiles of CRKP isolates between northern and southern China using data from 20 and 52 studies conducted in these regions respectively (Appendix S15). The analysis focused on 19 antibiotics. Carbapenems were excluded as CRKP is intrinsically resistant to this class.

Isolates from southern China demonstrated the highest levels of resistance to a specific set of antimicrobial agents. In contrast, while also exhibiting critically high resistance, the profile of the most resisted antibiotics in northern China differed. The highest resistance rates in the north were observed for a different combination of drugs (Table 1).

Table 1.

Comparison of antibiotics with the highest resistance rates in Southern and Northern China

Region	Antibiotic	Studies (k)	I2(%)	Pooled resistance rate (95% CI)
Southern China	Cefepime	35	0	1.00 (0.99–1.00)
Southern China	Cefotaxime	16	0	1.00 (0.98–1.00)
Southern China	Tetracycline	3	0	1.00 (0.95–1.00)
Northern China	Piperacillin-tazobactam	12	93	0.99 (0.95–1.00)
Northern China	Cefepime	12	89	0.94 (0.74–1.00)
Northern China	Cefotaxime	4	93	0.91 (0.73–1.00)

A provincial-level analysis of seven antibiotics further confirmed that cefotaxime resistance was consistently high across multiple provinces (Fig. 9).

Fig. 9.

Geographical distribution of antimicrobial resistance rates for seven antibacterial agents across China. The resistance rate of each agent is displayed by province, with color intensity indicating the magnitude of the resistance rate. a Chloramphenicol; b Cefotaxime; c Ceftazidime-avibactam; d Fosfomycin; e Minocycline; f Tetracycline; g Tobramycin

Statistically significant geographic variations were identified for three agents: chloramphenicol (Random effects model: Q_b = 4.97, df = 1, P = 0.030), ceftazidime-avibactam (Random effects model: Q_b = 8.58, df = 1, P < 0.010), and tetracycline (Random effects model: Q_b = 9.69, df = 1, P < 0.010). Resistance to chloramphenicol and ceftazidime-avibactam was significantly lower in the south than in the north, whereas the converse was true for tetracycline.

Variation in antimicrobial resistance of CRKP across age groups

We analyzed the antimicrobial resistance profiles of CRKP by age group, stratifying the population into minors (0–17 years), adults (18–59 years), and the elderly (≥ 60 years), according to World Health Organization definitions.

A subgroup analysis based on patient age revealed statistically significant differences in resistance rates for a substantial subset of the antibiotics studied. The analysis identified that age was a significant factor associated with variations in resistance to several key antimicrobial agents. These findings indicate that patient age may be an important consideration for predicting antibiotic resistance patterns (Table 2, Appendix S16).

Table 2.

Subgroup analysis of the association between patient age and antibiotic resistance in CRKP

Antibiotic	Subgroup	Studies (k)	Pooled resistance rate (95% CI)	I2 (%)a		Qb	P
Levofloxacin	0–17 years	9	0.02 (0.00–0.14)	77		62.82	< 0.010
	18–59 years	14	0.92 (0.76–0.99)	61
	≥ 60 years	16	0.95 (0.83–1.00)	86
Amikacin	0–17 years	10	0.04 (0.00–0.22)	79		26.55	< 0.010
	18–59 years	16	0.81 (0.60–0.95)	87
	≥ 60 years	22	0.70 (0.43–0.90)	91
Gentamicin	0–17 years	8	0.14 (0.01–0.40)	77		19.95	< 0.010
	18–59 years	6	0.89 (0.61–1.00)	94
	≥ 60 years	14	0.86 (0.65–0.98)	81
Ciprofloxacin	0–17 years	8	0.22 (0.00–0.68)	84		18.11	< 0.010
	18–59 years	9	0.90 (0.60–1.00)	69
	≥ 60 years	20	1.00 (0.98–1.00)	0
Colistin	0–17 years	5	0.00 (0.00–0.00)	0		7.64	0.020
	18–59 years	10	0.00 (0.00–0.02)	0
	≥ 60 years	13	0.10 (0.00–0.42)	83
Nitrofurantoin	0–17 years	2	0.25 (0.00–0.79)	39		7.15	< 0.010
	≥ 60 years	2	1.00 (0.92–1.00)	0
Aztreonam	0–17 years	10	0.81 (0.44–1.00)	77		6.45	0.040
	18–59 years	13	1.00 (0.99–1.00)	50
	≥ 60 years	15	0.97 (0.87–1.00)	73

^a The random-effects model was used to calculate the pooled resistance rate due to high heterogeneity

Factors influencing antimicrobial resistance in CRKP

We performed univariate meta-regression analyses to identify study-level factors associated with antimicrobial resistance in CRKP. Patient age was identified as a significant positive predictor, demonstrating a statistically significant association with increased resistance to colistin, gentamicin, levofloxacin, tetracycline, and ciprofloxacin (Table 3).

Table 3.

Univariable meta-regression analysis of the association between patient age and antibiotic resistance in CRKP

Antibiotic	Studies (k)	Coefficient (β) (95% CI)	z	P
Colistin	28	0.0027 (0.0002–0.0052)	2.08	0.037
Gentamicin	28	0.0056 (0.0030–0.0081)	4.27	< 0.001
Levofloxacin	39	0.0087 (0.0067–0.0108)	8.38	< 0.001
Tetracycline	3	0.0050 (0.0007–0.0094)	2.26	0.024
Ciprofloxacin	37	0.0081 (0.0059–0.0102)	7.25	< 0.001

Beyond age, we evaluated several other potential influencing factors: the diversity of antimicrobial susceptibility testing (AST) methods, geographical location (Appendix S17), economic factors (measured by provincial GDP levels), and population density (Appendix S18).

The number of AST methods used was identified as a significant moderator, showing divergent associations with resistance across different antimicrobial agents. It was negatively associated with resistance to chloramphenicol, minocycline, and piperacillin-tazobactam, while its association with gentamicin resistance was positive (Table 4).

Table 4.

Univariable meta-regression analysis of the association between AST methods and antibiotic resistance in CRKP

Antibiotic	Studies (k)	Coefficient (β) (95% CI)	z	P
Gentamicin	37	0.1357 (0.0001–0.2714)	1.96	0.049
Chloramphenicol	7	-0.5702 (-0.7955–-0.3449)	4.96	< 0.001
Minocycline	5	-0.3366 ( -0.5611–-0.1120)	2.94	0.003
Piperacillin-tazobactam	46	-0.0647 (-0.1099–-0.0194)	2.80	0.005

Geographical location was identified as another significant moderator of resistance patterns. CRKP isolates from southern China demonstrated significantly higher resistance rates to fosfomycin, tobramycin, minocycline, tetracycline, and cefotaxime compared to those from northern China. Conversely, resistance to chloramphenicol and ceftazidime-avibactam was significantly lower in southern China (Table 5).

Table 5.

Univariable meta-regression analysis of the association between geographical location and antibiotic resistance in CRKP

Antibiotic	Studies (k)	Coefficient (β) (95% CI)	z	P
Fosfomycin	17	0.2815 (0.0501–0.5129)	2.38	0.017
Tobramycin	17	0.6153 (0.4221–0.8085)	6.24	< 0.001
Minocycline	4	0.3232 (0.1017–0.5447)	2.86	0.004
Tetracycline	5	0.3684 (0.0590–0.6779)	2.33	0.020
Cefotaxime	20	0.1436 (0.0101–0.2771)	2.11	0.035
Chloramphenicol	6	-0.5095 (-0.7673–-0.2516)	3.87	< 0.001
Ceftazidime-avibactam	18	-0.5786 (-0.8182–-0.3389)	4.73	< 0.001

Furthermore, provincial GDP level was a positive predictor factor for minocycline (Meta-regression analysis: β = 0.0854, 95% CI: 0.0242–0.1465, P = 0.006) and tetracycline (Meta-regression analysis: β = 0.0553, 95% CI: 0.0052–0.1054, P = 0.030) resistance. Population density showed a positive correlation with Fosfomycin resistance (Meta-regression analysis: β = 0.0002, 95% CI: 0.0000–0.0003, P = 0.001).

Impact of antimicrobial resistance on mortality of CRKP infections

Antimicrobial resistance is a critical determinant of mortality in CRKP infections. We performed a univariate meta-regression analysis to assess the association between resistance rates of 35 antimicrobial agents and all-cause mortality. The analysis revealed that overall antimicrobial resistance exerted a positive effect on mortality (Fig. 10). Specifically, resistance rates for 14 of the 35 agents demonstrated a significant positive correlation with mortality.

Fig. 10.

Forest plot. Resistance rates of 14 antimicrobial agents show a significant positive correlation with increased all-cause mortality in CRKP infections. CAZ Ceftazidime, SAM Ampicillin sulbactam, PIP Piperacillin, FOX Cefoxitin, CTX Cefotaxime, AMK Amikacin, GEN Gentamicin, LEV Levofloxacin, TCY Tetracycline, CZO Cefazolin, CIP Ciprofloxacin, AMP Ampicillin, TIC Ticarcillin-clavulanic acid, PMB Polymyxin B. *: P < 0.05; **: P < 0.01; ***: P < 0.001

Notably, polymyxin B resistance exhibited a strong positive correlation with mortality (Meta-regression analysis: β = 3.7494, 95% CI: 2.5258–4.9730, P < 0.001), indicating that resistance to this last-line agent is significantly associated with an increased risk of death in patients with CRKP infections and underscoring its profound impact on clinical outcomes.

A notable exception was observed for ceftazidime (CAZ), where higher resistance rates were paradoxically associated with lower mortality (Meta-regression analysis: β = -4.1985, 95% CI: -7.5689–-0.8280, P = 0.015).

Discussions

Our study provides a comprehensive analysis of the resistance profile and clonal epidemiology of CRKP infections in China. A distinctive finding was the observed temporal and geographic heterogeneity in resistance patterns, most notably a significant increase in resistance to fluoroquinolones (ciprofloxacin and levofloxacin) during the COVID-19 pandemic. The extent of multidrug resistance is severe, with pooled resistance rates exceeding 95% for 16 out of 36 antibiotics. Furthermore, significant age-related disparities were identified; adults and the elderly patients exhibited higher resistance to colistin, fluoroquinolones and aminoglycosides compared to minors.

The pooled all-cause mortality rate among hospitalized patients was 27% (95% CI: 0.14–0.41), underscoring the considerable clinical burden of CRKP infections. Molecular epidemiology revealed the dominance of the ST11 lineage, which accounted for 64.19% of collected isolates, highlighting its pivotal role in the persistence and spread of CRKP in China.

To our knowledge, this systematic review and meta-analysis is the first to comprehensively evaluate the spatiotemporal dynamics and age-stratified resistance patterns of CRKP infections across China, with a particular emphasis on the impact of the COVID-19 on antimicrobial resistance. In contrast to previous investigations limited to specific regions or short timeframes, our study integrates nationwide surveillance data spanning nearly two decades, thereby providing a unique and panoramic overview of the evolving epidemiology of CRKP in China.

Since the initial report of a CRKP infection in Zhejiang province, CRKP has been detected in most regions of China. The clonal background of CRKP in China is diverse, with most sequence types (STs) exhibiting region-specific distributions, a pattern potentially attributable to variations in local medical environments and infection control measures (Fig. 11).

Fig. 11.

Distribution of CRKP sequence types (STs) across selected provinces in China. The map displays the distribution of STs in provinces meeting the following criteria: containing three or more distinct STs and having more than 10 CRKP isolates reported, based on extracted data from the included studies

Our analysis identified two major high-risk lineages: the hyperepidemic ST11 and ST15 clones. Among these, ST11 dominates the national epidemiology of CRKP, a phenomenon likely driven by a combination of clonal expansion, enhanced ecological fitness, and successful horizontal gene transfer. This clone was extensively disseminated across 13 provinces in both northern and southern China, a finding consistent with previous reports [25–29].

The dominance of ST11 can be attributed to several factors. First, its association with the carbapenemase gene bla_KPC-2, whose horizontal transfer is efficiently mediated by convergent IncFII plasmids [30], has greatly facilitated its spread. Second, ST11 strains often exhibit a high fitness advantage, possibly due to adaptations in metabolic efficiency, biofilm formation, or evasion of host immunity, enabling sustained transmission within healthcare settings [31]. Finally, its success reflects patterns of clonal expansion—founded by a few highly successful lineages that acquire advantageous genetic elements and subsequently disseminate widely.

In contrast, the ST15 clone, while also a significant CRKP lineage, displayed a more geographically restricted distribution, primarily in southern China. This pattern may partly reflect sampling bias, as more studies were conducted in the south, but may also suggest differences in ecological adaptation or selective pressures between regions. ST15 is frequently linked to the carbapenemase gene blaOXA-232 and has been shown to effectively spread within hospitals [32]. Both ST11 and ST15 are major causes of pediatric CRKP infections [10, 33], highlighting their clinical relevance across age groups.

In summary, while the CRKP population in China is genetically diverse, it is dominated by the ST11 high-risk clone, whose nationwide prevalence is likely due to a confluence of efficient horizontal gene transfer (especially of bla_KPC-2-bearing plasmids), enhanced bacterial fitness, and widespread clonal expansion. These findings underscore the urgent need for enhanced genomic surveillance and stricter infection control measures to limit the continued dissemination of these successful lineages.

A comprehensive understanding of temporal and spatial resistance patterns is essential for promoting rational antibiotic use and formulating effective strategies. Our analysis revealed an alarming prevalence multidrug resistance among CRKP isolates in China, with pooled resistance rates exceeding 95% for 16 out of 36 antibiotics, encompassing penicillins, carbapenems, and cephalosporins.

The COVID-19 pandemic has potentially exacerbated this crisis. Increased rates of secondary bacterial infections in critically ill patients led to heightened antibiotic consumption [34] creating significant challenges in preventing antibiotic overuse and misuse [35]. Our time-trend analysis confirms a significant surge in resistance rates during the pandemic period.

Most notably, resistance to ciprofloxacin rose from 87 to 100%, and levofloxacin resistance escalated sharply from 57 to 96%. This notable surge in fluoroquinolone resistance observed in our study aligns with broader national and global trends during this period. In China, despite a temporary overall decrease in antibiotic consumption in outpatients during the strict containment phases of the pandemic, the prescribing patterns for broad-spectrum antibiotics, including fluoroquinolones, remained a concern particularly in hospital settings [36]. Studies from other regions also reported increased use of broad-spectrum antibiotics as empirical therapy during the pandemic, driven by clinical uncertainties and challenges in distinguishing COVID-19 pneumonia from bacterial co-infections initially [37]. For instance, WHO data indicated that approximately 75% of COVID-19 inpatients globally received antibiotics empirically, despite a low prevalence of bacterial co-infections (around 8%) [38]. This widespread empirical use, especially of broader-spectrum agents, likely exerted significant selective pressure, contributing to the accelerated development of antimicrobial resistance, as observed in our findings for fluoroquinolones [36].

Significant regional differences in CRKP resistance profiles were observed across China (Appendix S17; Appendix S18). The chloramphenicol resistance rate was significantly higher in northern China (Random effects model: Q_b = 4.97, df = 1, P = 0.030), whereas the tetracycline resistance rate was lower in the north than in the south (Random effects model: Q_b = 9.69, df = 1, P < 0.010). Although global studies have not reported significant regional differences in resistance to ceftazidime-avibactam, our study identified regional variability, with higher resistance rates in northern China [39]. This variability may be attributed to the primary mechanism of CRKP resistance to ceftazidime-avibactam, which involves bla_KPC-2 variants such as bla_KPC-72 [40]. The regional distribution of these resistance genes likely contributes to the observed differences.

Geographic variations in antimicrobial resistance can be influenced by local antibiotic usage patterns, healthcare quality, and public health policies. A cross-sectional study underscored significant differences in antibiotic use between developed (e.g., Zhejiang) and less developed provinces (e.g., Shaanxi) in China. Children in economically underdeveloped provinces face higher risks of antibiotic misuse in both household and clinical settings [41]. These regional discrepancies likely contribute to geographic differences in resistance, underscoring the need for tailored public health policies and interventions.

Similarly, our geographic subgroup and regression analysis confirmed higher resistance rates to ceftazidime-avibactam and chloramphenicol in northern China, and higher tetracycline resistance in Southern China. This discrepancy likely arises from regional differences in antibiotic usage practices and healthcare standards. A study on antibiotic consumption trends in 76 countries highlighted that low and middle income countries are driving the global increase in antibiotic consumption, which correlates with growth in gross domestic product per capita (GDPPC) [42]. Our univariate regression analysis also revealed a positive correlation between the resistance rates of minocycline and tetracycline and GDP levels, supporting this observation.

The risk of antimicrobial resistance varies significantly across different age groups. Our analysis revealed that resistance rates for six out of nineteen drugs differed significantly across age groups, with adults exhibiting higher resistance rates than minors (Appendix S17; Appendix S19). Among adults, elderly individuals (≥ 60 years) bear a particularly high burden of antibiotic-resistant infections. This is partly due to frequent hospital and ICU admissions for severe health conditions, which increases their exposure to healthcare-associated pathogens [43, 44 ]. Additionally, older adults are more susceptible to antibiotic misuse as a result of cognitive and physical decline, frequent medical visits, and weakened immune function, all of which contribute to elevated resistance rates [45].

With advancing age, prolonged antibiotic exposure becomes more common. A study conducted in a suburban area of Shanghai demonstrated widespread antibiotic exposure among adults aged 21–75 years, with 45.9% of urine samples testing positive for 18 common antibiotics [ 46]. These factors collectively drive higher antimicrobial resistance in the adult population, underscoring the need for rational antibiotic use in clinical practice.

Age-stratified subgroup and regression analysis of four antibiotics —colistin, gentamicin, levofloxacin, and ciprofloxacin— revealed significant variations in resistance across age groups, with the highest rates observed in the adults. Both analyses consistently indicated a positive correlation between age and resistance for these agents. Notably, colistin resistance showed a marked increasing trend with age, highlighting a growing challenge in managing CRKP infections among the elderly.

A review of the global burden of bacterial antimicrobial resistance from 1990 to 2021, along with projections for the next 25 years, reported that although deaths due to antimicrobial resistance have decreased by more than 50% among children under 5 years of age, they have increased by over 80% among adults aged 70 years and older–a trend expected to continue [47]. The findings of our study are consistent with these global patterns.

A previous systematic review reported a CRKP infection mortality rate of 44.82% in Asia [8]. In contrast, the primary data from our study indicate that the overall mortality rate attributable to CRKP among hospitalized patients in China is 27% (95% CI: 0.14–0.41). Additionally, a large multicenter case–control and cohort study in China reported a 28-day mortality rate of 24.2% for CRKP infections [48]. These findings suggest that some other Asia countries may have higher CRKP-related mortality rates than China.

However, these mortality estimates should be interpreted with caution due to methodological limitations. Our analysis primarily focused on resistance profiles and molecular epidemiology, with mortality as a secondary outcome. Consequently, the data lacked sufficient detail to stratify by key variables such as infection type, comorbidities, or treatment regimens, which precluded adjusted analyses. Only a subset of studies provided stratified mortality data. Forcible stratification might introduce bias due to incomplete reporting. Moreover, the lack of standardized mortality attribution (e.g., CRKP-attributable versus all-cause death) across studies further limits comparability.

Despite these constraints, our results highlight the substantial clinical threat of CRKP in China and underscore the need for future prospective studies with standardized mortality assessments and stratified risk analyses.

The mortality associated with CRKP infection is influenced by various factors, including comorbidities and clinical treatment regimens [12, 44 ]. Univariate meta-regression analysis in our study indicated a significant association between age and mortality (Appendix S20), with mortality rising with advancing age (Meta-regression analysis: β = 0.0043, 95% CI: 0.0023–0.0064, P < 0.001). Advanced age has been identified as an independent risk factor for death in patients infected or colonized with CRKP [49–51 ], which is often attributable to diminished immune function and a higher burden of underlying diseases among older adults. Consequently, elderly patients present with more complex clinical conditions and face greater treatment challenges during CRKP infection.

The mortality rate of CRKP infection is significantly associated with antimicrobial resistance. Regression analysis in our study indicated that resistance rates to 14 antimicrobial agents were positively correlated with mortality. Previous studies comparing mortality between CRKP and carbapenem-sensitive K. pneumoniae (CSKP) infections have consistently reported higher fatality rates among CRKP-infected patients [52–54]. Notably, our analysis revealed that the positive correlation between polymyxin B resistance and mortality was more pronounced than other antibiotics (Fig. 10), suggesting that polymyxin B resistance poses a particularly critical challenge in the clinical management of CRKP infections. Genomic epidemiology analyses suggest that the emergence of polymyxin B resistance is not confined to a single clone but is observed across diverse genetic backgrounds of CRKP isolates in our cohort. This indicates that the resistance is likely driven by independent evolutionary events (e.g., mutations in chromosomal genes such as pmrAB, phoPQ or mgrB) under antimicrobial selective pressure, rather than solely by the clonal spread of a specific resistant lineage [55]. However, the potential role of plasmid-mediated mobile polymyxin B resistance genes (e.g., mcr variants), which can facilitate horizontal transfer across strains, cannot be ruled out and warrants further investigation [56]. Given that polymyxin B remains one of the last-line therapeutic options for CRKP infections, especially in the context of limited availability of novel antibiotics [9], the use of combination regimens —such as polymyxin B with tigecycline— is recommended to mitigate the emergence of colistin resistance [57]. Although polymyxin B resistance remains relatively uncommon, its presence is strongly associated with significantly higher mortality, underscoring the severely limited therapeutic options and grave implications for patients harboring such resistant strains. The polyclonal nature of polymyxin B resistance highlighted in this study underscores the importance of stringent infection control measures to prevent the spread of resistant strains and the critical need for antimicrobial stewardship programs to mitigate the selection pressure driving the emergence of such resistance.

However, our analysis revealed a negative correlation between ceftazidime (CAZ) resistance and CRKP mortality. As a third-generation cephalosporin, CAZ is generally ineffective against CRKP due to the presence of carbapenemase genes such bla_KPC, which encode enzymes that hydrolyze β-lactam antibiotics. In our study, the overall CAZ resistance rate reached 99%. Despite the lack of efficacy of CAZ monotherapy against CRKP, the combination agent ceftazidime-avibactam has demonstrated considerable clinical effectiveness in treating these infections [58–60].

The observed inverse relationship between CAZ resistance and CRKP mortality may be attributed to adaptive changes in clinical practice. With high resistance rates leading to reduced reliance on CAZ, clinicians are likely avoiding its ineffective use and increasingly turning to more potent alternatives such as ceftazidime-avibactam or colistin. This shift in prescribing behavior may help explain the associated reduction in mortality.

There are several limitations to our study. First, our sample consisted solely of clinically hospitalized patients, which may limit the generalizability of our findings, as the results may not be representative of community-based populations or outpatient settings. Consequently, the estimated mortality rates of CRKP infections apply specifically to hospitalized individuals and should not be extrapolated beyond this context. Second, the literature search was conducted using a single electronic database. Although no language restriction was applied, including additional databases and gray literature could have improved the comprehensiveness and robustness of the data synthesis.

Our study utilized a dataset encompassing 19 provinces; however, we acknowledge that it may not fully capture the socioeconomic and environmental characteristics of all key regions across China. This limitation arises primarily from the scarcity of published data in less-developed, rural, and remote areas, particularly in inland provinces such as Xinjiang, Tibet, and Qinghai, which have lower population densities and were not included in our analysis. Consequently, the overrepresentation of densely populated southeastern coastal provinces could introduce a regional bias by underestimating the conditions prevalent in underrepresented western and rural regions. While our findings provide valuable insights into the situations of the covered provinces, this geographical imbalance may reduce the national representativeness of our results and limits the direct extrapolation of our conclusions to all of China. Future studies should prioritize securing data from these critical but data-scarce regions to enable a more comprehensive national assessment.

Additionally, heterogeneity in laboratory methodologies —such as variations in susceptibility testing standards and clinical outcome definitions across studies— may introduce bias into estimates of resistance and mortality. Although substantial heterogeneity advises caution in interpretation, the stability of our findings across multiple subgroup analyses supports the robustness of the overall conclusions. Therefore, we have prioritized the presentation and synthesis of the full body of evidence.

Moreover, as most included studies reported aggregate mortality data without stratification by the site of infection, we were unable to perform a quantitative meta-analysis on the association between specific infection sites (e.g., respiratory vs. urinary tract) and mortality risk. This lack of granular data precludes a definitive conclusion on how infection site directly influences patient outcomes in CRKP infections, despite our observation of their prevalence.

Further limitations include constraints in the review process due to insufficient genomic data and inconsistent reporting of comorbidities, which restricted deeper mechanistic interpretation. Despite these limitations, our study highlights the considerable mortality burden associated with CRKP infections in China and reveals notable temporal and regional variations in resistance profiles. This work offers a comprehensive epidemiological and resistance characterization of CRKP in China and underscores the urgent need for enhanced infection control strategies and optimized antibiotic therapy to mitigate the impact of CRKP nationwide.

Conclusion

Our study presents a comprehensive analysis of the resistance profile and associated risk factors of CRKP infections in China. To our knowledge, this is the first systematic review and meta-analysis to extensively evaluate the spatiotemporal trends and age-stratified resistance patterns of CRKP infections across China, with particular emphasis on the impact of the COVID-19 pandemic on antimicrobial resistance.

A significant increase in resistance to ciprofloxacin and levofloxacin was observed during the pandemic, alongside pronounced regional variations in resistance patterns. Notably, resistance rates to 16 out of 36 antibiotics exceeded 95%. Substantial age-related disparities were identified: adults and elderly patients exhibited higher resistance to colistin, fluoroquinolones and aminoglycosides compared to minors, highlighting a severe multidrug resistance crisis. The overall mortality rate among hospitalized patients was 27% (95% CI: 0.14–0.41), with sequence type ST11 as the dominant sequence clone (64.19%), indicative of its widespread dissemination.

Our findings indicate that CRKP resistance in China is influenced by temporal, geographic, and demographic factors. The positive correlation between colistin resistance and mortality underscores the urgent need for targeted clinical and public health interventions. Given China’s aging population, the burden of antimicrobial resistance is anticipated to increase, necessitating more stringent infection control measures and rational antibiotic use policies nationwide.

Author contributions

X.J., H.C., C.F., and B.Z. were responsible for the study design. X.Y. wrote the manuscript, and Z.L. performed the data analysis. X.W., Z.L., and Q.L. contributed to data verification. Study oversight and supervision were provided by K.G., Y.Y., L.G., K.W., H.X., and W.L. The final manuscript was reviewed and approved by all authors.

Funding

National Key Research and Development Program of China, 2020YFE0204300, National Natural Science Foundation of China, 82072314, Shandong Provincial Laboratory Project, SYS202202, Fundamental Research Funds for the Central Universities, 2022ZFJH003.

Data availability

All raw data extracted from the 68 included studies, along with detailed documentation of the literature quality assessment, are provided in the Supplementary Materials in Microsoft Excel format.

Declarations

Competing interests

The authors declare no competing interests.