Incidence, antimicrobial resistance and mortality of Klebsiella pneumoniae bacteraemia in Shanghai, China, 2018–2022

Abstract

Background

Klebsiella pneumoniae (KP) accounts for high antimicrobial resistance and mortality rates of bloodstream infections (BSIs).

Objectives

To investigate incidence, antimicrobial resistance and risk factors for mortality of KP BSIs in East China.

Methods

A retrospective study of patients with KP BSIs was conducted in a tertiary care hospital from 2018 to 2022. Medical records of all hospitalised patients with KP BSIs were reviewed and analysed. The incidence, antimicrobial resistance and mortality of KP BSIs were evaluated. The Kaplan-Meier method was used to plot survival curves and logistic regression was used to analyse risk factors for crude 30-day mortality.

Results

A total of 379 inpatients with KP BSIs were enrolled. The incidence of patients with KP BSIs was fluctuating between 4.77 and 9.40 per 100,000 patient-days. The crude 30-day mortality rate of these patients was 26.39%. Of the 379 KPisolates, 197 (51.98%) were carbapenem-resistant (CR) and 252 (66.49%) were multidrug-resistant (MDR). All isolates showed the lowest resistance to tigecycline (13.77%) and polymyxin B (14.61%). Cases with MDR/CR isolates had significantly longer length of hospital stay, higher crude 30-day mortality and medical costs than non-MDR/non-CR isolates. Age, CR phenotype, paracentesis, indwelling central venous catheter (CVC), use of carbapenems, tetracyclines, polymyxins B, and irrational empiric treatment were independently associated with crude 30-day mortality.

Conclusion

MDR/CR KP BSIs are associated with increased mortality, healthcare costs and prolonged hospitalisation. Patients with advanced age, CR phenotype, paracentesis, CVC, exposure to some antibiotics, and irrational empirical antibiotic treatment are at higher mortality risk.

Keywords:

• Klebsiella pneumoniae
• bloodstream infections
• incidence
• mortality
• antimicrobial resistance

Introduction

Bloodstream infections (BSIs) are one of the leading causes of mortality. The growing number of immunocompromised individuals, use of invasive procedures and medications contributed to the high incidence of BSIs, which is receiving increasing attention worldwide [1]. Klebsiella pneumoniae (KP) is an important Gram-negative bacillus causing pneumonia, urinary tract infections, BSIs and abdominal infections [2]. It is ranked as the second Gram-negative organism causing BSIs [3]. Mortality of BSIs caused by KP ranged between 20% and 40% [4].

Carbapenems are broad-spectrum antimicrobials for treatment of infectious diseases and are also used as the last line drug in the treatment of Gram-negative bacteria [5]. In recent years, with the wide usage of cephalosporins, the proportion of extended-spectrum beta-lactamase producing KP has increased rapidly [6], resulting in significant increase of carbapenems usage and proportion of carbapenem-resistant KP (CRKP) [7]. According to the China Bacterial Resistance Monitoring Network, the resistance rate of CRKP increased by tenfold from 3% to 30% between 2005 and 2023 in China [8]. In addition, multidrug-resistant KP (MDRKP) has been on the rise in China in recent years [9], and infections caused by MDRKP result in mortality rates as high as 40%–50% [10]. Moreover, hypervirulent KP (hvKP) has been reported to be widespread in Asia-Pacific region, Europe and North America, the Middle East and Africa in recent years [11–13]. In China, the proportion of hvKP has been reported as high as 57.14% [14].

Mortality rate of BSIs is closely related to the drug-resistant phenotype and virulence of pathogens [15]. However, there were few studies on the epidemiology of KP BSIs in the context of high virulence and high drug resistance. In this study, we conducted a comparative analysis of the clinical characteristics, incidence, drug-resistant phenotypes and risk factors for 30-day mortality of KP BSIs among patients in a large tertiary care hospital in the eastern region of China.

Materials and methods

Ethics

The study was approved by the ethics committee of Ruijin Hospital (KY2023-083). Because of the retrospective nature of the study, the committee waived informed consent. Patient data was obtained from medical record systems and analysed anonymously to protect patient privacy.

Study design and setting

This is a retrospective study performed at Ruijin Hospital from 1 January 2018 to December 31 2022. Ruijin Hospital located in the eastern part of China, is a large-scale, tertiary-level A-class hospital with approximately 3400 beds. It has approximately 5.36 million outpatient and emergency visits and nearly 150,000 hospitalized patients per year. Patients with at least one positive blood culture of KP were enrolled in our study. Patients will be excluded in the following cases: (i) positive culture results are considered as contaminants (confirmed by physicians caring for the patient); (ii) patients with incomplete/inaccurate medical records; (iii) patients insisted on discharge from hospital against medical advice. Only the first episode of each patient was included and each patient was included only once.

Definitions

KP BSIs were defined as a positive blood culture of KP with fever, chills, shiver, stun, cutaneous mucous haemorrhage, hypotension, multiple organ dysfunction, increased heart rate (>90 times/min), increased respiratory rate (>20 times/min), low partial arterial carbon dioxide pressure (<32 mmHg), abnormal number of peripheral blood white blood cell or elevated inflammatory indexes such as C-reactive protein and procalcitonin [16]. The presence of KP BSIs was defined as the moment when the first blood culture was positive for KP. A case was defined as a patient diagnosed with KP BSIs. KP BSIs incidence was defined as the number of cases per 100,000 patient-days. Polymicrobial BSIs was defined as a set of blood cultures simultaneously isolating KP and other microorganisms [17]. According to the Centres for Disease Control and Prevention’s criteria, hospital-associated BSI (HABSI) was defined as a BSI that occurred more than 48 h after admission to hospital or within 48 h of the last hospital discharge [18]. Primary BSIs were defined as bacteria for which there was no documented focal source. Source of infection was defined as the most possible origin of infection responsible for KP BSIs according to medical records. MDR was defined as acquired non-susceptibility to at least one agent in three or more antimicrobial categories [19]. CR was defined as resistance to any carbapenem [20]. Inappropriate empirical antimicrobial therapy was defined as empirical dosing that did not include any drug with in-vitro activity before blood culture report, whereas inappropriate definitive antimicrobial therapy refers to the moment of receiving antimicrobial susceptibility test results [21,22].

Data collection

The following clinical information of the patient was extracted from the hospital electronic medical record system: gender, age, admission department, intensive care unit (ICU) stay, length of hospital stay, site of complicated infection, underlying diseases, history of invasive clinical procedures, special treatments (glucocorticoids, immunosuppressive agents, radiotherapy and chemotherapy, and dialysis), data of antibiotic therapy. Definition of each variable corresponding to these data was listed in Supplementary Table 1.

Microbiology

Isolates were identified using MALDITOF MS (bioMérieux, Marcy l‘Etoile, France) and drug sensitivity tests were performed using VITEK^®2 Compact (bioMérieux, Marcy l‘Etoile, France). Drug sensitivity tests were interpreted in accordance with Clinical and Laboratory Standard Institute standard [23]. For tigecycline, the results were analysed based on the breakpoint of European Committee On Antimicrobial Susceptibility [24].

Statistical analysis

Resistance rates of MDR/non-MDR, CR/non-CR groups of commonly used drugs were compared using Pearson’s chi-square test or Fisher’s exact test. Continuous variables that conform to normal distribution were expressed as mean and standard deviation, and continuous variables that did not conform to normal distribution were expressed as median and interquartile range (IQR). Categorical variables were compared using Pearson’s chi-square test and Fisher’s exact test. Logistic regression was used to analyse risk factors for crude 30-day mortality. Correlation and relevant interactions between variables with p < 0.05 in univariate analysis were checked. After removing variables with high-level correlation (correlation coefficient ≥0.70), the remaining variables were considered for inclusion in the multivariate model to assess independent risk factors associated with 30-day mortality. Odds ratio (OR) and 95% confidence interval (95% CI) were calculated to determine the strength of these associations. The Kaplan–Meier method was used to plot 30-day survival curves after the day of the first positive blood culture for KP and differences between survival curves by antimicrobial resistant phenotypes were evaluated by the log-rank test. P values were two-tailed and p < 0.05 was considered statistically significant. SPSS version 26.0 (IBM, Armonk, NY, USA) and GraphPad Prism version 9.0.0 (GraphPad Software, San Diego, CA, USA) were used for statistical analysis and data generation.

Results

Demographic and clinical details

A total of 379 cases were included in this study (Supplementary Figure 1). The clinical characteristics of enrolled patients are shown in Table 1. Their age ranged from 2 to 95 years, with a median age of 63 years, and 42.22% were 65 years or older. Males accounted for 71.24%. Of the 379 cases, 313 patients were HABSIs (82.59%) and 193 were primary BSIs (50.92%). There were 60 patients with polymicrobial BSIs (60/379, 15.83%). Of the 186 (49.08%) cases with known origin, gastrointestinal, abdominal, and pelvic infections were the most common source (86/186, 46.2%), followed by respiratory tract infections (51/186, 27.4%) and skin and soft tissue infections (35/186, 18.8%). About half of the cases were from surgical (174/379, 45.91%), 34.75% from haematology (41/118, 34.75%) and 22.96% from ICU (87/379, 22.96%).

Table 1. Clinical characteristics of 379 Klebsiella pneumoniae bloodstream infections cases.

Characteristics	Number (%)^§
Age, years	63 (51–71, 2–95)^¶
Age ≥ 65 years	160 (42.22)
Gender, male	270 (71.24)
Class, HA	313 (82.59)
Origin, primary	193 (50.92)
Origin, secondary	186 (49.08)
Source of BSIs
Respiratory tract infection	51 (27.42)
Urinary tract infection	9 (4.84)
Skin and soft tissue infection	35 (18.82)
Gastrointestinal, abdominal, and pelvic infection	86 (46.24)
Catheter-related infection	5 (2.69)
Polymicrobial BSIs	60 (15.83)
Ward
ICU	87 (22.96)
Surgical	174 (45.91)
Neurosurgery	1 (0.57)
Urological surgery	1 (0.57)
Burn unit	17 (9.77)
Cardiothoracic surgery	22 (12.64)
Transplantation unit	8 (4.60)
Medical	118 (31.13)
Cardiology	5 (4.24)
Gastroenterology	15 (12.71)
Gynaecology	1 (0.85)
Infection unit	15 (12.71)
Respiratory medicine	5 (4.24)
Dermatology	3 (2.54)
Neurology	5 (4.24)
Haematology	41 (34.75)
Oncology	4 (3.39)
Comorbidities
Agranulocytosis	6 (1.58)
Chemotherapy or radiotherapy	43 (11.35)
Malignancy	74 (19.53)
Disease of the circulatory system	138 (36.41)
Disease of the respiratory system	41 (10.82)
Endocrine, nutritional, and metabolic diseases	98 (25.86)
Chronic renal disease	41 (10.82)
Biliary tract and pancreas diseases	124 (32.72)
Burns	40 (10.55)
Healthcare exposure
Time at risk, days	12 (4–27, 0–210)^¶
Length of hospital stay, days	16 (6–32.50, 0–210)^¶
ICU stay	92 (24.27)
Length of ICU stay, days	0 (0–0, 0–205)^¶
Invasive procedures
Surgery	237 (62.53)
Paracentesis	112 (29.55)
Invasive ventilation	127 (33.51)
Indwelling catheterisation
AC	65 (17.15)
CVC	241 (63.59)
Urinary catheter	203 (53.56)
Gastric tube	127 (33.51)
Drainage tube	168 (44.33)
Drug usage
Corticosteroids	188 (49.60)
Immunosuppressor	33 (8.71)
Antibiotics	294 (77.57)

Abbreviations: BSIs, bloodstream infections; HA, hospital acquired; ICU, intensive care unit; AC, arterial catheter; CVC, central venous catheter.

^§ Number of cases with the characteristics, unless stated otherwise.

^¶ Median (interquartile range IQR, range).

Incidence

The incidence of KP BSIs remained stable between 8.52 and 8.72 per 100,000 patient-days in 2018–2020 followed by a significant decline in 2021 (4.77 per 100,000 patient-days, p = 0.0005) (Figure 1). The mean and standard deviation of incidence of KP BSIs is 7.99 per 100,000 patient-days and 1.65, respectively.

Figure 1. Incidence of Klebsiella pneumoniae bloodstream infections and percentage of antimicrobial resistant phenotypes between 2018 and 2022. Bar chart represents the incidence of K. pneumoniae BSIs. Scatter plot and line chart illustrate percentage of multidrug resistant (MDR) and carbapenem resistant (CR) K. pneumoniae isolates.

Antimicrobial resistance

Isolates with MDR and CR phenotypes accounted for 66.49% (252/379) and 51.98% (197/379), respectively. All isolates showed low resistance to tigecycline (13.77%) and polymyxin B (14.61%) (Figure 2). By contrast, resistance rate was the highest to cefuroxime (65.22%) (Figure 2). It is noteworthy that for all antibiotics tested, resistant rates of MDR isolates were higher than that of non-MDR isolates, and resistant rates of CR isolates were higher than that of non-CR isolates (Figure 2).

Figure 2. Radar plot of antimicrobial resistance rates of 379 K. pneumoniae isolates by antimicrobial resistant phenotypes. MDR, multidrug resistant; non-MDR, non-multidrug resistant; CR, carbapenem resistant; non-CR, non-carbapenem resistant.

Risk factors for crude 30-day mortality

The crude 30-day mortality rate of KP BSIs was 26.39%. Cases infected with MDR/CR isolates had significantly lower crude 30-day survival rates than these cases infected with non-MDR/non-CR isolates (Figure 3) (p < 0.001). Total length of hospital stay and length of hospital stay after KP isolation were significantly longer in cases with MDR/CR isolates than those with non-MDR/non-CR isolates (Table 2) (p < 0.001). Healthcare costs of cases with MDR/CR isolates were also significantly higher than cases with non-MDR/non-CR isolates (Table 2) (p < 0.001).

Figure 3. Survival (in days) of 379 Klebsiella pneumoniae bloodstream infections cases and comparison by antimicrobial resistant phenotypes. MDR, multidrug resistant; non-MDR, non-multidrug resistant; CR, carbapenem resistant; non-CR, non-carbapenem resistant.

Table 2. Comparison of total length of hospital stay and healthcare costs of 379 Klebsiella pneumoniae bloodstream infections cases by antimicrobial resistant phenotypes.

Variable	MDR versus non-MDR			CR versus non-CR
	MDR	non-MDR	P	CR	non-CR	P
Total length of hospital stay (days), median (IQR, range)^§	42.50 (26–74.50, 2–217)	23(13–42, 2–271)	<0.001	45 (28–77, 2–210)	27.50 (14–45.25, 2–271)	<0.001
Length of hospital stay after KP separation (days), median (IQR, range)	17 (7–37.75, 1–196)	14 (8–29, 1–250)	0.012	22 (6–43.50, 1–196)	14 (8–27.25, 1–250)	0.001
Healthcare costs (10,000 CNY), median (IQR, range)	30.19 (12.08–64.63, 0.65–407.02)	9.34 (4.07–20.22, 0.02–259.86)	<0.001	36.53 (17.44–73.83, 0.65–407.02)	10.54 (4.36–24.08, 0.02–259.86)	<0.001

Abbreviations: IQR, interquartile range; CNY, China Yuan; MDR, multidrug resistant; non-MDR, non-multidrug resistant; CR, carbapenem resistant; non-CR, non-carbapenem resistant.

^§ Total days of hospital stay for the index hospitalisation during which Klebsiella pneumoniae bloodstream infections occurred, that is, interval between hospital admission and dischargeof the index hospitalisation during which blood sample of the first positive culture for Klebsiella pneumoniae was collected.

The univariate analysis showed that age, MDR/CR phenotype, some medical exposures in the 90 days prior to the onset of BSI, cardiovascular disease, respiratory disease, several invasive procedures, use of certain medications in the 90 days before diagnosis of BSI, irrational empirical drug use and irrational targeted treatments were associated with crude 30-day mortality. The multivariate analysis showed that age, CR phenotype, paracentesis, indwelling CVC, carbapenems, tetracyclines, polymyxins and irrational empirical therapies were independent risk factors for crude 30-day mortality.

Discussion

This study provided a comprehensive analysis of the epidemiology of patients with KP BSIs and evaluated the incidence of and risk factors for mortality. To our knowledge, this is the first report to assess the epidemiology of KP BSIs in eastern China which is an area with high prevalence of resistant and virulent isolates in recent years.

The incidence of KP BSIs fluctuated between 9.40 and 4.77 per 100,000 patient-days during 2018–2022. The coronavirus disease 2019 (COVID-19) pandemic began in 2020, and some literature suggests that incidence of BSIs increased during COVID-19 pandemic [25]. However, maybe due to strict controls by the Chinese government and the closed-loop transfer of COVID-19 patients to sentinel hospitals, the incidence of KP BSIs reduced by nearly half to 4.77 per 100,000 patient-days in 2021. In contrast, the incidence of KP BSIs returned to pre-pandemic levels with the downgraded management of COVID-19 in 2022.

The crude 30-day mortality rate of KP BSIs patients was 26.39%, which is much higher than Japan [26]. Therefore, it is necessary to evaluate risk factors for mortality of KP BSIs. Consistent with other study, advanced age was found to be an independent variable that can significantly increase crude 30-day mortality in this study as it represented a vulnerable population with several underlying diseases for drug resistant pathogen infections [27]. Among comorbidities, ischaemic heart disease and pulmonary infection were associated with mortality. Bacteraemia which was an important driver of a fatal outcome generally begins with respiratory tract infections [28]. Moreover, this study showed that 27.42% of all the cases stemmed from respiratory tract infection (Table 1). Similar to other studies, time at risk, length of hospital stay and ICU stay were associated with crude 30-day mortality which may be explained by the increased risk of contracting resistant bacterial infection over time and antimicrobial selection pressure due to antibiotics used for infections at other sites given that hospital is a reservoir of resistant bacteria [29, 30]. As reported in the literature [31], invasive procedures are independent risk factors for mortality, our study also found that patients who experienced invasive procedures had a higher mortality rate. Previous studies have shown that intravenous catheter insertion was strongly associated with disease severity and increased the risk of poor prognosis [28, 31]. Paracentesis such as thoracentesis and abdominal paracentesis were independent risk factors for CRKP infections and associated with higher mortality rate [28]. Also, paracentesis was an independent risk factor for 30-day mortality in this study. It is well-known that KP typically colonises the respiratory and gastrointestinal tract of the human body, and invasive procedures or devices could increase the probability of bacterial colonisation or infection by rupturing the mucous membranes of the respiratory and gastrointestinal tracts, thereby increasing the chances of KP entering the bloodstream [32]. It is noteworthy that invasive procedures should be interpreted with caution as they may present surrogate markers of critical illness rather than reflecting a direct association with mortality. Consistent with this interpretation, our study also found that ICU stay was associated with crude 30-day mortality (Table 3).

Table 3. Risk factors associated with crude 30-day mortality of 379 Klebsiella pneumoniae bloodstream infections cases.

Characteristics	Non-survivor (%)^§ N = 100	Survivor (%)^§ N = 279	Univariate		Multivariate
			OR (95%CI)	P	aOR (95%CI)	P
Age (years), median (IQR)	64 (55–73)	61 (50–70)	1.02 (1.00–1.03)	0.026	1.02 (1.00–1.05)	0.037
Age ≥ 65 years	49 (49.00)	111 (39.78)	1.45 (0.92–2.30)	0.110
Gender, male	78 (78.00)	192 (68.82)	1.61 (0.94–2.75)	0.083
Smoking	18 (18.00)	41 (14.70)	1.27 (0.69–2.34)	0.435
Alcohol drinking	16 (16.00)	32 (11.47)	1.47 (0.77–2.81)	0.245
Polymicrobial BSIs	17 (17.00)	43 (15.41)	1.12 (0.61–2.08)	0.709
MDR	86 (86.00)	166 (59.50)	4.18 (2.26–7.72)	<0.001
CR	79 (79.00)	118 (42.29)	5.13 (3.00–8.78)	<0.001	6.19 (1.94–19.76)	0.002
Monotherapy	7 (7.00)	40 (14.34)	0.45 (0.20–1.04)	0.062
Combination therapy	85 (85.00)	163 (58.42)	4.03 (2.22–7.34)	<0.001
Healthcare exposure
Time at risk (days), median (IQR)	19.50 (11–40.75)	10 (3–22)	1.03 (1.02–1.04)	<0.001	1.012 (0.990–1.034)	0.282
Length of hospital stay (days), median (IQR)	25 (11.25–48.50)	14 (4–28)	1.02 (1.01–1.03)	<0.001
ICU stay	40 (40.00)	52 (18.64)	2.91 (1.76–4.80)	<0.001	1.20 (0.57–2.47)	0.627
Length of ICU stay (days), median (IQR)	0 (0–15.75)	0 (0–0)	1.02 (1.01–1.03)	0.001
Comorbidities
Agranulocytosis	3 (3.00)	3 (1.08)	2.85 (0.57–14.34)	0.205
Chemotherapy or radiotherapy	14 (14.00)	29 (10.39)	1.40 (0.71–2.78)	0.331
Malignancy	41 (41.00)	106 (37.99)	1.13 (0.71–1.81)	0.597
Disease of the circulatory system	48 (48.00)	90 (32.26)	1.94 (1.22–3.09)	0.005	1.26 (0.63–2.52)	0.506
Hypertension	37 (37.00)	80 (28.67)	1.46 (0.90–2.37)	0.123
IHD	15 (15.00)	22 (7.89)	2.06 (1.02–4.15)	0.043
Disease of the respiratory system	18 (18.00)	23 (8.24)	2.44 (1.26–4.75)	0.008	1.32 (0.57–3.00)	0.515
pulmonary infection	13 (13.00)	16 (5.73)	2.46 (1.14–5.31)	0.022
Endocrine, nutritional, and metabolic diseases	32 (32.00)	66 (23.66)	1.52 (0.92–2.51)	0.103
Diabetes mellitus	24 (24.00)	55 (19.71)	1.29 (0.75–2.22)	0.366
Chronic renal disease	13 (13.00)	28 (10.04)	1.34 (0.66–2.70)	0.414
Biliary tract and pancreas diseases	33 (33.00)	91 (32.62)	1.02 (0.63–1.66)	0.994
Burns	8 (8.00)	32 (11.47)	0.67 (0.30–1.51)	0.335
Invasive procedures
Surgery	56 (56.00)	181 (64.87)	0.69 (0.43–1.10)	0.117
Paracentesis	47 (47.00)	65 (23.30)	2.92 (1.09–4.72)	<0.001	2.22(1.02–4.41)	0.043
Invasive ventilation	48 (48.00)	79 (28.32)	2.34 (1.46–3.74)	<0.001	1.13 (0.52–2.46)	0.752
AC	25 (25.00)	40 (14.34)	1.99 (1.13–3.50)	0.016	1.08 (0.49–2.32)	0.844
CVC	78 (78.00)	163 (58.42)	2.52 (1.49–4.29)	0.001	2.45(1.04–5.79)	0.041
Urinary catheter	60 (60.00)	143 (51.25)	1.43 (0.90–2.27)	0.133
Gastric tube	43 (43.00)	84 (30.11)	1.75 (1.09–2.81)	0.020	0.67 (0.33–1.33)	0.260
Drainage tube	53 (53.00)	115 (41.22)	1.61 (1.02–2.55)	0.043
Drug use
Corticosteroids	66 (66.00)	122 (43.73)	2.50(1.55–4.02)	<0.001	1.09 (0.55–2.15)	0.796
Immunosuppressor	11 (11.00)	22 (7.89)	1.44 (0.67–3.10)	0.345
Antibiotics	91 (91.00)	203 (72.76)	3.79 (1.82–7.89)	<0.001	0.84 (0.31–2.31)	0.726
Total quantity (DDD), median (IQR)	26.09 (6.22–50.38)	6 (0–21)	1.01 (1.00–1.01)	0.016	0.99 (0.98–1.00)	0.120
Glycopeptides	50 (50.00)	88 (31.54)	2.17 (1.36–3.46)	0.001	0.84 (0.43–1.64)	0.614
Quantity (DDD), median (IQR)	0.13 (0–5)	0 (0–1.25)	1.01 (0.99–1.04)	0.310
Carbapenems	75 (75.00)	125 (44.80)	3.70 (2.22–6.16)	<0.001
Quantity (DDD), median (IQR)	7.75 (0.03–17.19)	0 (0–5)	1.05 (1.03–1.07)	<0.001	1.04 (1.00–1.08)	0.044
Macrolides	0 (0.00)	5 (1.79)		0.999
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)		0.990
Fluoroquinolones	30 (30.00)	37 (13.26)	2.80 (1.62–4.86)	<0.001	1.81 (0.90–3.61)	0.093
Quantity (DDD), median (IQR)	0 (0–3.75)	0 (0–0)	1.07 (0.03–1.12)	0.002
Cephalosporins	46 (46.00)	104 (37.28)	1.43 (0.90–2.28)	0.127
Quantity (DDD), median (IQR)	0 (0–3)	0 (0–2)	1.02 (1.02–1.06)	0.034	1.01 (0.98–1.05)	0.538
Penicillins	3 (3.00)	6 (2.15)	1.41 (0.35–5.74)	0.634
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	1.07 (0.92–1.25)	0.379
β-lactam/β-lactamase inhibitor combinations	10 (10.00)	33 (11.83)	0.83 (0.39–1.75)	0.621
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	1.01 (0.98–1.05)	0.046	1.01 (0.96–1.05)	0.748
Aminoglycosides	13 (13.00)	18 (6.45)	2.17 (1.02–4.60)	0.044	0.73 (0.25–2.01)	0.554
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	1.04 (0.99–1.11)	0.108
Sulfonamides	0 (0.00)	1 (0.36)		1.000
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)		1.000
Nitrofurans	3 (3.00)	6 (2.15)	1.40 (0.35–5.74)	0.634
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	1.02 (0.97–1.08)	0.428
Linkeramides	2 (2.00)	6 (2.15)	0.93 (0.18–4.68)	0.928
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	0.98 (0.88–1.10)	0.777
Tetracyclines	24 (24.00)	22 (7.89)	3.69 (1.96–6.94)	<0.001	2.82 (1.06–7.45)	0.037
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	1.00 (1.00–1.01)	0.952
Polymyxin B	6 (6.00)	5 (1.79)	3.50 (1.04–11.73)	0.042	5.72(1.01–32.46)	0.049
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	1.01 (0.99–1.03)	0.165
Oxazolidinones	27 (27.00)	37 (13.26)	2.42 (1.38–4.24)	0.002	0.96 (0.45–2.02)	0.925
Quantity (DDD), median (IQR)	0 (0–1)	0 (0–0)	1.06 (1.01–1.12)	0.028
Fosfomycin	6 (6.00)	11 (3.94)	1.56 (0.56–4.32)	0.397
Quantity (DDD), median (IQR)	0 (0–0)	0 (0–0)	0.99 (0.93–1.06)	0.846
Antimicrobial therapy after BSIs onset
Inappropriate empiric therapy	60 (60.00)	77 (27.60)	3.94 (2.44–6.35)	<0.001	2.95 (1.24–7.03)	0.015
Inappropriate definitive therapy	39 (39.00)	30 (10.75)	5.31 (3.06–9.22)	<0.001

Note: OR (95%CI), odds ratio (95% confidence interval); aOR (95%CI), adjusted odds ratio (95% confidence interval); IQR, interquartile range; BSIs, bloodstream infections; MDR, multidrug resistant; non-MDR, non-multidrug resistant; CR, carbapenem resistant; non-CR, non-carbapenem resistant; IHD, ischaemic heart disease; ICU, intensive care unit; AC, arterial catheter; CVC, central venous catheter; DDD, defined daily dose.

^§ Number of non-survivors/survivors with the characteristics (percentage of non-survivors/survivors with the characteristics), unless stated otherwise.

Our results showed a high rate of resistance in KP, except for two drugs, tigecycline and polymyxin B, which are consistent with other studies [20]. Polymyxin B and tigecycline are often used as last resort drugs for CR infections while carbapenems are often used as last resort drugs for MDR infections [33]. Use of polymyxin B, tetracyclines and amount of carbapenems 90 days before diagnosis of KP BSIs were determined as independent risk factors for crude 30-day mortalityin this study (Table 3). Majority of tetracyclines used in this study was tigecycline (39/46, 84.78%). This may be explained by the fact that, non-survivors were more extensively treated with these antibiotics than survivors as the proportion of non-survivors with CR/MDR infections were significantly higher than that of survivors (Table 3). Moreover, combination therapy including tigecycline, polymyxins and carbapenems were reported to be more effective than active monotherapy for KP BSIs [34, 35]. However, antimicrobial toxicity of combination therapy may increase the probability of adverse effects and the risk of mortality afterwards [36]. Additionally, use of carbapenems may induce the production of carbapenemases, which is one of the main resistant mechanisms of CRKP [37]. Again, this could be explained by critical illness as these antibiotics are last-line therapeutic options for MDR/CR infections.

The crude 30-day mortality rate increased to 34.12% in MDRKP BSI patients and higher in CRKP BSI patients (40.10%). CR phenotype was an independent risk factor for crude 30-day mortality. In this study, the majority of CR isolates (n = 197) are also MDR isolates (n = 194). MDR and CR phenotypes were widely reported to be predictors of poorer prognosis due to quite few effective therapeutic options for treatment [22, 38, 39]. We found that irrational empirical antibiotic therapy was an independent predictor for 30-day mortality. Empirical antibiotics are started before the availability of microbiological results. This is vital because inappropriate, inadequate or delayed administration of active antibiotics has been reported to be associated with increased mortality [4, 22, 34]. Therefore, better antibiotic stewardship and appropriate drug use is critical.

Interestingly, we found that total days of hospitalisation, days of hospitalisation after KP separation, and total hospitalisation costs of cases with MDR/CR phenotype were significantly higher than those of cases with non-MDR/non-CR phenotype. It is well-known that prolonged hospitalisation is a risk factor for drug-resistant bacterial infections [40]. Approximately 10% of hospitalisations are complicated by healthcare-associated infections, and up to 75% of infections are due to microorganisms resistant to first-line therapeutic antibiotics [41]. Increased medical costs are due to higher drug costs for MDR/CR BSIs, increased frequency of procedures such as intravenous antibiotics, and increased risk of complications [42]. Drug-resistant bacterial infections make the disease more severe and complex, making treatment more difficult and ultimately leading to longer hospital stays and higher healthcare costs.

This study has some limitations. First, it is a retrospective, single-center study that may subject to selection and recall bias. Second, we have not molecularly characterised these clinical isolates to dissect their potential resistance mechanisms and virulence which may contribute to high mortality. Bioinformatics and phylogenetic studies are needed in future studies to better understand the phylogeny and pathogenicity of these isolates.

Conclusion

The incidence of KP BSIs was fluctuating during the study period. The crude 30-day mortality rate of KP BSIs patients was 26.39%. Awareness is required to prevent widespread of highly resistant KP. KP BSIs cases with non-MDR/non-CR phenotypes had lower mortality, shorter hospital stay, and lower healthcare costs compared to those of cases with MDR/CR phenotypes. Age, CR phenotype, paracentesis, CVC, exposure to some antibiotics, and irrational empirical treatment are independent risk factors for mortality.

Ethics approval and consent to participate

The study was approved by the ethics committee of Ruijin Hospital (KY2023-083). Because of the retrospective nature of the study, the committee waived informed consent. Patient data was obtained from medical record systems and analysed anonymously to protect patient privacy.

Acknowledgement

We would like to thank all the technicians of Department of Clinical Microbiology of Ruijin Hospital for their assistance in data collection.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The datasets used during the current study are available from the corresponding author on reasonable request.

Additional information

Funding

None.