Association Between Red Cell Distribution Width and Mortality in Patients with Klebsiella pneumoniae Bloodstream Infection: A Cohort Study

Introduction

Klebsiella pneumoniae (KP) is a gram-negative, encapsulated bacillus whose global burden of severe infection remains substantial, with estimates of hundreds of thousands of attributable deaths annually.¹ KP pathogenicity is mediated by multiple virulence determinants including a polysaccharide capsule that impairs phagocytosis, lipopolysaccharide (LPS) that triggers strong inflammatory responses, siderophores that enhance iron acquisition, and various adhesins and secreted effectors that facilitate tissue invasion and immune evasion.² Hypervirulent and multidrug-resistant strains are associated with more severe clinical courses and higher mortality.^3,4 Bloodstream infection (BSI) due to KP (KP-BSI) commonly progresses to sepsis, septic shock and multiorgan dysfunction, with reported 30-day mortality rates frequently in the range of 20~50% depending on host factors and strain virulence.^3,5–7 Klebsiella pneumoniae bloodstream infections (KP-BSI) are a leading cause of sepsis in hospitalized patients, with mortality rates ranging from 20% to 50% depending on host factors and pathogen virulence Immunocompromised states, older adults, and chronic comorbidities (eg, diabetes, renal failure), delays in appropriate antimicrobial treatment, delayed appropriate antimicrobial therapy and infection with resistant strains further worsen outcomes.^8–12 These observations underscore the urgent need for improved prognostic stratification tools that can guide clinical management and optimize resource allocation.

Several biomarkers and clinical parameters, such as C-reactive protein (CRP), procalcitonin (PCT), and comorbidity burden, have been investigated for their prognostic value in KP-BSI.^13,14 However, these markers may not be consistently available, and their predictive accuracy remains limited. Red blood cell distribution width (RDW), a routine hematology parameter reflecting erythrocyte size heterogeneity, has emerged as an inexpensive, widely available marker associated with inflammation, oxidative stress and impaired erythropoiesis. In patients with bloodstream infection, the severe inflammatory responses elicited by bacterial infections can result in alterations to blood components, including red blood cells.¹⁵ Emerging evidence has established RDW as a novel prognostic marker across diverse clinical conditions. Particularly, recent investigations have demonstrated significant associations between elevated RDW levels and adverse outcomes in sepsis,¹⁶ sepsis-associated liver injury,¹⁷ coronavirus disease 2019 (COVID-19),¹⁸ heart failure,¹⁹ and non-alcoholic fatty liver disease.²⁰ Nevertheless, the prognostic relevance of RDW, in predicting in-hospital-mortality among KP-BSI patients remains insufficiently characterized in current clinical research.

Thus, in the cohort study, we explore whether it associated KP-BSI prognosis. This approach aims to assist physicians in promptly evaluating the patient’s condition and guiding further prognostic interventions. Our primary hypothesis posits that RDW levels demonstrate significant prognostic associations with clinical outcomes in patients with KP-BSI.

Materials and Methods

Data Source

This retrospective analysis was performed at Beijing Luhe Hospital, Capital Medical University (Beijing, China), targeting patients with Klebsiella pneumoniae bloodstream infections (KP-BSI). Beijing Luhe Hospital is a tertiary hospital with 1300 inpatient beds, located in eastern Beijing, China. The study encompassed a six-year period from January 2019 through December 2024, with all individuals meeting KP-BSI diagnostic criteria at this tertiary center being systematically enrolled. The study was conducted in accordance with the Declaration of Helsinki, and received approval from the Ethics Committee of Beijing Luhe Hospital, Capital Medical University (NO. 2025-LHKY-017-01). The ethics committee approved the retrospective study as anonymous, thus waiving informed consent. The studies were conducted in accordance with the local legislation and institutional requirements. We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines in reporting this study.²¹

Study Population

Clinical data of each patient were obtained from the electronic medical record system. This database was built using patients’ clinical records after removing any personally identifiable information. The criteria for inclusion were: 1) A diagnosis of KP-BSI was established when Klebsiella pneumoniae was isolated from at least one positive blood culture, with the samples preserved in our laboratory. Only patients with compatible clinical manifestations of bloodstream infection were included, while cases deemed to represent contamination were excluded. 2) age >18 years, 3) complete data of the complete blood cell analysis. For patients with multiple positive culture results, data from only their first culture was incorporated. Patients younger than 18 years, incomplete data, or unspecified infections, were excluded from the study. Figure 1 showed the flow chart of the study.

Figure 1 Flow chart of study. KP-BSI Klebsiella pneumoniae bloodstream infections.

Microbiological Tests

All clinical isolates underwent standardized species identification and antibiotic susceptibility profiling via the VITEK 2 Compact platform (bioMérieux, France). Antimicrobial susceptibility testing panels were calibrated in accordance with Clinical and Laboratory Standards Institute (CLSI-M100) interpretive criteria for standardized antibiotic resistance profiling.²²

Variable Extraction

RDW

RDW was evaluated and entered into the clinical records within 24 hours of blood culture. RDW was analyzed as both continuous and categorical variables. Patients were classified into 3 groups based on tertile of RDW: T1 < 43.3fL, T2 43.3–50.8 fL, T3> 50.8fL.

Covariates

Demographic data, such as gender and age, comorbidities included diabetes, septic shock, malignant tumor, hematological indices collected at the time of blood culture, including the C-reactive protein (CRP), creatinine, albumin, white blood cell (WBC) count, platelet, were assessed and recorded in the clinical records within 24 hours of blood culture. Treatments included continuous renal replacement therapy (CRRT), use of ventilator. Microorganism factors included carbapenem-resistant Klebsiella pneumoniae (CRKP). The severity of the disease includes the sequential organ failure assessment (SOFA) score. Other covariates included: intensive care unit (ICU) admission, nosocomial infection, infection site. Septic shock was defined according to the Sepsis-3 consensus as persistent hypotension requiring vasopressors to maintain a mean arterial pressure of ≥65 mmHg, together with a serum lactate level >2 mmol/L despite adequate fluid resuscitation.

Outcomes

The primary outcome was in-hospital mortality.

Statistical Analysis

Descriptive statistics were performed for all patients. Categorical variables were presented as counts and percentages, while continuous variables were reported as means with standard deviations (SD) for normally distributed data or as medians with interquartile ranges for skewed distributions. The chi-square test, one-way ANOVA, and Kruskal–Wallis test was employed to compare categorical, normally distributed, and non-normally distributed continuous variables, respectively. Missing covariates were less than 1%, so we did not perform any imputation. We conducted multivariable logistic regression analyses and to explore the independent relationship between RDW and in-hospital death. Covariates were selected for multivariable adjustment based on a combination of prior literature, statistical significance in univariate analyses, and clinical relevance. Model 1 was crude. Model 2 was adjusted for sex and age. Model 3 was adjusted further for Adjusted for sex, age, diabetes, malignant tumor, chronic kidney disease, central venous catheter placement (CVC), septic shock, white blood cell (WBC), HCT, PCT, albumin. based on Model 2. Tests for trend were performed by incorporating the median value of each tertile as a continuous variable in the models. Additionally, we performed subgroup analysis and sensitivity analysis to explore the effects within specific populations. The same set of covariates was consitently applied in multivariable logistic regression Model 3, as well as in Figure 2, subgroup and sensitivity analyses to ensure comparability and interpretability. To evaluate the discriminative ability of RDW for predicting in-hospital mortality, receiver operating characteristic (ROC) curve analysis was performed. The area under the ROC curve (AUC) and its 95% confidence interval were calculated to quantify the model’s predictive performance.

All analyses were performed using the R statistical software, version 4.3.2 (http://www.R-project.org, The R Foundation) and Free Statistics software version 2.2.0.²³ A two-tailed test was utilized, and a p-value of less than 0.05 was deemed statistically significant.

Table 1 Baseline Characteristics of KP-BSI Patients

Figure 2 Subgroup analyses of RDW associated with mortality in patients with KP-BSI. Odds ratios (ORs) were adjusted for sex, age, diabetes, malignant tumor, chronic kidney disease, CVC, septic shock, WBC, hematocrit, procalcitonin, albumin. Abbreviations: RDW, red cell distribution width; KP-BSI, Klebsiella pneumoniae bloodstream infections; CVC, central venous catheter placement; WBC, white blood cell; CRKP, carbapenem-resistant Klebsiella pneumoniae; CRRT, continuous renal replacement therapy.

Results

Baseline Characteristics of Participants

The characteristics of participants with KP-BSI were analyzed according to the tertiles of RDW (Table 1). Within the original cohort 267 patients were included in our study. Among all these patients, the average age of the participants was 65.9 years and 67.5% were men (Table 1). A total of 122 patients died in, resulting in a mortality rate of 45.7%. Non-survivors were older and had higher rates of CRKP infection, septic shock, ICU admission, CRRT, and mechanical ventilation. They also had higher RDW, CRP, PCT, serum creatinine, and SOFA scores, and lower albumin levels (Supplementary table S1).

Table 2 Association Between RDW and Morality in Patients with KP-BSI

The Association Between RDW and In-Hospital Mortality

Table 2 presents the findings from the multivariable logistic regression analysis regarding the association between RDW and mortality in KP-BSI patients. When RDW was examined as a continuous factor, a significant independent positive relationship was observed between RDW and mortality in the unadjusted model (OR: 1.06, 95% CI:1.03–1.09; p < 0.001). Furthermore, additional adjustments did not substantially alter these findings. Following adjustments for sex, age, diabetes, malignant tumor, chronic kidney disease, CVC, septic shock, WBC, hematocrit, procalcitonin, albumin and SOFA score in model 3, the adjusted OR was 1.05 (95% CI 1.02–1.08, p=0.003). As RDW tertiles elevated, the incidence of mortality increased, with the OR for T3 being higher than that for T1 (OR: 7.27, 95% CI: 3.66–14.41; p<0.001, model 1) in the crude model, and the association between RDW and mortality is statistically significant (OR: 7.63, 95% CI: 2.69–19.69; p<0.001) in Model 3. The p-values for the trend tests were all less than 0.001 across all three models.

Subgroup Analysis

In this study, we conducted stratified and interaction analyses to assess whether the relationship between RDW levels and mortality incidence was consistent across different subgroups. Stratified analysis indicated that the results remained consistent across the groups when categorized by sex, age, diabetes, septic shock, CRKP. Interaction analysis indicated that there was no interaction between RDW and the subgroups (p for interaction were both >0.05) (Figure 2).

Sensitivity Analysis

To confirm the robustness of our results, we performed sensitivity analyses excluding patients with hematological diseases, chronic heart failure, or chronic liver disease, with additional adjustment for SOFA score. As shown in Table 3, RDW remained significantly associated with in-hospital mortality.

Table 3 Sensitivity Analysis

Receiver Operating Characteristic (ROC) Curve Analysis

The discriminative ability of RDW, CRP, and PCT for predicting in-hospital mortality in patients with KP-BSI was assessed using receiver operating characteristic (ROC) curve analysis. As shown in Figure 3, RDW demonstrated the highest predictive value with an AUC of 0.705 (95% CI: 0.641–0.766), followed by PCT with an AUC of 0.698 (95% CI: 0.635–0.761), whereas CRP showed a relatively lower predictive ability with an AUC of 0.632 (95% CI: 0.565–0.699). These results suggest that RDW may serve as a moderate prognostic marker compared to CRP and PCT in this patient population.

Figure 3 Receiver Operating Characteristic (ROC) Curves Assessing RDW’s Predictive Performance for In-Hospital Mortality. Abbreviations: AUC, area under the curve; RDW, red cell distribution width.

Discussion

In this retrospective cohort study, we aimed to investigate the association between RDW and in-hospital mortality in patients with KP-BSI. A positive relationship was identified between RDW and mortality within the cohort. Our research confirms the clinical utility of RDW in enhancing mortality risk assessment for patients with KP-BSI.

The RDW quantifies variability in erythrocyte volumes within the bloodstream. Elevated RDW levels frequently originate from hematological disturbances that modify erythrocytic morphology or accelerate premature erythrocyte release into systemic circulation. Emerging clinical research has validated RDW as a prognostic parameter across different diseases, including COVID-19,^18,24 sepsis,^16,25,26 sepsis associated liver injury,¹⁷ severe osteomyelitis,²⁷ heart failure,¹⁹ and non-alcoholic fatty liver disease.²⁰ For example, Wu²⁶ et al analyzed data from 256,387 critically ill septic patients and revealed that those with an RDW >16% exhibited a significantly higher mortality risk, with a hazard ratio (HR) of 1.887 (95% CI, 1.847–1.928) in comparison to patients with an RDW < 16%. Liu et al²⁷ included 1109 patients with osteomyelitis, and founded that elevated RDW levels increased risk of mortality (HR = 2.128, 95% CI 1.324–3.420). Furthermore, Chun-Fu Yeh et al²⁸ have reported that RDW is an independent predictor of 28-day mortality in patients visiting the emergency department with BSI. These studies focused on various disease populations and indicated significant associations of higher RDW with adverse outcomes. These studies, covering diverse patient populations, consistently demonstrated that elevated RDW is associated with adverse outcomes. To our knowledge, this is the first study to specifically evaluate the association between RDW and mortality in patients with KP-BSI. Our findings are consistent with prior reports and provide strong evidence that elevated RDW independently predicts in-hospital mortality in this population.

The biological mechanisms between elevated RDW and adverse outcomes in KP-BSI patients remains unclear. The possible mechanisms between RDW and mortality risk in KP-BSI emerges from a tripartite interplay of erythrocyte dysregulation, systemic inflammation, and microbial virulence factors. First, LPS from KP can induce pyroptosis in erythroid precursors, leading to increased phosphatidylserine exposure and subsequent clearance of damaged red blood cells by macrophages.^29,30. This elevates RDW as spleen macrophages target damaged red cells.³¹ Second, Infection triggers the release of pro-inflammatory cytokines such as interleukin-1 (IL-1), IL-6, and tumor necrosis factor-alpha (TNF-α), which can disrupt erythropoiesis by inhibiting red blood cell maturation and promoting the release of reticulocytes into the peripheral blood. This disruption contributes to an increase in RDW.³² Third, The presence of LPS on the surface of KP can activate Toll-like receptor 9 (TLR9) in erythroblasts, impairing enucleation and leading to erythropoietic dysfunction.³³ Additionally, oxidative stress during infection generates reactive oxygen species (ROS) and reactive nitrogen species (RNS), which can damage red blood cell membranes, alter their deformability, and shorten their lifespan. These changes contribute to an increase in RDW and may exacerbate organ dysfunction and mortality.^34,35 Moreover, oxidative stress can also damage other tissue cells, impair organ function, worsen the condition, and increase the risk of death.³⁶ While RDW serves as a valuable prognostic marker, it should not be used in isolation. Our study’s findings are consistent with existing literature; however, further research is needed to elucidate the molecular mechanisms linking RDW to mortality in KP-BSI patients and to assess its utility in conjunction with other clinical parameters for comprehensive risk stratification.

Our study has several advantages: First, this is the first study to examine the connection between RDW and mortality among Chinese patients suffering from KP-BSI. Second, we utilized various statistical methods, including logistic regression and analysis of subgroups, to confirm the robustness of the findings. However, several limitations of the present study should be acknowledged. First, the research was conducted at a single center in China and utilized a retrospective design, which may limit its accuracy compared to more rigorous multicenter prospective studies from various countries. Second, there is a risk of selection bias due to the study’s reliance on a single measurement of RDW, with no follow-up assessments. This approach precludes analysis of how RDW levels at different time points may have influenced the outcomes. Additional research is needed to explore the mechanisms underlying the relationship between RDW and clinical outcomes in patients with KP-BSI. Conducting more multicenter studies on this readily accessible variable would be crucial for the early identification of outcomes, ultimately benefiting treatment strategies.

Conclusion

To summarize, RDW was associated with the prognosis of KP-BSI patients at the time of blood culture collection. Patients with KP-BSI and high RDW levels demonstrated a higher mortality rate. Therefore, RDW may act as an independent risk factor for predicting outcomes in individuals with KP-BSI.