Abstract
Background
Patients with cirrhosis are highly susceptible to pneumonia, which significantly increases mortality. Existing prognostic scores have limitations in capturing the dual burden of systemic inflammation and impaired hepatic synthesis. The neutrophil-to-albumin ratio (NAR), a composite biomarker reflecting both inflammatory activity and nutritional status, may offer more comprehensive risk stratification in this population.
Methods
This multicenter retrospective cohort study enrolled 504 hospitalized patients with concurrent cirrhosis and pneumonia from five centers between November 2013 and December 2024. The primary outcome was in-hospital mortality. Patients were stratified by admission NAR tertiles. Multivariable logistic regression models were used to assess the independent association between NAR (analyzed as both continuous and categorical variable) and mortality, with progressive adjustment for demographics, comorbidities, and key laboratory parameters. Sensitivity analyses, including E-value calculation and subgroup analyses, were performed. Mediation analysis evaluated the role of white blood cell count (WBC).
Results
The cohort’s mean age was 64.7 ± 12.4 years, with 64.9% males. Higher NAR was associated with worse liver function (Albumin-Bilirubin grade, Model for End-Stage Liver Disease score), greater pneumonia severity (CURB-65), and elevated inflammatory markers. In-hospital mortality was significantly higher in the highest NAR tertile (14.9% vs. 5.4% in the lowest, P < 0.002). After full adjustment, each unit increase in NAR was associated with a 2.23-fold higher mortality odds (OR = 2.23, 95% CI: 1.25–4.00, P = 0.007). Patients in the highest tertile had 2.48 times the odds of death compared to the lowest (OR = 2.48, 95% CI: 1.08–5.70, P = 0.032), with a significant dose-response trend (P = 0.017). Sensitivity analyses confirmed robustness. Mediation analysis indicated that approximately 50% of NAR’s effect on mortality was mediated through WBC.
Conclusions
Admission NAR is strongly and independently associated with in-hospital mortality in cirrhotic patients with pneumonia, integrating systemic inflammation and nutritional status. This readily calculable biomarker may enhance risk assessment in this high-risk population and warrants prospective validation.
Trial registration
Chinese Clinical Trial Registry, ChiCTR2500097772. Registered 25 February 2025.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12879-026-12585-3.
Keywords: Cirrhosis, Pneumonia, Neutrophil-to-albumin ratio, Mortality, Biomarker, Prognosis
Introduction
Cirrhosis represents the end-stage manifestation of various chronic liver diseases, imposing a substantial global health burden, with its complications being a primary cause of patient mortality [1]. Infection, particularly pneumonia, stands as one of the most common complications and a significant precipitant of death in cirrhotic patients [2, 3]. Patients with cirrhosis are predisposed to infections due to immune dysfunction, and these infections often progress to sepsis and acute-on-chronic liver failure (ACLF), leading to sharply increased mortality [4, 5].Therefore, early identification of high-risk patients is crucial for guiding intervention and improving prognosis.
Currently, clinicians often combine liver disease-specific scores and infection severity scores to assess prognosis in such patients. We commonly evaluate liver functional reserve using the Model for End-Stage Liver Disease (MELD) score and the Albumin-Bilirubin (ALBI) grade [6, 7], while we frequently gauge pneumonia severity with the CURB-65 score [8]. However, these scoring systems have respective limitations. The MELD score does not include infection-related parameters, and the CURB-65 score fails to fully account for the unique pathophysiological alterations in the context of liver disease, such as the systemic inflammatory response and nutritional/immune status [9, 10]. The mortality risk in cirrhotic patients with pneumonia results from the combined effects of liver failure and severe infection. Consequently, we need convenient biomarkers that integrate both systemic inflammation and liver function/nutritional status to provide more comprehensive risk stratification.
In this context, composite biomarkers that combine inflammatory and nutritional parameters, such as the neutrophil-to-albumin ratio (NAR), may offer a more holistic assessment of patient risk by simultaneously reflecting the pro-inflammatory state and immunonutritional reserve. The neutrophil count provides a rapidly responsive inflammatory marker, while serum albumin reflects hepatic synthetic function and serves as a key negative acute-phase protein crucial for immune homeostasis [11, 12]. The neutrophil-to-albumin ratio (NAR) theoretically captures both enhanced inflammation and weakened defense. Preliminary studies have reported prognostic associations of NAR in conditions like heart failure and stroke-associated pneumonia [13–15]. However, we still lack clarity on its association with outcomes in cirrhotic patients with pneumonia, as well as its independence from traditional scores and its underlying mechanisms.
Therefore, we conducted this multicenter retrospective cohort study to evaluate the association between the admission NAR and in-hospital mortality in cirrhotic patients with pneumonia, and to explore its potential prognostic utility in this high-risk group.
Materials and methods
Data source
This retrospective cohort study utilized data extracted from the electronic medical record systems of five hospitals (Beijing Mentougou District Hospital, n = 110; Chinese People’s Armed Police Force Characteristic Medical Center, n = 126; Qingdao Central Hospital, n = 114; Xiangtan Central Hospital, n = 95; and Chuiyangliu Hospital Affiliated to Tsinghua University, n = 59). Data were collected sequentially over the study period. The study enrolled a total of 504 hospitalized patients diagnosed with both cirrhosis (ICD-10: K74.6) and pneumonia (ICD-10: J12-J18) between November 2013 and December 2024. Data extraction and database construction were performed using the “FREE Electronic Data Capture System” (V2.0, Beijing FreeClinical Medical Technology Co., Ltd.).
Albumin and neutrophil counts were measured using standardized automated analyzers at each participating hospital’s clinical laboratory. While inter-laboratory variability exists as a common limitation in multicenter retrospective studies, all laboratories operated under accredited national quality control programs.
Study population and data collection
The study population comprised adults aged ≥ 18 years with a discharge or admission diagnosis of both cirrhosis and pneumonia, supported by adequate imaging, laboratory, or clinical documentation. Exclusion criteria were age < 18 years, insufficient or unconfirmed diagnostic evidence, missing data on the primary outcome (in-hospital death), and retention of only the first admission for patients with multiple hospitalizations. From the electronic medical records, we collected demographic information (age and sex); comorbidities (history of malignant tumor, coronary atherosclerotic heart disease [CAD], hypertension, diabetes mellitus [DM], heart failure [HF], and hepatitis type [viral, alcoholic, other, or none]); the first available laboratory measurements after admission, including serum creatinine (Cr), urea, total bilirubin (TBIL), direct bilirubin (DBIL), alanine aminotransferase (ALT), aspartate aminotransferase (AST), fibrinogen, prothrombin time/international normalized ratio (PT/INR), activated partial thromboplastin time (APTT), platelet count (PLT), hemoglobin (Hb), white blood cell count (WBC), D-dimer, neutrophil count, and albumin; as well as calculated disease severity scores (Albumin-Bilirubin [ALBI] grade, CURB-65 score, and Model for End-Stage Liver Disease [MELD] score).
Variable definitions
The diagnosis of cirrhosis was based on meeting one of the following four criteria: imaging evidence suggestive of nodular liver surface, hepatic lobe volume redistribution, and splenomegaly; endoscopic confirmation of esophageal or gastric varices; laboratory evidence of hepatic decompensation (e.g., hypoalbuminemia, elevated transaminases, hyperbilirubinemia, or prolonged coagulation time); or histopathological confirmation of pseudolobule formation surrounded by fibrous septa [16]. The diagnosis of pneumonia required meeting both radiological and clinical-laboratory criteria: radiological findings showing new infiltrates on chest X-ray or CT; and clinically presenting with at least one of the following: new or worsening cough, sputum production, fever (oral temperature ≥ 37.3 °C), signs of lung consolidation/moist rales, or abnormal peripheral white blood cell count [17]. The neutrophil-to-albumin ratio (NAR) was defined as the peripheral blood neutrophil count (×10⁹/L) divided by the serum albumin concentration (g/L). The primary outcome measure of this study was in-hospital death. The ALBI grade was calculated using the formula: ALBI score = (log₁₀ bilirubin [µmol/L] × 0.66) + (albumin [g/L] × -0.085). Patients were classified into Grade 1 (score ≤ -2.60), Grade 2 (-2.60 < score ≤ -1.39), or Grade 3 (score > -1.39) [18]. The MELD score was calculated using the standard formula: MELD = 3.78 × ln(bilirubin [mg/dL]) + 11.20 × ln(INR) + 9.57 × ln(creatinine [mg/dL]) [19]. The CURB-65 score was assessed based on the following indicators: Confusion, Urea > 7 mmol/L, Respiratory rate ≥ 30 breaths/min, low Blood pressure (systolic < 90 mmHg or diastolic ≤ 60 mmHg), and Age ≥ 65 years. One point was assigned for each criterion met, resulting in a total score ranging from 0 to 5 [20].
Statistical analysis
The 504 patients were stratified into three groups based on the tertiles of the neutrophil-to-albumin ratio (NAR).This approach was chosen to ensure sufficient statistical power in each group for comparison and to avoid arbitrary cut-off selection. In the baseline characteristics, categorical variables were presented as frequency (percentage), normally distributed continuous variables as mean ± standard deviation, and non-normally distributed variables as median (interquartile range). Group comparisons were performed using one-way ANOVA, the Kruskal-Wallis test, or the Chi-square test, depending on the data type. Multivariable logistic regression was employed to evaluate the independent association between NAR and the risk of in-hospital death.
Four progressively adjusted models were constructed: an unadjusted model (crude model), a model adjusted for age and sex (Model 1), a model further adjusted for comorbidities (coronary atherosclerotic heart disease, hypertension, diabetes mellitus) and hepatitis type (Model 2), and a fully adjusted model additionally incorporating key laboratory parameters (alanine aminotransferase, D-dimer, hemoglobin, activated partial thromboplastin time) (Model 3). Covariates were selected based on clinical relevance and preliminary statistical analyses. Variables exhibiting multicollinearity were assessed and excluded. The laboratory parameters (ALT, D-dimer, Hb, APTT) in the fully adjusted model were selected based on their known clinical relevance to liver function, systemic inflammation/coagulation, and prognosis in cirrhosis, as well as statistical significance in preliminary analyses. To test for a dose-response relationship between NAR and mortality risk, the median value of each NAR tertile group was treated as a continuous variable for trend testing.
To verify the robustness of the primary findings, a series of sensitivity analyses were conducted, including: (1) analysis based on a complete-case dataset (excluding any missing values); (2) calculation of the E-value to assess the strength of association an unmeasured confounder would need to have to explain away the observed association [21]; (3) subgroup analyses stratified by age, sex, comorbidities, hepatitis type, and disease severity scores (MELD, ALBI, CURB-65), with tests for interaction. Furthermore, to explore the potential mechanism through which NAR influences mortality risk, we employed a mediation analysis model to examine the possible mediating role of white blood cell count (WBC). We acknowledge that mediation analysis in observational studies assumes specific conditions (e.g., temporality, no unmeasured confounding between the mediator and outcome) that cannot be fully verified with retrospective data. Therefore, these results should be interpreted as exploratory, suggesting a shared pathway rather than establishing causal mediation.
For variables with a missing rate below 20% (see Supplementary Table 1 for details), multiple imputation was performed using the MICE package in R software (version 4.2.2) [22]. This threshold was chosen as it represents a commonly accepted balance between bias and precision in epidemiological studies [23–25]. All statistical analyses were performed using R 4.2.2 and Free Statistics 2.2.0 software. A two-sided P-value < 0.05 was considered statistically significant. The design and reporting of this study strictly adhered to the STROBE statement [26].
Results
Baseline characteristics
A total of 504 hospitalized patients concurrently diagnosed with cirrhosis and pneumonia were included in this study. The flowchart of patient selection is presented in Fig. 1. The baseline characteristics of the entire cohort are summarized in Table 1. The cohort had a mean age of 64.7 ± 12.4 years and was predominantly male (64.9%). After stratification by tertiles of the neutrophil-to-albumin ratio (NAR), no significant differences were observed among the three groups regarding sex and age (P > 0.05).
Fig. 1.
Flowchart of patient selection for the study
Table 1.
Baseline characteristics of cirrhotic patients with pneumonia stratified by NAR tertiles
| Variables | Total (n = 504) | NAR | P-value | ||
|---|---|---|---|---|---|
| T1 (n = 168) | T2 (n = 168) | T3 (n = 168) | |||
| Male c | 327 (64.9) | 98 (58.3) | 112 (66.7) | 117 (69.6) | 0.079 |
| Agea (years) | 64.7 ± 12.4 | 63.0 ± 12.0 | 66.2 ± 11.7 | 64.9 ± 13.1 | 0.059 |
| Crb (umol/L) | 70.0 (54.0, 99.0) | 64.0 (52.8, 85.2) | 70.3 (51.8, 92.0) | 78.0 (57.0, 132.5) | < 0.001 |
| Ureab (umol/L) | 6.5 (4.6, 10.3) | 5.1 (4.0, 7.0) | 6.8 (5.0, 9.5) | 8.6 (5.5, 14.5) | < 0.001 |
| TBILb(µmol/L) | 21.4 (11.9, 40.0) | 21.1 (15.4, 36.2) | 22.2 (12.1, 39.5) | 20.9 (8.5, 46.9) | 0.409 |
| DBILb(µmol/L) | 13.0 (6.8, 31.1) | 11.0 (6.9, 19.7) | 13.1 (6.2, 29.0) | 17.9 (7.7, 62.4) | 0.002 |
| ALTb(U/L) | 24.1 (15.8, 39.1) | 23.0 (14.0, 36.2) | 23.0 (15.0, 36.0) | 27.5 (18.0, 49.2) | 0.008 |
| ASTb(U/L) | 41.0 (24.9, 65.3) | 37.5 (24.1, 59.9) | 37.8 (24.0, 65.1) | 51.2 (27.0, 73.1) | 0.005 |
| Fibrinogenb(g/L) | 259.0(166.0, 368.0) | 252.0 (170.8, 341.5) | 238.0 (164.8, 318.0) | 293.5 (166.0, 436.8) | 0.011 |
| PT/INR a | 1.4 ± 0.5 | 1.3 ± 0.3 | 1.3 ± 0.4 | 1.5 ± 0.6 | 0.015 |
| APTTa (s) | 39.9 ± 10.1 | 39.6 ± 7.3 | 39.1 ± 10.1 | 40.9 ± 12.3 | 0.231 |
| PTa (s) | 16.3 ± 4.9 | 15.8 ± 3.3 | 16.2 ± 4.9 | 17.0 ± 6.1 | 0.095 |
| PLTb(×109/L) | 96.0 (65.8, 144.2) | 81.0 (55.2, 116.5) | 86.5 (62.5, 139.0) | 120.0 (80.8, 192.2) | < 0.001 |
| Hba (g/L) | 102.1 ± 30.8 | 102.1 ± 30.5 | 101.9 ± 30.1 | 102.5 ± 31.9 | 0.987 |
| WBCb(×109/L) | 5.3 (3.6, 8.3) | 3.2 (2.5, 4.2) | 5.2 (4.2, 6.5) | 10.3 (7.9, 13.0) | < 0.001 |
| D-dimerb (g/L) | 1.9 (0.8, 4.0) | 1.7 (0.6, 3.0) | 2.0 (0.8, 4.4) | 2.1 (1.0, 5.1) | 0.012 |
| Hepatitis types c | 0.161 | ||||
| None | 64 (12.7) | 18 (10.7) | 18 (10.7) | 28 (16.7) | |
| Viral | 225 (44.6) | 85 (50.6) | 73 (43.5) | 67 (39.9) | |
| Alcoholic | 111 (22.0) | 28 (16.7) | 44 (26.2) | 39 (23.2) | |
| Others | 104 (20.6) | 37 (22) | 33 (19.6) | 34 (20.2) | |
| ALBI grade c | < 0.001 | ||||
| Grade 1 | 55 (10.9) | 28 (16.7) | 17 (10.1) | 10 (6) | |
| Grade 2 | 271 (53.8) | 111 (66.1) | 97 (57.7) | 63 (37.5) | |
| Grade 3 | 178 (35.3) | 29 (17.3) | 54 (32.1) | 95 (56.5) | |
| CURB-65 score c | < 0.001 | ||||
| <3 | 233 (46.2) | 99 (58.9) | 75 (44.6) | 59 (35.1) | |
| ≥ 3 | 271 (53.8) | 69 (41.1) | 93 (55.4) | 109 (64.9) | |
| MELD score c | < 0.001 | ||||
| <10 | 387 (76.8) | 140 (83.3) | 136 (81) | 111 (66.1) | |
| ≥ 10 | 117 (23.2) | 28 (16.7) | 32 (19) | 57 (33.9) | |
| CAD c | 104 (20.6) | 25 (14.9) | 50 (29.8) | 29 (17.3) | 0.001 |
| DM c | 147 (29.2) | 43 (25.6) | 56 (33.3) | 48 (28.6) | 0.29 |
| HF c | 21 (4.2) | 6 (3.6) | 11 (6.5) | 4 (2.4) | 0.144 |
| Hypertension c | 179 (35.5) | 44 (26.2) | 68 (40.5) | 67 (39.9) | 0.008 |
| Malignant tumor history c | 140 (27.8) | 48 (28.6) | 40 (23.8) | 52 (31) | 0.33 |
| Non-survival c | 44 (8.7) | 9 (5.4) | 10 (6) | 25 (14.9) | < 0.002 |
Notes: data presented are mean ± SDa, median (Q1-Q3)b, or N (%)c
Abbreviations: SD: standard deviation; NAR: neutrophil-to-albumin ratio; Cr: serum creatinine; TBIL: total bilirubin; DBIL: direct bilirubin; ALT: alanine aminotransferase; AST: aspartate aminotransferase; PT/INR: prothrombin time/international normalized ratio; APTT: activated partial thromboplastin time; PT: prothrombin time; PLT: platelet count; Hb: hemoglobin; WBC: white blood cell count; CAD: Coronary Atherosclerotic Heart Disease; DM: Diabetes Mellitus; HF: Heart Failure; MELD score: Model for End-Stage Liver Disease score; ALBI: Albumin-Bilirubin grade
Clinically important differences were observed across NAR tertiles. Compared to patients in the lowest tertile (T1), those in the highest tertile (T3) showed signs of greater disease severity: significantly elevated levels of Cr, urea, DBIL, ALT, AST, fibrinogen, PT/INR, D-dimer, and higher PLT and WBC counts (all P < 0.05).
Regarding disease severity scores, a clear gradient was evident. Patients in the T3 group had significantly worse liver function, characterized by a higher proportion with ALBI grade 3 (56.5% vs. 17.3% in T1, P < 0.001) and a MELD score ≥ 10 (33.9% vs. 16.7%, P < 0.001). Concurrently, pneumonia severity was greater in the T3 group, with a higher proportion having a CURB-65 score ≥ 3 (64.9% vs. 41.1%, P < 0.001). Among comorbidities, the prevalence of hypertension and CAD was higher in the T3 group (P < 0.01). Critically, in-hospital mortality differed significantly across tertiles, with the highest rate observed in the T3 group (14.9%), compared to 5.4% in T1 (P < 0.002), suggesting a preliminary association between elevated NAR and adverse outcomes.
Association between the NAR and in-hospital mortality in cirrhotic patients with pneumonia
Table 2 presents the results of the multivariable logistic regression analyses assessing the association between NAR and in-hospital mortality. When analyzed as a continuous variable, NAR remained significantly associated with an increased risk of mortality after sequential adjustment for potential confounders (Model 3: OR = 2.23, 95% CI: 1.25–4.00, P = 0.007). To contextualize the effect size, given the median neutrophil count (3.6 × 10⁹/L) and albumin level (30.9 g/L) in our cohort, a one-unit increase in NAR roughly corresponds to an increase in neutrophils by ~ 3.6 × 10⁹/L, a decrease in albumin by ~ 3.6 g/L, or a combination thereof.
Table 2.
Multivariate logistic regression analysis of the NAR and in-hospital mortality
| Variable | Crude | Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|---|---|
| OR(95%CI) | P-value | OR (95%CI) | P-value | OR(95%CI) | P-value | OR(95%CI) | P-value | |
| NAR | 2.37 (1.38 ~ 4.06) | 0.002 | 2.29 (1.33 ~ 3.95) | 0.003 | 2.41 (1.38 ~ 4.21) | 0.002 | 2.23 (1.25 ~ 4) | 0.007 |
| NAR | ||||||||
| T1 | 1(Ref) | 1(Ref) | 1(Ref) | 1(Ref) | ||||
| T2 | 1.12 (0.44 ~ 2.83) | 0.813 | 1.06 (0.42 ~ 2.7) | 0.903 | 0.95 (0.37 ~ 2.45) | 0.912 | 0.9 (0.34 ~ 2.34) | 0.823 |
| T3 | 3.09 (1.4 ~ 6.84) | 0.005 | 2.95 (1.32 ~ 6.56) | 0.008 | 2.91 (1.29 ~ 6.56) | 0.01 | 2.48 (1.08 ~ 5.7) | 0.032 |
| P for trend | 0.003 | 0.004 | 0.004 | 0.017 | ||||
Model 1: adjusted for age, sex
Model2: adjusted for age, sex, CAD, hypertension, DM, hepatitis types
Model3: adjusted for age, sex, CAD, hypertension, DM, hepatitis types, ALT, D-dimer, Hb, APTT
Abbreviations: OR: odds ratio; CI: confidence interval; NAR: neutrophil-to-albumin ratio; ALT: alanine aminotransferase; APTT: activated partial thromboplastin time; Hb: hemoglobin; CAD: Coronary Atherosclerotic Heart Disease; DM: Diabetes Mellitus
Analysis by NAR tertiles revealed that, compared to the lowest tertile (T1), patients in the highest tertile (T3) had a significantly elevated risk of death in the fully adjusted model (OR = 2.48, 95% CI: 1.08–5.70, P = 0.032). A significant dose-response trend was also confirmed (P for trend = 0.017). These results indicate that a higher admission NAR is independently associated with increased odds of in-hospital death in this population.
Sensitivity analysis
To test the robustness of the primary findings, sensitivity analyses were performed. As shown in Supplementary Table 2, analysis based on the complete-case dataset (n = 141) reinforced the main conclusions, with the association between continuous NAR and mortality remaining significant (Model 3: OR = 7.98, 95% CI: 2.29–27.86, P = 0.001).
To further assess the potential impact of unmeasured confounding, we calculated the E-value. As illustrated in Fig. 2, the E-value for this cohort was 3.89 (lower limit of 95% CI: 1.81). This suggests that an unmeasured confounder would need to be associated with both NAR and mortality by risk ratios of at least 3.89-fold each to fully explain away the observed association, indicating relative robustness to potential confounding.
Fig. 2.
E-value analysis assessing the potential influence of unmeasured confounding on the association between NAR and in-hospital mortality. Abbreviations: RR: risk ratio
The results of subgroup analyses are presented in the forest plot (Fig. 3). The positive association between elevated NAR and increased in-hospital mortality remained consistent across all subgroups stratified by age, sex, comorbidities (hypertension, diabetes, CAD), hepatitis type, and different disease severity scores (MELD, ALBI, CURB-65). Notably, all tests for interaction yielded P-values > 0.05, indicating no statistically significant heterogeneity in the association across patient subgroups with different clinical characteristics.
Fig. 3.
Forest plot of subgroup analyses for the association between NAR and in-hospital mortality. Abbreviations: OR: odds ratio; CI: confidence interval; NAR: neutrophil-to-albumin ratio; CAD: Coronary Atherosclerotic Heart Disease; MELD score: Model for End-Stage Liver Disease score; ALBI: Albumin-Bilirubin grade
Mediation analysis
To explore the potential pathway underlying the association between NAR and mortality risk, a mediation analysis was performed to examine the role of WBC. As shown in Fig. 4, the total effect of NAR on in-hospital death was significant (estimate = 0.0905, 95% CI: 0.0307–0.1615, P < 0.001). WBC played a significant mediating role, with an indirect effect estimate of 0.0446 (95% CI: 0.007–0.086, P = 0.024), accounting for approximately 49.99% of the total effect. Furthermore, WBC itself was independently associated with mortality risk (OR = 1.09, 95% CI: 1.04–1.14, P < 0.001), and NAR was strongly positively correlated with WBC levels (β = 7.51, 95% CI: 6.51–8.52, P < 0.001). This suggests that nearly half of the association between NAR and mortality risk operates through a pathway shared with elevated systemic inflammation, as reflected by WBC.
Fig. 4.
Mediation analysis model examining the role of WBC in the association between NAR and in-hospital mortality. Abbreviations: OR: odds ratio; CI: confidence interval; WBC: white blood count; NAR: neutrophil-to-albumin ratio
Discussion
This multicenter retrospective cohort analysis found that an elevated admission neutrophil-to-albumin ratio (NAR) is strongly and independently associated with in-hospital mortality in cirrhotic patients with pneumonia. This observed association remained robust after extensive adjustment for demographics, comorbidities, hepatitis etiology, and key laboratory indicators, and a significant dose-response relationship was noted. Sensitivity analyses supported the reliability of this finding. Notably, our exploratory mediation analysis suggests that approximately 50% of the association between NAR and mortality risk aligns with elevated WBC levels, highlighting a shared pathway involving systemic inflammation.
The value of composite ratios integrating inflammatory and nutritional markers for prognostic assessment is increasingly recognized. Our findings extend previous observations of NAR in other conditions to the high-risk population of cirrhotic patients with pneumonia [13, 15]. Within hepatology, NAR represents a physiologically rational prognostic indicator that simultaneously captures two critical axes in decompensated cirrhosis with infection: neutrophilic inflammation and hypoalbuminemia, the latter reflecting diminished synthetic function and immune capacity [11, 12]. Compared to established scores like MELD, ALBI, and CURB-65, which primarily focus on single-organ dysfunction or infection severity [6–8], NAR provides an integrated snapshot of the host’s inflammatory-nutritional state. Our results indicate that NAR may offer incremental prognostic information, as its association with mortality persisted in subgroup analyses stratified by these scores. This is pertinent because mortality in this population results from the synergy between hepatic failure and systemic infection [2, 3].
The biological plausibility of NAR as a prognostic indicator is supported by the pathophysiology of cirrhosis and pneumonia [27]. Cirrhosis involves “cirrhosis-associated immune dysfunction,” featuring both systemic inflammation and immunodeficiency [28, 29]. A high NAR likely identifies patients in a state of severe dysregulation: experiencing a vigorous inflammatory response alongside failure of homeostatic and defensive mechanisms in which albumin plays a key role [30]. Moreover, the strength of the association observed—an odds ratio of approximately 2.2 per unit increase in NAR, and 2.48 for the highest versus lowest tertile—corresponds to a clinically meaningful shift in laboratory values (approximating a change in neutrophil count by ~ 3.6 × 10⁹/L or albumin by ~ 3.6 g/L). This bridges the statistical finding to a tangible clinical context, suggesting that the degree of NAR elevation is related to a substantial difference in the risk of in-hospital death. We note the conceptual overlap between WBC and the neutrophil component of NAR. Therefore, the mediation analysis should be interpreted as exploratory, quantifying the extent to which the association of NAR with mortality coincides with broader systemic inflammation as reflected by WBC. This finding is consistent with the concept of cirrhosis-associated immune dysfunction, where systemic inflammation is a key driver of poor outcomes. The link between higher NAR and worse liver disease and pneumonia severity scores further supports its utility in reflecting overall disease burden.
Strengths of this study include its multicenter design, which enhances generalizability, and the application of rigorous statistical methods including multiple imputation and comprehensive sensitivity analyses, which bolster the credibility of the findings. The exploration of WBC’s role provides contextual insight, though we acknowledge the exploratory nature of this analysis due to the conceptual overlap mentioned and the inherent limitations of mediation analysis in observational data.
However, several important limitations must be acknowledged. First, the retrospective observational design precludes causal inference, and despite multivariable adjustment, residual confounding may persist. Specifically, although we adjusted for key confounders, detailed information on microbiological etiology, antibiotic regimens, timing of treatment, and multidrug-resistant organisms—factors known to influence outcomes in cirrhotic patients with infection [7, 31]—was unavailable and represents a potential source of residual confounding. While the E-value suggests relative robustness to unmeasured confounding, the possibility of residual confounding from these specific, clinically pertinent factors cannot be excluded. Second, NAR was calculated from a single measurement at admission; its dynamic changes during hospitalization, which might offer additional prognostic information, were not assessed. Future prospective studies should incorporate serial measurements and collect detailed microbiological and treatment data.
Clinical implications and future perspectives
From a clinical perspective, the admission NAR, derived from routine blood tests, could be considered a simple and rapid adjunctive tool for initial risk stratification at the bedside. For instance, a high NAR might alert clinicians to a patient with cirrhosis and pneumonia who is at substantially increased risk of death, prompting closer monitoring, earlier consideration for intensive care unit consultation, or more aggressive management of both the infection and underlying liver dysfunction. We emphasize that these are hypothetical applications derived from an observed association. Future research should prospectively validate predefined NAR thresholds in independent cohorts and evaluate whether integrating NAR into clinical decision-making actually improves patient outcomes through targeted interventions.
Conclusion
In summary, this observational study suggests that the easily calculable admission NAR is a novel indicator strongly associated with in-hospital mortality in cirrhotic patients with pneumonia. By integrating information on systemic inflammation and hepatic synthetic function, it provides a holistic view of the patient’s risk profile. These findings highlight its potential prognostic utility. Further prospective studies are needed to validate this association in independent cohorts, explore its dynamic changes, and assess whether it can contribute to risk stratification to guide management in this vulnerable population.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
In addition, we would like to thank Jie Liu, PhD (Chinese PLA General Hospital) for their feedback.
Abbreviations
- SD
Standard deviation
- OR
Odds ratio
- CI
Confidence interval
- NAR
Neutrophil-to-albumin ratio
- Cr
Serum creatinine
- TBIL
Total bilirubin
- DBIL
Direct bilirubin
- ALT
Alanine aminotransferase
- AST
Aspartate aminotransferase
- PT/INR
Prothrombin time/international normalized ratio
- APTT
Activated partial thromboplastin time
- PT
Prothrombin time
- PLT
Platelet count
- Hb
Hemoglobin
- WBC
White blood cell count
- CAD
Coronary Atherosclerotic Heart Disease
- DM
Diabetes Mellitus
- HF
Heart Failure
- MELD score
Model for End-Stage Liver Disease score
- ALBI grade
Albumin-Bilirubin grade
- RR
Risk ratio
- ACLF
Acute-on-chronic liver failure
- NLR
Neutrophil-to-lymphocyte ratio
- CAID
Cirrhosis-associated immune dysfunction
- STROBE
Strengthening the Reporting of Observational Studies in Epidemiology
- MICE
Multivariate Imputation by Chained Equations
Author contributions
The contributions of each author are as follows: Y.Z. conceived the study, drafted the initial manuscript, and participated in data interpretation. X.C. was responsible for the statistical analysis, including software implementation and result validation. X.H., Y.G., L.S., H.Z., Y.Z., and B.L. contributed to data acquisition, curation, and database management from their respective centers. X.W. and Y.M. M. supervised the entire study, provided critical revisions to the manuscript for important intellectual content, and approved the final version for submission. All authors reviewed and agreed upon the final manuscript.
Funding
This work was supported by the Tianjin Key Medical Discipline Construction Project (Grant No. TJYXZDXK-3-001D).
Data availability
The datasets generated and analyzed during the current study are not publicly available due to patient privacy regulations, restrictions stipulated by the participating institutions’ ethical review boards, and the fact that the data form part of an ongoing research program. However, de-identified data underlying the reported findings may be made available to qualified researchers for the purpose of replicating the results, upon reasonable request to the corresponding author and subject to a formal data sharing agreement that complies with institutional and ethical guidelines. The analytic codes used in this study are available from the corresponding author upon reasonable request.
Declarations
Ethical approval
The study protocol was approved by the Ethics Committee of Beijing Mentougou District Hospital (Approval Number: 2025-QYYPJ-A01). Given the retrospective nature of the study, the requirement for informed consent was waived by the same Ethics Committee, and the entire process adhered to the principles of the Declaration of Helsinki. The study was registered with the National Medical Research Archive System (MR-11-25-033636) and the Chinese Clinical Trial Registry (Registration Number: ChiCTR2500097772; Date of Registration: February 25, 2025).
Consent for publication
Not applicable.
Declarations
The opinions stated in this article are exclusively those of the writers and may not necessarily reflect the views of their associated institutions or the publisher, editors, and reviewers.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Ye Zhang and Xian Chen contributed equally to this work.
Contributor Information
Xin Wang, Email: wangxin12091209@126.com.
Yingmin Ma, Email: mayingmin2002@126.com.
References
- 1.Devarbhavi H, Asrani SK, Arab JP, et al. Global burden of liver disease: 2023 update. J Hepatol. 2023;79(2):516–37. [DOI] [PubMed] [Google Scholar]
- 2.Arvaniti V, D’Amico G, Fede G, et al. Infections in patients with cirrhosis increase mortality four-fold and should be used in determining prognosis. Gastroenterology. 2010;139(4):1246–56. 1256.e1-5. [DOI] [PubMed] [Google Scholar]
- 3.Wong F, Piano S, Singh V, et al. Clinical features and evolution of bacterial infection-related acute-on-chronic liver failure. J Hepatol. 2021;74(2):330–9. [DOI] [PubMed] [Google Scholar]
- 4.Jalan R, Fernandez J, Wiest R, et al. Bacterial infections in cirrhosis: a position statement based on the EASL special conference 2013. J Hepatol. 2014;60(6):1310–1324. [DOI] [PubMed]
- 5.Propst-Graham KL, Preheim LC, Vander Top EA, et al. Cirrhosis-induced defects in innate pulmonary defenses against Streptococcus pneumoniae. BMC Microbiol. 2007;7(1):94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lindenmeyer CC, Flocco G, Sanghi V, et al. LIV-4: a novel model for predicting transplant-free survival in critically ill cirrhotics. World J Hepatol. 2020;12(6):298–311. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gupta T, Lochan D, Verma N, et al. Prediction of 28-day mortality in acute decompensation of cirrhosis through the presence of multidrug-resistant infections at admission. J Gastroenterol Hepatol. 2020;35(3):461–6. [DOI] [PubMed] [Google Scholar]
- 8.Hung T-H, Tseng C-W, Tsai C-C, et al. Effect of proton pump inhibitors on mortality of cirrhotic patients with pneumonia. PLoS ONE. 2019;14(4):e0216041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Department of Gastroenterology and Hepatology, the First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China, Gao F, Cai M-X, et al. Model for end-stage liver disease and pneumonia: an improved scoring model for critically ill cirrhotic patients with pneumonia. Turk J Gastroenterol. 2019;30(6):532–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Xu L, Ying S, Hu J, et al. Pneumonia in patients with cirrhosis: risk factors associated with mortality and predictive value of prognostic models. Respir Res. 2018;19(1):242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Tsao FHC, Li Z, Amessoudji AW, et al. The role of serum albumin and secretory phospholipase A2 in sepsis. Int J Mol Sci. 2024;25(17):9413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Arnau-Barrés I, Güerri-Fernández R, Luque S, et al. Serum albumin is a strong predictor of sepsis outcome in elderly patients. Eur J Clin Microbiol Infect Dis Off Publ Eur Soc Clin Microbiol. 2019;38(4):743–6. [DOI] [PubMed] [Google Scholar]
- 13.Hu Z, Wang J, Xue Y, et al. The neutrophil-to-albumin ratio as a new predictor of all-cause mortality in patients with heart failure. J Inflamm Res. 2022;15:701–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Liu C-F, Chien L-W. Predictive role of neutrophil-percentage-to-albumin ratio (NPAR) in nonalcoholic fatty liver disease and advanced liver fibrosis in nondiabetic US adults: evidence from NHANES 2017–2018. Nutrients. 2023;15(8):1892. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wan J, Liu Y, Yuan X, et al. Neutrophil-to-albumin ratio predicts stroke-associated pneumonia in patients with intracerebral hemorrhage. J Inflamm Res. 2025;18:10835–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Smith A, Baumgartner K, Bositis C. Cirrhosis: diagnosis and management. Am Fam Physician. 2019;100(12):759–70. [PubMed] [Google Scholar]
- 17.Cao B, Huang Y, She D-Y, et al. Diagnosis and treatment of community-acquired pneumonia in adults: 2016 clinical practice guidelines by the Chinese thoracic society, Chinese medical association. Clin Respir J. 2018;12(4):1320–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Johnson PJ, Berhane S, Kagebayashi C, et al. Assessment of liver function in patients with hepatocellular carcinoma: a new evidence-based approach-the ALBI grade. J Clin Oncol Off J Am Soc Clin Oncol. 2015;33(6):550–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease. Hepatol Baltim Md. 2001;33(2):464–70. [DOI] [PubMed] [Google Scholar]
- 20.Lim WS, van der Eerden MM, Laing R, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax. 2003;58(5):377–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167(4):268–74. [DOI] [PubMed] [Google Scholar]
- 22.Ciarleglio A, Petkova E, Harel O. Elucidating age and sex-dependent association between frontal EEG asymmetry and depression: an application of multiple imputation in functional regression. J Am Stat Assoc. 2022;117(537):12–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Shaw C, Wu Y, Zimmerman SC, et al. Comparison of imputation strategies for incomplete longitudinal data in life-course epidemiology. Am J Epidemiol. 2023;192(12):2075–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Garayemi S, Khoei RAA, Friedrich S. Missing data imputation in the context of propensity score analysis: a systematic review [Internet]. arXiv; 2025 [cited 2025 Dec 30]. Available from: http://arxiv.org/abs/2510.05857.
- 25.Afkanpour M, Hosseinzadeh E, Tabesh H. Identify the most appropriate imputation method for handling missing values in clinical structured datasets: a systematic review. BMC Med Res Methodol. 2024;24(1):188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Elm EV, Altman DG, Egger M, et al. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ. 2007;335(7624):806–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hasa E, Hartmann P, Schnabl B. Liver cirrhosis and immune dysfunction. Int Immunol. 2022;34(9):455–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Albillos A, Lario M, Álvarez-Mon M. Cirrhosis-associated immune dysfunction: distinctive features and clinical relevance. J Hepatol. 2014;61(6):1385–96. [DOI] [PubMed] [Google Scholar]
- 29.Wilde B, Katsounas A. Immune dysfunction and albumin-related immunity in liver cirrhosis. Mediators Inflamm. 2019;2019:7537649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Curi R, Levada-Pires AC, da Silva EB, et al. The critical role of cell metabolism for essential neutrophil functions. Cell Physiol Biochem Int J Exp Cell Physiol Biochem Pharmacol. 2020;54(4):629–47. [DOI] [PubMed] [Google Scholar]
- 31.Onorato L, Monari C, Capuano S, et al. Prevalence and therapeutic management of infections by multi-drug-resistant organisms (MDROs) in patients with liver cirrhosis: a narrative review. Antibiotics. 2022;11(2):232. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated and analyzed during the current study are not publicly available due to patient privacy regulations, restrictions stipulated by the participating institutions’ ethical review boards, and the fact that the data form part of an ongoing research program. However, de-identified data underlying the reported findings may be made available to qualified researchers for the purpose of replicating the results, upon reasonable request to the corresponding author and subject to a formal data sharing agreement that complies with institutional and ethical guidelines. The analytic codes used in this study are available from the corresponding author upon reasonable request.