Risk Factors and Mortality of Elderly Patients with Hospital-Acquired Pneumonia of Carbapenem-Resistant Klebsiella pneumoniae Infection

Introduction

Hospital-acquired pneumonia (HAP) is an infection of lung parenchyma caused by pathogens within the hospital environment. It is the second-most common nosocomial infection and the leading cause of death.¹ The incidence of HAP varies from 5 to more than 20 cases per 1000 hospitalized patients.² The incidence of HAP in Chinese is 1.74% to 6.4%.^3,4 Elderly patients are more likely to develop severe pneumonia due to decreased immunity, vomiting reflex disorders, weakened fever response, and cardiopulmonary dysfunction.

The most common cause of HAP in the elderly is bacterial infection, particularly K. pneumoniae, a gram-negative capsulated bacterium that can cause severe pneumonia, especially in immunocompromised patients.⁵ Since carbapenems were first used to treat infections caused by K. pneumoniae, the emergence of CRKP has become a major concern worldwide.⁶ In 2017, the World Health Organization listed CRKP as one of the resistant pathogens in urgent need of new antimicrobial agents.⁷ The number of CRKP had increased significantly year by year. According to the China Antimicrobial Surveillance Network (CHINET, http://www.chinets.com/), the isolation rate of CRKP based on meropenem resistance increased significantly from 2.9% in 2005 to 27.5% in 2022.

The mortality rate of patients with CRKP pulmonary infection was high. A previous study showed that the 14-day mortality rate of patients with CRKP pulmonary infection was 23.2%, and the in-hospital mortality rate was 34.8% (73.9% of carbapenem-resistant Enterobacterales were CRKP).⁸ Risk factors associated with mortality of CRKP infection included older age, malignancy, high SOFA score, and inappropriate therapy.^8–10 Additionally, several studies had demonstrated that CRKP infected patients had higher mortality and longer hospital stays than CSKP infected patients.^11,12

However, it is worth noting that there are few studies researching the risk factors and mortality of elderly patients with CRKP infection. The objective of this study is to identify the risk factors and mortality associated with CRKP pulmonary infection in the elderly.

Materials and Methods

Study Design

This retrospective case–control study was conducted from 1 January 2019 to 31 December 2021 at Peking University First Hospital, a tertiary hospital with 1805 beds in Beijing, China. Elderly inpatients (≥65 years) with confirmed with K. pneumoniae HAP were recruited. K. pneumoniae HAP was defined as a new pulmonary infiltrate confirmed by X-ray or computed tomography (occurred ≥48 h after admission) associated with at least one of the following: new or worsening cough with or without sputum production, fever (temperature >37.8◦C) or hypothermia (<35.6◦C), leukocytosis, left shift, or leukopenia based on local normal values, as well as isolation of K. pneumoniae from lower respiratory tract secretions (sputum, bronchoalveolar lavage fluid, bronchoscopic needle aspiration, or blood).¹³ The exclusion criteria were: a) age <65 years; b) incomplete medical records; c) K. pneumoniae colonization (did not meet the criteria for infection);¹⁴ d) outpatients or emergency patients; e) infection acquired from outside of hospital setting. Only the first episode of K. pneumoniae HAP was included. A total of 193 episodes were enrolled in this study.

Data Collection

The patient data were collected from the hospital’s computerized microbiology laboratory database and information system. The main variables included demographic characteristics; underlying diseases; Charlson comorbidity index;¹⁵ past medical history (surgery, smoke, alcohol, exposure to antibiotics, and K. pneumoniae infection or colonization); sepsis/septic shock,¹⁶ Pitt bacteremia score,¹⁷ and quick Sepsis-related Organ Failure Assessment (qSOFA) score¹⁶ at infection onset; laboratory findings; co-infection with other pathogens; therapeutic processes (admission to ICU after infection, mechanical ventilation); antibiotics therapy; clinical outcomes (30-day mortality, failure rate of microbiological clearance, length of hospital stay, length of ICU stay). The events that occurred before CRKP infection were considered for analyzing risk factors. Thirty-day mortality was defined as death within 30 days from diagnosis of infection (including patients who died within 30 days, and patients who discharged within 30 days were confirmed dead by telephone follow-up). To further compare the clinical outcomes between CRKP and CSKP groups, propensity score matching was used to minimize the effect of potential confounding variables, including age, admission to ICU before infection, and Charlson comorbidity index.

Definition of Terms

Patient conditions to be considered immunocompromised were: a) primary immune deficiency diseases; b) active malignancy or malignancy within 1 year of community-acquired pneumonia, excluding patients with localized skin cancers or early-stage cancers; c) receiving cancer chemotherapy; d) human immunodeficiency virus infection with a CD4 T-lymphocyte count <200 cells/mL or percentage <14%; e) solid organ transplantation; f) hematopoietic stem cell transplantation; g) receiving corticosteroid therapy with a dose ≥20 mg prednisone or equivalent daily for ≥14 d or a cumulative dose >600 mg of prednisone; h) receiving biological immune modulators; i) receiving disease-modifying antirheumatic drugs or other immunosuppressive drugs (eg, cyclosporin, cyclophosphamide, hydroxychloroquine, methotrexate).¹⁸ Severe pneumonia was diagnosed if one or more major criteria (invasive mechanical ventilation or septic shock with the need for vasopressors) and/or three minor criteria [respiratory rate ≥30 breaths/min, PaO2/FiO2 ratio ≤250, multilobar infiltrates, confusion/disorientation, uremia (blood urea nitrogen level ≥20 mg/dL), leukopenia (white blood count <4000 cells/mm³), thrombocytopenia (platelet count <100,000 cells/mm³), hypothermia (core temperature <36°C), hypotension requiring aggressive fluid resuscitation] met.¹⁹ Failure rate of microbiological clearance was defined as the persistence of positive cultures (presence of pathogens initially isolated from the site of infection).^20,21 The definition of co-infection was defined as detection 2 or more bacteria in the same sample. Longer hospitalization time was defined as the patient’s hospital stay more than 30 days after the first positive culture.

Empirical and targeted therapies were defined as antibiotics administered before and after obtaining the result of an antimicrobial susceptibility test. Appropriate therapy was defined as using at least one in vitro active agent, with or without inactive companion agents. Monotherapy was defined as only one active agent using, combination therapy was defined as more than one active agent using. Inappropriate therapy was defined as selection of agents without activity against responsible pathogens.²²

Antimicrobial Susceptibility Test

K. pneumoniae was identified by Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (Bruker Microflex LT, Bremen, Germany).²³ Antibiotics susceptibility testing was completed using a Vitek 2 automated system (bio-Mérieux, France), which was performed by adopting the broth microdilution method. Gram-negative susceptibility (GNS) cards were used to determine the minimal inhibitory concentration (MIC) of strains for various antibiotics (8–64 ug/mL for amikacin, 2–32 ug/mL for aztreonam, 0.25–32 ug/mL for cefepime, 8–32 ug/mL for cefoperazone/sulbactam, 0.25–32 ug/mL for ceftazidime, 0.5–4 ug/mL for ciprofloxacin, 1–16 ug/mL for doxycycline, 1–12 ug/mL for imipenem, 0.25–8 ug/mL for levofloxacin, 0.5–12 ug/mL for meropenem, 2/4-48/8 ug/mL for piperacillin/tazobactam, 2–8 ug/mL for minocycline, 0.75–4 ug/mL for tigecycline, 8–64 ug/mL for tobramycin, 8/2-64/2 ug/mL for ticarcillin/clavulanate and 1/19-16-304 ug/mL for trimethoprim/sulfamethoxazole). Antibiotic susceptibility testing for colistin was performed using the test kit (BIO-KONT^®), with the method of broth microdilution. Susceptibility testing results were interpreted according to the criteria recommended by the Clinical and Laboratory Standards Institute (CLSI M100 31th),²⁴ except tigecycline, which was interpreted according to the American Food and Drug Administration (FDA) standard (https://www.fda.gov/drugs/development-resources/tigecycline-injection-products).

Statistical Analysis

SPSS version 26.0 and MatchIt package in R 3.5 were used for statistical analyses. For continuous variables, patient characteristics were presented as median and IQR (interquartile range), Student’s t-test was used to compare normally distributed variables, and Mann–Whitney U-test was used to compare non-normally distributed variables. For categorical variables, patient characteristics were presented with rate. Chi-square test or Fisher’s exact test was used to compare categorical variables. In the multivariate logistic regression model, variables with P <0.01 in the univariate analysis were analyzed. Variables whose P <0.05 and OR >1 was considered as independent risk factors. Additionally, OR and the 95% confidence interval (CI) were calculated for each variable.

Results

A total of 193 patients were included in this study, including 115 patients with CRKP infection, and 78 patients with CSKP infection (Figure 1). The mean age was 76.0 years old, and the age ranged from 65 to 99 years. 66.8% (129/193) of patients were male, 33.2% (64/193) of patients were female.

Figure 1 Flowchart outlining of the patients included in this study.

Results of K. pneumoniae Susceptibility Test

Detailed findings for susceptibilities of CRKP and CSKP are shown in Figure 2. In vitro antimicrobial susceptibility test showed that CRKP strains were significantly resistant to ciprofloxacin, piperacillin-tazobactam, aztreonam, ceftazidime, levofloxacin, cefoperazone, cefepime, minocycline, doxycycline, tobramycin, and amikacin, the resistance rate was 50%–100%. The proportion of CRKP resistant to tigecycline and colistin was 5.9% and 14.8%, respectively. CSKP strains were sensitive to most antibiotics.

Figure 2 The results of antimicrobial susceptibility test of K. pneumoniae strains.

The Risk Factors for CRKP Infection

The demographic and clinical characteristics of patients are summarized in Table 1. The majority of patients infected with CRKP (77/115, 67.0%) and CSKP (52/78, 66.7%) were male. Patients with CRKP infection (median age = 79.0 years) were older than patients with CSKP infection (median age = 73.0 years). Patients with history of ICU stays from hospital admission to positive respiratory specimen culture for K. pneumoniae, Charlson comorbidity index score ≥3, long hospital stay before infection, prior K. pneumoniae infection or colonization in the preceding 3 years, exposure to carbapenems or tigecycline within 3 months before a positive culture were more likely to be infected with CRKP.

Table 1 Univariate Analysis of Risk Factors for CRKP Infection

Multivariate logistic regression analysis showed that independent risk factors for CRKP infection were history of ICU stays from hospital admission to positive respiratory specimen culture for K. pneumoniae (OR = 2.530, 95% CI: 1.195–5.356; P=0.015), Charlson comorbidity index score ≥3 (OR = 2.420, 95% CI: 1.097–5.338; P=0.029), prior exposure to carbapenems (OR = 5.280, 95% CI: 2.330–11.966; P<0.001), and prior K. pneumoniae infection or colonization in the preceding 3 years (OR = 18.529, 95% CI: 3.801–90.336; P<0.001) (Table 2).

Table 2 Multivariate Analysis of Risk Factors for CRKP Infection

The Clinical Outcomes Between CRKP and CSKP Infected Patients

The all-cause 30-day mortality was 22.3% (43/193) in all patients, the 30-day mortality of CRKP infection was 28.7% (33/115), and the 30-day mortality of CSKP infection was 12.8% (10/78). The all cause in-hospital mortality was 41.5% (80/193) in all patients, the in-hospital mortality of CRKP infection was 53.9% (62/115), and the in-hospital mortality of CSKP infection was 23.1% (18/78). Patients with CRKP infection had higher mortality than patients with CSKP infection (P=0.009). Compared with patients with CSKP infection, patients with CRKP infection were more likely to have sepsis/septic shock (29/78 vs 8/73, P<0.001), severe pneumonia (31/78 vs 12/78 P=0.001), and long hospital stay (40/78 vs 20/78, P<0.001). In addition, CRKP strains were more difficult to clear up than CSKP strains (49/78 vs 16/78 P<0.001) (Figure 3).

Figure 3 The clinical outcomes of patients with CRKP and CSKP infection.

The Risk Factors for All-Cause 30-Day Mortality

In univariate analysis, the following factors were associated with 30-day mortality, including history of ICU stays from hospital admission to positive respiratory specimen culture for K. pneumoniae, immunocompromised patients, exposure to carbapenems or tigecycline within 3 months prior to a positive culture, Pitt bacteremia score ≥3, qSOFA ≥2, received mechanical ventilation treatment, sepsis/septic shock occurred, ICU admission after infection, severe pneumonia, CRKP infection, and prolonged ICU stay. Some laboratory variables, such as high procalcitonin, sensitivity C-reactive protein, D-dimer, neutrophils, and low lymphocytes, were also associated with mortality (Table 3).

Table 3 Univariate Analysis of Risk Factors for 30-Day Mortality

Multivariate logistic regression showed that older age (OR = 1.107; 95% CI: 1.016–1.206; P=0.020), immunocompromised status (OR = 8.632; 95% CI: 1.926–38.680; P=0.005), severe pneumonia (OR = 51.244; 95% CI, 10.181–257.918; P<0.001), qSOFA score ≥2 (OR = 6.187; 95% CI, 1.445–26.481; P=0.014), exposure to tigecycline before infection (OR = 24.702; 95% CI, 2.418–252.408; P=0.007), and prolonged ICU stay (OR = 0.987; 95% CI, 0.977–0.997; P=0.013) were independent risk factors for 30-day mortality (Table 4). Kaplan–Meier analyses also confirmed that 30-day mortality was higher in patients with CRKP infection than patients with CSKP infection (Figure 4A). The patients with severe pneumonia had higher 30-day mortality than patients without severe pneumonia (P<0.001) (Figure 4B). The 30-day mortality of elderly patients with severe pneumonia was 62.9%. In addition, immunocompromised patients had higher 30-day mortality than immunocompetent patients (P<0.001) (Figure 4C). The 30-day mortality of patients with immunocompromised pneumonia was 42.3%.

Table 4 Multivariate Analysis of Risk Factors for 30-Day Mortality

Figure 4 Kaplan–Meier analysis was used to compare the 30-day survival rate for (A) CRKP vs CSKP infection; (B) severe pneumonia vs non-severe pneumonia; (C) immunocompromised vs immunocompetent patients, and (D) patients receiving CAZ-AVI containing regimes vs polymyxin B sulfate (PB) containing regimes. The data were analyzed using SPSS 26 to obtain p-values (SPSS 26_analyze_survival_Kaplan-Meier_log-rank_p value), and the drawing was made using Prism 8 app.

Antimicrobial Therapy

Among 193 patients included, 122 patients received appropriate targeted therapy. Among those received appropriate targeted therapy, 102 patients received monotherapy, 20 patients received combination therapy. Tigecycline and carbapenems were the most commonly used monotherapy options for patients with CRKP and CSKP infection, respectively. In combination therapy, amikacin combined with tigecycline, and PB combined with tigecycline were common treatment options (Supplemental Table 1). In particular, 30-day mortality was significantly lower in patients received CAZ-AVI containing regimens than in patients received PB containing regimens (25.0% vs 64.3%, P=0.048) (Figure 4D).

Relationship Between qSOFA Score and the Prognosis of Patients

The distribution of qSOFA score and its relationship with 30-day mortality are shown in Figure 5. When qSOFA score was 0 point, the 30-day mortality rate of patients was 3.0% (2/65), qSOFA score was 1 point, the 30-day mortality of patients was 8.0% (4/50), qSOFA score ≥2, the 30-day mortality of patients was 47.4% (37/78). Among the died, 86.0% of patients had a qSOFA ≥2, the AUROC was 0.838.

Figure 5 Distribution and mortality of patients by qSOFA score. (A) Distribution of qSOFA score (0–3) and its association with 30-day mortality (the x axis indicated the specific value of the qSOFA score when the patient developed K. pneumoniae infection) (B) the area under receiver operating characteristic curve (AUROC) predicted the prognosis of patients using qSOFA score. The (A) was made using Prism 8 app., the (B) was made using SPSS 26 analysis (SPSS 26_analyze_category_ROC curve).

Discussion

HAP caused by CRKP infection was commonly reported to be associated with higher mortality,²⁵ but data regarding risk factors and clinical outcomes for CRKP pulmonary infection in the older patients was still limited. Therefore, this purpose of this study was to systematically review HAP caused by CRKP infection, especially the risk factors of morbidity and mortality and clinical outcomes, so as to provide clinical value for the treatment and control of CRKP infection.

Early identification of people at high-risk of CRKP infection is important. Previous studies demonstrated that prior central venous catheter use, sputum suction, continuous renal replacement therapy, exposure to fluoroquinolones,²⁶ hypertension, exposure to carbapenems, ICU stay,²⁷ hematologic malignancies, ICU acquired infection,²⁸ β-lactam/β-lactamase inhibitor combinations exposure, solid organ transplantation,²⁹ liver cirrhosis, and length of hospital stay over the previous 14 days³⁰ were independent risk factors for the morbidity of CRKP infection. In this study, we found that history of ICU stays from hospital admission to positive respiratory specimen culture for K. pneumoniae, Charlson comorbidity index score ≥3, prior exposure to carbapenems, and prior K. pneumoniae infection or colonization in the preceding 3 years were independent risk factors for CRKP infection.

ICU patients were more likely to undergo invasive procedures, which may hinder airway clearance, increase the risk of CRKP colonization. In addition, such invasive measures exposed ICU patients to pathogens carried by roommates or presented in the hospital environment. Therefore, the use of invasive devices should be minimized for the elderly. In humans, K. pneumoniae often colonizes at various mucosal surfaces, including the upper respiratory tract and the gut. In hospitalized patients, colonization rates in the nasopharynx rise to 19%, while it can be as high as 77% in the gastrointestinal tract.^31,32 Gastrointestinal colonization serves as a major reservoir for transmission and infection to other sites. Previous study found that 16% of the patients colonized with K. pneumoniae were found to be infected, compared to only 3% in the non-carriers, whole-genome sequencing revealed that the patients were infected with the same strain they carried in the form of colonization.³³ This indicated that K. pneumoniae colonization in high-risk units (such as intensive care, transplant surgery, hematology oncology wards) at admission should be actively monitored and continuously monitored during subsequent hospitalization.

Comparing with CSKP infected patients, CRKP infected patients had higher mortality. Regarding risk factors for mortality, different populations, regions and types of infection could contribute to different results. Previous studies showed that treatment failure for infection,²⁶ CRKP infection and isolation in ICU,²⁷ severe sepsis or septic shock incidents, inadequate empirical antimicrobial therapy, corticosteroids use,²⁸ CRKP bloodstream infections, mechanical ventilation >96 hours, and platelet counts <100×10⁹/L²⁹ were independent risk factors for mortality of CRKP infection. In this study, we found that the independent risk factors for mortality included older age, immunocompromised status, severe pneumonia, qSOFA score ≥2, empirical use of tigecycline, and prolonged ICU stay.

It has been demonstrated that age ≥65 years was an independent risk factor for hospital mortality of patients with pneumonia.³⁴ The incidence of pneumonia increased rapidly with age and peaked between 60 and 79 years of age.³⁵ Immunocompromised patients were prone to acquire pneumonia with high mortality. Different from the general immunocompetent population, the etiologies of lung infection in immunocompromised patients were more diverse, and the possibility of co-infection should be considered. Due to the failure to implement early and rapid diagnosis, many patients were initially treated with inappropriate antibiotics, which may cause several serious comorbidities such as respiratory failure and infectious shock, further increasing mortality.³⁶ In this study, empirical use of tigecycline before infection was associated with death. This is probably because the standard dose of tigecycline (50 mg of tigecycline every 12 h after a 100-mg loading dose) was absolutely insufficient to eradicate strains with tigecycline MIC >0.5 mg/L.³⁷ In this study, the tigecycline MIC >0.5 mg/L in 71.3% of the strains. With suboptimal pharmacokinetics properties in certain sites of infection increased potential for toxicity. High doses of tigecycline (received 100 mg of tigecycline every 12 h after a 200-mg loading dose) and combination therapy can be considered for HAP caused by multidrug-resistant bacterial infections, but caution is needed for its side effects, such as gastrointestinal.³⁸

In addition, treatment options against CRKP infections are very limited. Clinicians are compelled to use last-resort antimicrobials, like CAZ-AVI and PB. In this study, we found that the 30-day mortality was significantly lower in patients received CAV-AZI containing regimens than patients received PB containing regimens, providing encouraging results. More recently, several new agents with activity against carbapenem-resistant pathogens have been approved for clinical use or are reaching late-stage clinical development, they include β-lactam-β-lactamase inhibitor combinations (ceftolozane-tazobactam, meropenem-vaborbactam, imipenem-cilastatin-relebactam), plazomicin, eravacycline, and cefiderocol. Clinicians appear to generally favor newer agent over previous best available therapy.³⁹

The knowledge of the exact mechanism of CRKP was crucial for the selection of the most appropriate antimicrobial.^39,40 For example, ceftazidime/avibactam was not active against strains producing class B metallo-β-lactamases (MBLs), the combination of aztreonam-ceftazidime/avibactam had a treating effect against 86% of MBLs producing members of Enterobacteriaceae family.⁴¹ In addition, the antibiotic resistance and virulence factors could co-exist in CRKP isolates.⁴² Antibiotic resistance and virulence genes can be transmitted horizontally through plasmids, which lead to the accelerated spread of multidrug resistance among different populations of pathogens.⁴³ Understanding the direct relationship between the frequency of the genes involved in the development of virulence and resistance, which could provide a highly effective model for physicians to prescribe suitable antibiotics.⁴⁴ Determining genetic diversity for the identification of dominant species and complete knowledge of biofilm formation were necessary to stop the spread of infections in hospitals and to control and manage associated infections.⁴⁵ The information on genetic resistance profile is not reported in this article, we have collected all strains, and will further study the genetic resistance profile of strains in the future.

qSOFA has been proposed as a screening and prognostic tool in patients with suspected infection, sepsis, or septic shock, it can be used to predict the prognosis of patients.¹⁶ In this study, we found that qSOFA score ≥2 could predict the mortality of patients with HAP infected with K. pneumoniae. Previous study had shown that the accuracy of qSOFA score in predicting in-hospital mortality of elderly hospitalized patients with suspected infection was low (AUROC = 0.680).⁴⁶ One plausible reason for this difference may be that this study only focused on older patients with HAP who were infected with K. pneumoniae. The accuracy of qSOFA score in predicting the prognosis of patients with HAP needs to be further verified by more studies in the future.

The present work has several limitations: first, the number of cases were relatively small; second, this study mainly focused on HAP caused by K. pneumoniae, and lacked analysis of K. pneumoniae infections in other sites; third, due to the limited data available in retrospective studies, some predictors may be missing from the data analysis.

Conclusion

In summary, history of ICU stays from hospital admission to positive respiratory specimen culture for K. pneumoniae, Charlson comorbidity index score ≥3, prior exposure to carbapenems, and prior K. pneumoniae infection or colonization in the preceding 3 years were associated with the development of CRKP pulmonary infection. We also found a high mortality caused by K. pneumoniae pulmonary infection in elderly patients with older age, immunocompromised status, severe pneumonia, qSOFA ≥2, exposure to tigecycline, and prolonged ICU stays. Ceftazidime-avibactam-based regimes were associated with better outcomes. Active screening of CRKP for high-risk groups, especially for elderly patients, is conducive to the early identification, treatment and control of CRKP infections, so as to achieve the successful management.