Neutrophil-to-lymphocyte ratio: a potential supportive marker for elderly community-acquired bloodstream infections—a retrospective study

Research design

This retrospective analysis enrolled 232 patients aged ≥ 60 years admitted to Xuanwu Hospital of Capital Medical University between January 2012 and December 2021. Participants were stratified into BSI and non-infection groups. The BSI group comprised 58 patients diagnosed with CA-BSI infection, while 174 inpatients with non-infectious diseases were matched during the same period based on age, sex, and concomitant diseases. Ethical approval was granted by the Institutional Ethics Committee of Xuanwu Hospital (Lin Yan Shen [2022] No. 131), with waived written informed consent due to the retrospective analysis of anonymized data.

Participant selection criteria

Inclusion required: (a) age ≥ 60 years and (b) confirmed CA-BSI diagnosis. Exclusion criteria encompassed: recent major surgery (within 3 months), chronic liver disease, acquired immunodeficiency syndrome, sustained glucocorticoid use (≥ 10 mg prednisone daily or equivalent for ≥ 1 month), recent radiotherapy/chemotherapy (≤ 3 months), granulocytosis, rheumatic immune diseases (e.g., systemic lupus erythematosus, rheumatoid arthritis), hematological disorders (e.g., aplastic anemia, leukemia), or hypersplenism. The diagnosis of BSI was based on the criteria established by the Ministry of Health of the People’s Republic of China, with CA-BSI defined by absence of recent healthcare facility exposure and symptom onset before or within 48 h of admission (Ministry of Health of the People’s Republic of China, 2001).

Blood culture protocol

For BSI-group patients prior to antibiotic administration, two sets of blood cultures were drawn from distinct anatomical sites. Each set contained one anaerobic and one aerobic culture bottle (BD BACTEC™ system).

Clinical data collection

General patient information included name, age, sex, and comorbid conditions (hypertension, type 2 diabetes, coronary heart disease).

For the BSI group, data encompassed:

1. Infection manifestations: chills, fever (axillary temperature > 37 °C), sweats, fatigue, myalgia, pharyngitis, respiratory symptoms (cough, dyspnea), gastrointestinal disturbances (nausea, diarrhea), urinary complaints, neurological alterations (drowsiness, coma), rash, hepatosplenomegaly, and vital signs.

2. Suspected infection sites.

3. Pre-antibiotic laboratory tests: complete blood count: White blood cell (WBC), neutrophil count (NEUT), lymphocyte count (LYM), NLR; analyzed on Sysmex XS-800i with original reagents) and blood culture results (BD 9120 system).

The non-infection group was evaluated for absence of the aforementioned clinical manifestations and underwent identical blood tests. All laboratory analyses were performed by the Central Laboratory of Xuanwu Hospital, with microbial identification conducted using the VITEK 32 system (Meniere, France).

For the BSI group, blood samples were collected before the administration of any antibiotics. For the non-infection group, samples were collected upon hospital admission.

Statistical analysis

Quantitative data were summarized using means with standard deviations for normally distributed variables, or medians with interquartile ranges for non-normally distributed variables. Categorical variables were expressed as frequencies and percentages. Normality of data was assessed using the Shapiro–Wilk test.

Intergroup comparisons employed:

1. Independent samples t-tests for normally distributed continuous data;

2. Mann–Whitney U tests for non-normally distributed continuous data;

3. Pearson’s χ² tests for categorical variables.

Binary logistic regression identified factors associated with CA-BSI in elderly patients. Multivariable logistic regression models were adjusted for age, sex, and comorbidities (hypertension, type 2 diabetes, coronary heart disease).

Diagnostic performance of leukocyte indices was evaluated through receiver operating characteristic (ROC) curve analysis, determining optimal thresholds with corresponding sensitivity and specificity. Statistical significance was defined as P < 0.05 (two-tailed).

Clinical and microbiological profiles

We enrolled 232 hospitalized elderly patients (≥ 60 years): 58 with CA-BSI and 174 controls admitted for non-infectious conditions. Hypertension, diabetes, and coronary heart disease prevalence did not differ significantly between groups (all P > 0.05; Table 1).

BSI-group patients exhibited diverse clinical manifestations (detailed in Tables 2–3), which were absent in control subjects. Among 58 BSI cases, monomicrobial infections were identified in 30 Gram-positive and 24 Gram-negative cases, while polymicrobial infections occurred in four cases (specifically: two with Gram-positive/fungal co-infections and two with mixed bacterial infections) (detailed in Table 4).

Microbiological analysis revealed 65 total isolates: Gram-positive pathogens accounted for 55.4% (36/65), predominantly Staphylococcus species (25 strains including eight S.hominis, five S. haemolyticus, four S. aureus, and four S. epidermidis) and Enterococcus species (eight strains comprising six E. faecium and two E. faecalis). Gram-negative isolates represented 40.0% (26/65), with major pathogens being Escherichia coli (20.0%, 13/65) and Klebsiella species (12.3%, 8/65; including five K. pneumoniae). Fungal pathogens constituted 4.6% (3/65) of isolates, identified as Candida parapsilosis, C. glabrata, and C. albicans.

Pathogen distribution analysis showed that 49 primary BSI cases yielded 55 pathogens, whereas nine secondary BSI cases originated from urinary (n = 8) or pulmonary (n = 1) sources, with full pathogen characteristics provided in Table 4.

Diagnostic utility of leukocyte indices for CA-BSI

Peripheral blood analysis revealed significantly elevated WBC counts (P < 0.01), neutrophil counts (P < 0.01), and NLR (P < 0.01), but reduced lymphocyte counts (P < 0.01) in BSI patients versus non-infected controls (Table 5).

Correlation analysis identified demographic(age), comorbidities (hypertension, type 2 diabetes and coronary heart disease), laboratory tests (WBC, NEUT, LYM, and NLR), manifestations (fever, chills and rigors and productive cough), as significant risk factors for CA-BSI (all P < 0.05). (Table 6)

Univariate logistic regression identified laboratory tests (WBC, NEUT, LYM, and NLR), comorbidities (hypertension and coronary heart disease) as significant risk factors for CA-BSI (all P < 0.001). Multivariate analysis (Hosmer–Lemeshow P = 0.733) confirmed only NLR elevation independently predicted CA-BSI (OR = 3.416; 95% CI [1.861–6.268]; P < 0.001) (Tables 7–8).

ROC curve analysis demonstrated NLR’s superior diagnostic performance for CA-BSI (AUC = 0.95, 95% CI [0.89–1.00], P < 0.05). At the optimal cutoff of 8.0, sensitivity was 0.72 and specificity 0.99, with positive/negative predictive values of 0.95 and 0.09, respectively (Youden’s index maximized; Table 9, Fig. 1).

Clinical significance of early CA-BSI diagnosis in elderly

The case fatality rate of BSI in China reaches 26.8%, underscoring the critical need for early intervention (Yang et al., 2010). Current guidelines mandate intravenous antimicrobial therapy within 1 h of sepsis diagnosis (Evans et al., 2021). Although blood culture remains the diagnostic gold standard, its 15%–25% positivity rate and 24–72 h turnaround limit utility in community settings (Book et al., 2013). Molecular methods offer faster results but are cost-prohibitive for widespread community use (Mwaigwisya, Assiri & O’Grady, 2015). Thus, identifying rapid, economical diagnostic markers for elderly CA-BSI is imperative.

Atypical presentations in elderly CA-BSI

Elderly patients frequently exhibit non-classical symptoms. In this study, 53.4% lacked fever—consistent with evidence suggesting altered thermoregulation in aging. The rectal temperature cutoff for fever in elderly is >37.8 °C, yet only 8% exceed 38.5 °C (Norman, 2000). While rectal thermometry detects 86% of febrile cases versus 32% for axillary measurements (Darowski et al., 1991). However, Singler et al. (2013) noted that although rectal thermometry is more accurate, its implementation in elderly patients is often hindered by procedural difficulties and practical barriers. Zhang, W & Lu (2022) reported 135 patients with S. aureus BSI cases and found that the proportion of febrile patients was 31.1%, while the rate of cough with sputum production was even lower at 22.2%. These findings confirm that reliance on typical symptoms risks missed diagnoses.

Diagnostic utility of leukocyte indices

Superiority of NLR

In this study, multivariate analysis of BSI revealed that among white blood cell indicators, only an elevated NLR was independently associated with community-acquired BSI in the elderly. Changes in NLR reflect the dynamic equilibrium between inflammatory response and immune status. As the numerator, NEUT—innate immune cells—are mobilized from the bone marrow into peripheral blood during BSI to phagocytose pathogens (Jackaman et al., 2017). Conversely, the denominating LYM—adaptive immune cells—may undergo apoptosis or immune suppression in BSI, resulting in peripheral lymphocytopenia (Hawkins et al., 2006). Scholars have reported that during bacterial invasion, limited receptor expression on NEUT surfaces leads to the release of immature or dysfunctional neutrophils from the bone marrow, while substantial LYM migration to infection sites causes peripheral LYM depletion (Summers et al., 2010). In a retrospective analysis of CA-BSI in adults, De Jager et al. (2010) assessed the diagnostic utility of white blood cell parameters. Elevated NLR and reduced LYM were the only significant indicators; NLR demonstrated superior diagnostic sensitivity (0.77) and specificity (0.63), with an optimal cutoff of 10.

Similarly, Tan et al. (2019) retrospectively analyzed 70 adult patients with urosepsis: Neutrophils significantly increased during sepsis but showed attenuated elevation upon progression to septic shock. This suggests that while NEUT counts have diagnostic value for sepsis, their limited change in severe BSI (e.g., septic shock) reduces utility. Concurrently, rapid LYM apoptosis due to systemic immune suppression maintains NLR dynamics. Thus, elevated NLR may hold greater diagnostic significance than isolated NEUT or LYM changes in severe BSI.

Multivariate analysis confirmed NLR elevation as the sole hematologic predictor of community-acquired BSI in the elderly. Single parameters—WBC and NEUT counts—exhibit marked fluctuations influenced by non-infectious factors, while LYM primarily reflects immune status, offering incomplete diagnostic insight. NLR integrates NEUT-driven inflammation and LYM-mediated immune response, providing a stable diagnostic marker for elderly CA-BSI. Elevated NLR outperforms elevated WBC/NEUT or reduced LYM, serving as a more reliable early indicator. Setting the NLR cutoff at > 8.0 yields high sensitivity and specificity, alerting clinicians to probable CA-BSI in elderly patients.

NLR’s broader diagnostic context

Zahorec (2021) have established reference ranges for the NLR based on extensive clinical trial data: The normal threshold for NLR ranges from 1 to 2, while values for mild, moderate, and severe infections or stress responses fall within 3–7, 7–11, and 11–17, respectively. NLR demonstrates diagnostic utility with disease-specific cutoffs across various infectious conditions.

In a study by Farah et al. (2017) involving 126 patients, the mean NLR was significantly higher in hospital-acquired pneumonia (14.37) than in community-acquired pneumonia (9.39) (P < 0.01). NLR has been established as a powerful prognostic predictor in community-acquired pneumonia patients (de Jager et al., 2012), including older adults (Cataudella et al., 2017). Khanzadeh et al. (2022) analyzed 25 studies encompassing 6,410 stroke patients, revealing that post-stroke pneumonia cases exhibited substantially elevated NLR levels (mean range: 4.6–8.2) compared to non-pneumonia cases. Asik (2021) demonstrated NLR’s diagnostic value for urinary tract infections in 406 patients, identifying an optimal cutoff of 4.6. Similarly, Festa et al. (2022) reported pooled sensitivity, specificity, and AUC of 0.72, 0.74, and 0.73, respectively, for NLR in diagnosing prosthetic joint infections among 7,537 patients undergoing total hip or knee arthroplasty. Among systemic inflammatory and infectious processes, NLR, combined with CRP, served as a useful prognostic predictor in COVID-19 patients (Regolo et al., 2022). Furthermore, NLR demonstrated an inverse relationship with PaO2/FiO2, a recognized marker of respiratory failure (Regolo et al., 2023). Li et al. (2025) explored the increased diagnostic power of combining NLR with several biomarkers to diagnose BSI in children. Furthermore, NLR is an emerging marker of the relationships between the immune system and diseases (Buonacera et al., 2022).

While NLR shows partial diagnostic value for pneumonia, urinary tract infections, and postoperative joint infections, it carries limitations. Farkas (2020) caution that exogenous steroids may elevate NLR, whereas adrenal insufficiency may reduce it. Furthermore, NLR’s applicability remains unvalidated in patients with human immunodeficiency virus or active hematologic disorders (e.g., leukemia, chemotherapy). Thus, clinical interpretation of NLR for infectious diseases requires comprehensive consideration of comorbidities, medications, and auxiliary test results.

Innovations and limitations

This study establishes NLR > 8.0 as a novel, high-specificity indicator for elderly CA-BSI—surpassing traditional markers. While retrospective data inherently restrict the precision of symptom-to-sampling time assessments, our protocol adheres to best practices in early clinical diagnosis by enforcing pre-antibiotic sampling procedures. Limitations include its retrospective, single-center design and modest sample size. Future multicenter prospective studies should validate NLR’s diagnostic utility.