Abstract
Purpose
Acute cholangitis (AC) is an emergency disease with a high risk of mortality. Red cell distribution width (RDW) has been proven to correlate with the adverse outcomes of several clinical conditions. The present study aims to investigate the association between RDW and 30-day mortality in critical AC.
Methods
This retrospective cohort study collected patients with AC from the Medical Information Mart for Intensive Care (MIMIC)-IV database. Patients were divided into tertiles according to baseline RDW values. We performed multivariable Cox regression and Kaplan-Meier survival analysis to evaluate the association between RDW and 30-day mortality. The predictive performance of RDW was assessed by the receiver operating characteristic (ROC) curve, along with integrated discrimination improvement (IDI) analysis. Subgroup analyses were employed to explore the interactions between selected covariates in RDW-mortality associations. Lastly, sex-stratified restricted cubic spline (RCS) was depicted to verify the sex-specific differences.
Results
Out of 756 patients, 146 cases experienced death within 30 days. Fully multivariable Cox regression analysis demonstrated elevated RDW levels were positively associated with an increased risk of 30-day mortality (adjusted hazard ratio [HR] 1.08, 95% confidence interval [CI] 1.01–1.15). The mortality risk followed an equidistantly gradual increase according to RDW tertiles (P for trend = 0.027). Kaplan-Meier survival analysis also displayed mortality disparities (log-rank test: P < 0.001). ROC analysis indicated that RDW exhibited an AUC of 0.733. IDI analysis revealed the significant incremental predictive value of RDW for the Sequential Organ Failure Assessment score (SOFA), Charlson Comorbidity Index (CCI), and Acute Physiology Score III (APS III) (P < 0.001 for all). The association of RDW with 30-day mortality in AC was more pronounced in males (P for interaction = 0.007). The sex-stratified RCS curves also suggested a heterogeneity in the RDW-mortality association between males and females. Sensitivity analysis corroborated the sex difference in the RDW-mortality association in AC.
Conclusion
RDW is associated with 30-day mortality in critically ill patients with AC, and this association demonstrates a heterogeneity in different sexes, underscoring the potential prognostic values of RDW in AC.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12876-025-04549-9.
Keywords: Acute cholangitis, Red cell distribution width, Mortality, Sex differences, Intensive care unit, MIMIC-IV
Introduction
Acute cholangitis (AC), commonly known as ascending cholangitis, is one of the diseases of concern due to its high risk of mortality. It is reported that the outcomes of AC differed greatly depending on the severity of the disease, with sepsis and multi-organ failure as the major causes of death [1]. Despite guidelines (e.g., Tokyo Guidelines, last updated in 2018 [TG18]) having been developed [2], challenges and difficulties for optimal diagnosis and management still exist. The sensitivity of typical presentations (e.g., absence of Charcot’s triad) for the diagnosis of AC is not satisfactory [3]; besides, current prognostic tools (e.g., TG18, Dutch Pancreatitis Study Group criteria) demonstrate limited dynamic predictive capacity [4]. Therefore, it is demanding to explore dynamic indicators that are associated with the prognosis of AC.
Red cell distribution width (RDW), a hematologic index quantifying erythrocyte size variability, originally used for anemia classification, has emerged as a bellwether of prognosis and indicator in acute medical admissions [5–7]. Wu et al. reported that high RDW levels were linked to a significant increase in 30-day mortality among critically ill patients [8]. Song et al. also found elevated RDW is associated with increased all-cause mortality in acute pancreatitis patients with sepsis [9]. RDW can be a crucial predictor of outcomes in patients with sepsis caused by Escherichia coli infection [10]. In addition, the combination of RDW with other biomarkers or illness scores, such as high-density lipoprotein, Sequential Organ Failure Assessment (SOFA) score, and Simplified Acute Physiology Score II (SAPS II), etc., can enhance the diagnostic accuracy for sepsis severity and serve as a valuable tool for risk stratification in septic patients [11, 12]. More intriguingly, Zalawadiya et al. [13] discovered that among adult Americans, RDW had a stronger correlation with cardiovascular and all-cause mortality in men than in women. Nevertheless, the sex-specific RDW prognostic values in other disease states have not been reported.
In hepatobiliary disorders, multiple studies have also found significantly elevated RDW levels in patients with chronic hepatitis B, liver cirrhosis, liver failure, and hepatocellular carcinoma [14–16]; besides, RDW, along with total bile acid or lymphocyte, can be a predictor of histologic severity in primary biliary cholangitis [17, 18]. However, most of these researches mainly focused on the association between RDW and the progression of the chronic hepatobiliary diseases. The prognostic values of RDW for the clinical outcomes of acute hepatobiliary diseases were seldom reported. In this study, we aimed to investigate the association between RDW and the 30-day mortality of critical AC, and further find out whether the association involves sex differences, thereby providing new insights into the prognostic significance of RDW in acute biliary diseases.
Methods
Study design
This retrospective analysis examined clinical data derived from the Medical Information Mart for Intensive Care-IV (MIMIC-IV v3.1) database. As a publicly available research resource, MIMIC-IV provides comprehensive critical care data from Beth Israel Deaconess Medical Center (BIDMC) in Massachusetts, United States. Given the retrospective nature and complete anonymization of all protected health information, the institutional review board at BIDMC granted ethical approval with waived individual consent requirements. Author Jie Liu obtained access to the database through PhysioNet credentialing following completion of mandatory research ethics training (Record ID: 61737483).
Patient selection
In this study, we focused on the cholangitis patients due to infection. The patient selection process was rigorously documented following Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines, with all screening phases visually presented in Fig. 1. Briefly, a total of 960 adult critical AC patients, confirmed by the 9th and 10th International Classification of Diseases (Supplementary Table S1), were first consecutively identified. The patients listed below were then eliminated one by one: first, we eliminated 15 patients with a primary sclerosing cholangitis diagnosis; next, we eliminated 20 patients without RDW measurements within 24 h of admission; and finally, we eliminated 169 patients who had an ICU stay of less than 24 h. Finally, 756 patients were recruited.
Fig. 1.
Flowchart of the patient selection.ICU intensive care unit; MIMIC-IV (V3.1) Medical Information Mart for Intensive Care-IV (Version 3.1); RDW red cell distribution width
Data extraction
Data extraction was performed through Structured Query Language (SQL) queries implemented via PostgreSQL with the Navicat Premium interface. The analytical framework was adapted from the publicly accessible MIMIC-IV code repository (https://github.com/MIT-LCP/mimic-code). Baseline characteristics comprising demographic profiles (age, sex, race), physiological parameters (cardiorespiratory indices, hemodynamic measurements), pre-existing comorbidities, and admission diagnoses were systematically recorded. Comprehensive laboratory profiles encompassing hematological indices (RDW, hemoglobin, white blood cell count [WBC]), metabolic markers (anion gap, bicarbonate), blood biochemical parameters (blood urea nitrogen [BUN], creatinine, total bilirubin, albumin, alanine aminotransferase [ALT], aspartate aminotransferase [AST], alkaline phosphatase [ALP]), and coagulation parameters (prothrombin time [PT], partial thromboplastin time [PTT]) were analyzed. Disease severity stratification scoring systems, including Charlson Comorbidity Index (CCI), SOFA, Acute Physiology Score III (APS III), SAPS II, and Oxford Acute Severity of Illness Score (OASIS), were also extracted. Only the first records of laboratory indicators within 24 h of ICU admission were selected. In the MIMIC-IV database, some electronic medical records are missing, which is quite common. The extracted data with missingness in our research was provided in Supplementary Table S2.
Study outcome
The outcome of the study was the all-cause mortality of critical AC patients within 30 days.
Statistical analysis
To reduce sample size loss and facilitate further analysis, the Multiple Imputation by Chained Equations (MICE) method was used to handle the missing data [19]. One of the five complete datasets generated during multiple imputation was randomly selected in the study. All of the analytic results were based on the selected complete dataset.
Patients were divided into tertiles according to the admission RDW values. Continuous variables were expressed as mean ± standard deviation (SD) or median (interquartile range), while categorical variables were presented as frequencies (percentages). Intergroup comparisons across RDW tertiles were performed using one-way ANOVA or Kruskal-Wallis tests for continuous variables, depending on distribution normality, and χ² tests for categorical variables.
Multivariable Cox proportional hazards models were constructed to evaluate the independent association between RDW and 30-day mortality. RDW was analyzed both as a continuous variable and categorized into tertiles (T1: reference group). The covariate selection protocol integrates: (1) clinical relevance prioritization, (2) altering RDW-30-day mortality associations by ≥ 10% when added, and (3) a variance inflation factor (VIF) less than five. The actual VIF values of the selected covariates were provided in Supplementary Table S3. Four sequential models were developed: (1) Non adjusted model; (2) Adjustment for age, sex, hemoglobin, and WBC; (3) Additional adjustment for bicarbonate, PTT, albumin, and total bilirubin; (4) Fully adjusted model incorporating CCI and APS III scores. Trend tests across tertiles were conducted using median values within each stratum.
Kaplan-Meier survival curves were generated to compare 30-day mortality risk across RDW tertiles, with between-group differences assessed using log-rank tests.
Receiver operating characteristic (ROC) curves were created to evaluate the capacity of RDW, SOFA, CCI, and APS III for predicting the risk of 30-day mortality. Additionally, the area under the ROC curve (AUC) values and integrated discrimination improvement (IDI) were used to assess the incremental predictive value of RDW.
Stratified multivariable Cox regression models were constructed to assess the consistency of RDW-mortality associations across clinically relevant subgroups. Prespecified stratification variables included age (< 72 vs. ≥ 72 years), sex (male vs. female), diabetes mellitus (no vs. yes), antibiotic use (no vs. yes), biliary drainage (no vs. yes) and sepsis (no vs. yes). Interaction tests between RDW and stratification factors were incorporated using multiplicative models.
With adjusting for the same covariates as in the fully-adjusted Cox model, a sex-stratified restricted cubic spline (RCS) model was employed to explore the sex-specific differences between RDW and 30-day mortality. RDW was treated as a continuous variable, incorporating four knots positioned at the 5th, 35th, 65th, and 95th percentiles in the males and females, respectively. The overall significance of the association was assessed using a Wald test.
Sensitivity analyses were performed by analyzing the complete cases without imputation to verify the robustness of the findings.
All statistical analyses were conducted using the statistical software packages R 4.2.2 and the Free Statistics analysis platform (Version 2.2, Beijing, China), with two-tailed P values < 0.05 being deemed statistically significant.
Results
Patient characteristics and outcomes
Table 1 demonstrates significant heterogeneity across RDW tertiles in critical AC patients (n = 756). Overall, 146 patients, accounting for 19.3% of all patients, died within 30 days. Higher RDW levels correlated with progressive hemodynamic deterioration, evidenced by descending blood pressure gradients. Hematologic parameters revealed an inverse RDW-hemoglobin relationship, with parallel leukocytosis escalation. Hepatic dysfunction markers showed divergent trends: total bilirubin median increased, while albumin decreased. Comorbidity indices, including CCI, SOFA, APS III, and SAPS II scores, progressively worsened with RDW elevation. Notably, the 30-day mortality gradient was particularly striking (T1: 6.9% vs. T3: 36.4%).
Table 1.
Baseline characteristics of patients according to RDW tertiles
| Variables | RDW tertiles | P-value | |||
|---|---|---|---|---|---|
| Total (n = 756) | T1 (n = 248) | T2 (n = 250) | T3 (n = 258) | ||
| Age, years | 72.1 ± 14.8 | 74.1 ± 14.0 | 72.9 ± 14.8 | 69.3 ± 15.1 | < 0.001 |
| Gender, n (%) | 0.181 | ||||
| Male | 402 (53.2) | 120 (48.4) | 138 (55.2) | 144 (55.8) | |
| Female | 354 (46.8) | 128 (51.6) | 112 (44.8) | 114 (44.2) | |
| Race, n (%) | 0.048 | ||||
| White | 538 (71.2) | 183 (73.8) | 185 (74.0) | 170 (65.9) | |
| Black | 43 (5.7) | 9 (3.6) | 11 (4.4) | 23 (8.9) | |
| Other | 175 (23.1) | 56 (22.6) | 54 (21.6) | 65 (25.2) | |
| Heart rate, bpm | 86.8 ± 16.9 | 86.7 ± 16.8 | 85.8 ± 17.3 | 87.9 ± 16.6 | 0.373 |
| SBP, mmHg | 112.7 ± 14.9 | 114.7 ± 15.3 | 114.2 ± 15.7 | 109.4 ± 13.1 | < 0.001 |
| DBP, mmHg | 60.2 ± 9.8 | 61.5 ± 9.6 | 60.4 ± 10.1 | 58.7 ± 9.4 | 0.004 |
| MBP, mmHg | 73.9 ± 9.4 | 75.4 ± 9.5 | 74.4 ± 9.7 | 72.0 ± 8.8 | < 0.001 |
| Respiratory rate, bpm | 19.9 ± 3.8 | 20.4 ± 3.6 | 19.8 ± 3.9 | 19.7 ± 4.0 | 0.101 |
| Temperature, ℃ | 36.9 ± 0.5 | 36.9 ± 0.5 | 36.8 ± 0.4 | 36.8 ± 0.5 | 0.015 |
| Hemoglobin, g/dL | 9.9 ± 2.0 | 10.9 ± 1.8 | 10.0 ± 1.9 | 8.8 ± 1.9 | < 0.001 |
| WBC, 109/L | 16.2 (10.5, 24.0) | 11.8 (7.6, 17.2) | 16.4 (10.8, 22.4) | 17.2 (12.1, 24.5) | < 0.001 |
| Anion gap, mEq/L | 17.2 ± 4.5 | 16.7 ± 4.2 | 17.1 ± 4.3 | 17.6 ± 4.9 | 0.071 |
| Bicarbonate, mEq/L | 20.0 ± 4.6 | 20.5 ± 4.0 | 20.5 ± 4.5 | 19.1 ± 5.1 | < 0.001 |
| BUN, mg/dL | 24.0 (16.0, 39.0) | 21.0 (14.0, 31.0) | 26.0 (17.0, 39.0) | 27.5 (16.0, 45.8) | < 0.001 |
| Creatinine, mg/dL | 1.2 (0.8, 1.8) | 1.1 (0.8, 1.6) | 1.2 (0.9, 1.9) | 1.3 (0.9, 2.0) | 0.006 |
| ALT, IU/L | 128.0 (59.0, 256.0) | 167.0 (91.0, 294.2) | 133.5 (61.8, 272.0) | 90.0 (44.0, 192.0) | < 0.001 |
| AST, IU/L | 237.0 (148.0, 413.2) | 169.5 (116.8, 270.8) | 234.0 (152.2, 396.0) | 335.0 (198.2, 515.0) | < 0.001 |
| ALP, IU/L | 141.0 (70.0, 293.0) | 156.5 (74.5, 314.0) | 136.5 (68.5, 270.0) | 132.5 (67.2, 278.5) | 0.527 |
| Total bilirubin, mg/dL | 4.1 (2.3, 6.9) | 3.5 (2.0, 5.1) | 3.9 (2.2, 6.2) | 6.1 (2.8, 11.4) | < 0.001 |
| Albumin, g/dL | 2.9 ± 0.6 | 3.1 ± 0.5 | 2.9 ± 0.5 | 2.6 ± 0.6 | < 0.001 |
| PT, sec | 20.4 ± 13.2 | 19.1 ± 15.0 | 20.3 ± 13.2 | 21.8 ± 11.1 | 0.075 |
| PTT, sec | 42.5 ± 25.3 | 38.4 ± 21.9 | 41.9 ± 25.5 | 47.1 ± 27.3 | < 0.001 |
| Congestive heart failure, n (%) | 0.187 | ||||
| No | 565 (74.7) | 182 (73.4) | 180 (72.0) | 203 (78.7) | |
| Yes | 191 (25.3) | 66 (26.6) | 70 (28.0) | 55 (21.3) | |
| Chronic pulmonary disease, n (%) | 0.896 | ||||
| No | 570 (75.4) | 185 (74.6) | 188 (75.2) | 197 (76.4) | |
| Yes | 186 (24.6) | 63 (25.4) | 62 (24.8) | 61 (23.6) | |
| Liver disease, n (%) | 0.006 | ||||
| No | 625 (82.7) | 216 (87.1) | 211 (84.4) | 198 (76.7) | |
| Yes | 131 (17.3) | 32 (12.9) | 39 (15.6) | 60 (23.3) | |
| Renal disease, n (%) | 0.019 | ||||
| No | 602 (79.6) | 212 (85.5) | 190 (76.0) | 200 (77.5) | |
| Yes | 154 (20.4) | 36 (14.5) | 60 (24.0) | 58 (22.5) | |
| CCI, scores | 6.7 ± 3.0 | 5.7 ± 2.6 | 6.5 ± 2.7 | 7.8 ± 3.2 | < 0.001 |
| SOFA, scores | 4.3 ± 2.2 | 3.9 ± 2.0 | 4.2 ± 2.1 | 4.8 ± 2.5 | < 0.001 |
| APS III, scores | 60.7 ± 26.0 | 53.5 ± 24.1 | 58.9 ± 23.8 | 69.5 ± 27.5 | < 0.001 |
| SAPS II, scores | 43.8 ± 14.8 | 40.6 ± 14.0 | 42.7 ± 13.9 | 47.9 ± 15.4 | < 0.001 |
| OASIS, scores | 34.2 ± 9.6 | 33.9 ± 9.6 | 33.9 ± 9.6 | 34.8 ± 9.5 | 0.500 |
| RDW, % | 15.8 ± 2.6 | 13.5 ± 0.6 | 15.1 ± 0.5 | 18.7 ± 2.3 | < 0.001 |
| 30-day mortality, n (%) | < 0.001 | ||||
| No | 610 (80.7) | 231 (93.1) | 215 (86.0) | 164 (63.6) | |
| Yes | 146 (19.3) | 17 (6.9) | 35 (14.0) | 94 (36.4) |
Data are presented as mean ± SD, median (IQR), or n (%)
SD standard deviation, IQR interquartile range, SBP systolic blood pressure, DBP diastolic bloodpressure, MBP mean blood pressure, WBC white blood cell count, BUN blood urea nitrogen, ALT alanine aminotransferase, AST aspartate aminotransferase, ALP alkaline phosphatase, PT prothrombin time, PTT partial thromboplastin time, CCI Charlson Comorbidity Index, SOFA Sequential Organ Failure Assessment, APS III Acute Physiology Score III, SAPS II SimplifiedAcute Physiology Score II, OASIS Oxford Acute Severity of Illness Score, RDW red cell distribution width
Multivariable cox regression analysis
Table 2 demonstrates a robust positive relationship between RDW elevation and mortality risk across analytical models. In the non-adjusted model, each 1% increase in RDW conferred a 26% higher mortality risk (HR 1.26, 95% CI 1.20–1.31, P < 0.001). After adjusting for a series of relevant covariates, this relationship persisted. The fully adjusted model (Model 3) demonstrated a consistent 8% mortality risk increase per 1% RDW increment (HR 1.08, 95% CI 1.01–1.15, P = 0.020). Meanwhile, when RDW was examined as a categorized variable, tertile analysis revealed progressive risk stratification: T3 patients exhibited 95% greater mortality versus T1 (HR 1.95, 95% CI 1.07–3.54, P = 0.028) in Model 3, with the linear trend significant (P for trend = 0.027).
Table 2.
Multivariable Cox regression analysis between RDW and 30-day mortality in critical cholangitis patients in different models
| Variables | Non-adjusted Model | Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|---|---|
| HR (95%CI) | P-value | HR (95%CI) | P-value | HR (95%CI) | P-value | HR (95%CI) | P-value | |
| RDW | 1.26 (1.20–1.31) | < 0.001 | 1.25 (1.19–1.31) | < 0.001 | 1.11 (1.05–1.18) | 0.001 | 1.08 (1.01–1.15) | 0.020 |
| RDW Tertiles | ||||||||
| T1 | Ref | Ref | Ref | Ref | ||||
| T2 | 2.13 (1.20–3.81) | 0.010 | 1.93 (1.07–3.46) | 0.028 | 1.73 (0.96–3.11) | 0.067 | 1.51 (0.83–2.73) | 0.175 |
| T3 | 6.42 (3.83–10.76) | < 0.001 | 5.31 (3.07–9.18) | < 0.001 | 2.71 (1.51–4.85) | 0.001 | 1.95 (1.07–3.54) | 0.028 |
| P for trend | < 0.001 | < 0.001 | < 0.001 | 0.027 | ||||
Model 1 adjusted for age, sex, hemoglobin and WBC; Model 2 adjusted for age, sex, hemoglobin, WBC, bicarbonate, PTT, albumin and total bilirubin; Model 3 adjusted for age, sex, hemoglobin, WBC, bicarbonate, PTT, albumin, total bilirubin, CCI and APS III
HR hazard ratio; CI confidence interval; RDW red cell distribution width; Ref reference; WBC white blood cell count; PTT partial thromboplastin time; CCI Charlson Comorbidity Index; APS III acute physiology score III
Survival analysis
A significant survival disparity was found across RDW tertiles (Fig. 2, log-rank test: P < 0.001). The T3 group exhibited a precipitous survival decline, reaching 63.6% vs. 93.1% for T1 at 30 days.
Fig. 2.
Kaplan-Meier analysis of 30-day survival in critical AC patients according to RDW tertiles. Notes: T1 (RDW < 14.4%), T2 (RDW ≥ 14.4%, < 16.2%), T3 (RDW ≥ 16.2%). The colored shadows represent 95% CIs of survival probabilities in RDW tertiles, respectively RDW red cell distribution width; T tertiles; CI confidence interval
Predictive value of RDW
ROC curves revealed that RDW had a moderate predictive accuracy for 30-day mortality, with an AUC value of 0.733 (Fig. 3), which was higher than SOFA (0.592) and CCI (0.726), but only lower than APS III (0.809). Additionally, when RDW was added to the basic model (constructed with all of the covariates in Model 3), the AUC increased from 0.878 to 0.883 (Fig. 3). The incremental predictive value of RDW for the SOFA, CCI, and APS III was also evaluated. With the addition of RDW, the AUC values for SOFA, CCI and APS III improved to 0.734, 0.793 and 0.842, and the incremental effect was statistically significant (P < 0.001 for SOFA and CCI, P = 0.007 for APS III, Table 3). IDI analysis further confirmed that RDW provided incremental predictive value over SOFA, CCI and APS III (IDI > 0, P < 0.001 for all, Table 3).
Fig. 3.
The predictive power assessment of RDW, SOFA, CCI, APS III and the basic model in critical AC patients. Notes: Basic model was constructed with age, sex, hemoglobin, WBC, bicarbonate, PTT, albumin, total bilirubin, CCI and APS III (all the adjusted covariates in Model 3 in Table 2) AC acute cholangitis; AUC area under the receiver operating characteristic curve; RDW red cell distribution width; SOFA Sequential Organ Failure Assessment; CCI Charlson Comorbidity Index; APS III Acute Physiology Score III
Table 3.
The incremental predictive value of RDW on 30-day mortality in critical AC patients
| AUC (95%CI) | P-value | IDI | P-value | |
|---|---|---|---|---|
| SOFA | 0.592 (0.541–0.644) | Ref | ||
| SOFA + RDW | 0.734 (0.687–0.782) | < 0.001 | 0.109 | < 0.001 |
| CCI | 0.726 (0.680–0.773) | Ref | ||
| CCI + RDW | 0.793 (0.753–0.832) | < 0.001 | 0.065 | < 0.001 |
| APS III | 0.809 (0.769–0.849) | Ref | ||
| APS III + RDW | 0.842 (0.806–0.877) | 0.007 | 0.071 | < 0.001 |
Subgroup analysis
Subgroup analysis was conducted, based on six stratifications: age, sex, diabetes mellitus, antibiotic use, biliary drainage, and sepsis. The forest plot demonstrated that the risk estimates for RDW and 30-day mortality remained consistent across different subgroups, except for sex strata (Fig. 4, male: HR 1.16 vs. female: HR 0.99, P for interaction = 0.007).
Fig. 4.
Forest plot for the stratified multivariable analysis of the association between RDW and 30-day mortality in critical AC patients. Notes: Except for the stratification component itself, each stratification analysis was adjusted for age, sex, hemoglobin, WBC, bicarbonate, PTT, albumin, total bilirubin, CCI, and APS III. The age group were stratified according to the mean value. Sepsis was diagnosed based on the Sepsis 3.0 Criteria RDW red cell distribution width; Ref reference; WBC white blood cell count; PTT partial thromboplastin time; CCI Charlson Comorbidity Index; APS III acute physiology score III; HR hazard ratio; CI confidence interval
Sex-stratified RCS analysis
To further verify the significant heterogeneity in RDW-mortality associations between males and females, we constructed sex-stratified RCS curves. Among males, the HR for 30-day mortality increased consistently with higher RDW values (P for overall = 0.012, Fig. 5), which followed a linear dose-response relationship (P for nonlinearity = 0.480). In contrast, the association in females was less pronounced, with a relatively flat curve and no significant trend (P for overall = 0.886, Fig. 5).
Fig. 5.
Sex-stratified restricted cubic spline curve of RDW and 30-day mortality in critical AC patients. Notes: adjusted for age, hemoglobin, WBC, bicarbonate, PTT, albumin, total bilirubin, CCI and APS III, respectively. Male (n = 402), Female (n = 354). The blue and red histograms represent the distribution of RDW values in males and females, whereas the blue and red shadows represent 95% CIs of HRs for males and females, respectively RDW red cell distribution width; WBC white blood cell count; PTT partial thromboplastin time; CCI Charlson Comorbidity Index; APS III acute physiology score III; HR hazard ratio; CI confidence interval
Sensitivity analysis
In the sensitivity analysis, complete cases without any of the covariates missing (n = 542) were included. The multivariable Cox analysis suggested an attenuated RDW-mortality association, with borderline significance (HR 1.07, 95% CI 0.99–1.15, P = 0.064) in the fully adjusted model (Supplementary Table S4). Interestingly, in the sex-stratified subgroup analysis and RCS analysis, RDW was consistently associated with 30-day mortality in males, whereas no such association was observed in females (Supplementary Table S5 and Supplementary Fig. S1), indicating the robustness of the sex differences in the association between RDW and 30-day mortality.
Discussion
In this study, we investigated the association between RDW and 30-day mortality in the critically ill patients with AC derived from the MIMIC-IV database. The results showed that elevated RDW levels were associated with the 30-day mortality of critical AC, even after adjusting for a series of confounders. This relationship remained consistent across RDW tertiles, which was corroborated by the Kaplan-Meier survival analysis. ROC curve analysis indicated that RDW exhibited a moderate predictive capability. Incorporating RDW into existing risk stratification tools (SOFA, CCI, and APS III) can significantly enhance their predictive values for 30-day mortality in critical AC. More interestingly, we identified a statistically significant interaction in RDW-mortality associations between male and female groups in subgroup analysis. Male patients demonstrated a more pronounced relationship for the RDW-mortality association than their female counterparts. Collectively, these findings indicate that RDW may serve as a potential biomarker for AC mortality risk stratification.
Emerging evidence positions RDW as a robust prognostic biomarker across diverse critical illnesses. Notably, COPD patients experiencing acute exacerbations demonstrated a significant association between elevated RDW levels and 28-day all-cause mortality [20]. In acute pancreatitis cohorts, admission RDW also served as a predictor of 30-day adverse clinical outcomes [21]. Moreover, the biomarker’s prognostic value extends to cerebrovascular emergencies, showing statistically significant associations with both stroke-associated pneumonia risk and acute ischemic stroke patients treated with thrombolysis [22]. In a similar study that included 104 patients, Yildiz et al. identified RDW as a factor predicting mortality in acute suppurative cholangitis [23]. Consistent with their findings, our study also demonstrated a statistically significant association between RDW and 30-day mortality in critically ill patients with AC, with a much larger sample (756 patients), multivariable adjustment, and comprehensive survival analysis. Besides, combined biomarkers based on RDW, such as RDW-to-albumin ratio (RAR) and hemoglobin-to-RDW ratio (HRR), have also been proved to correlate with the prognosis of several clinical conditions [24, 25]. In recent research from Ankara City Hospital, Acehan et al. found that RAR at admission can be a prognostic marker in AC patients requiring biliary drainage [26]. Similarly, our study showed that RDW, as a single biomarker, can be substantially linked to the short-term outcomes of AC in a relatively large-sample US patient population, adding new evidence to the prognostic value of RDW for demographic diversity in different populations.
As is known, critical AC is a severe infection of the biliary system, mostly accompanied by liver dysfunction and systemic inflammatory response. Cumulative evidence suggests RDW showed a strong, independent association with established systemic inflammatory markers such as C-reactive protein and erythrocyte sedimentation rate [27–29]. Furthermore, RDW is also positively correlated with markers of liver dysfunction (bilirubin, CCI) and negatively correlated with albumin [30, 31]. Under acute inflammatory conditions, elevated RDW may reflect inflammation severity, oxidative stress, and metabolic or nutritional imbalances [32, 33]. Hence, one possible inference is that RDW, along with other inflammatory markers and oxidative stressors, constitutes a complex regulatory network, which influences the prognosis of critical AC.
Importantly, we found a significant heterogeneity in the RDW-mortality associations between male and female AC patients, with a statistically significant linear dose-response association in male groups, while the female counterparts exhibited a non-significant dose-response association across RDW ranges. Interestingly, in a recent study of critically ill patients, Lee et al. found that, although both sexes showed increased one-year mortality risk with higher RDW, males demonstrated a more pronounced effect [34]. While the specific pathophysiological mechanisms for the difference of prognostic value of RDW in different sexes remained unclear, the stronger inflammation and oxidative stress in male patients can be a plausible explanation for the worse outcomes. In the state of illness, males tend to exhibit stronger oxidative stress than females [35, 36]. Meanwhile, higher RDW levels are associated with greater oxidative stress and persistent inflammation, a well-recognized cause of disease progression. In addition, the protective role of estrogen against iron-dependent cell death might be another potential mechanism for explaining the sex-specific differences [37]. However, it is not always the case that men’s risk is consistently higher than women’s. In frailty, women demonstrated a dramatically higher risk of frailty with elevated RDW [38]. For the prevalence of cardiovascular diseases, the association was also more pronounced in females [39]. Collectively, sex differences for the association between RDW and diseases may vary across different diseases, suggesting complex underlying biological mechanisms that warrant further investigation. Anyway, our study indicated that RDW may serve as a more robust predictor of 30-day mortality in male AC patients compared to the female groups.
RDW appears to be a reliable predictor for AC, which has been demonstrated to enhance traditional risk stratification when integrated into predictive scoring systems [23]. Our results provided additional evidence that RDW can be an effective prognostic indicator for AC. Furthermore, we have confirmed that incorporating RDW into traditional disease scoring tools (e.g., SOFA, CCI, and APS III) significantly enhances their predictive value for mortality. Accordingly, predictive models or innovative machine learning-based risk modeling frameworks can be developed by combining these indices to direct future risk assessment and treatment.
A strength of this study is represented by the comprehensive exploration of RDW-mortality, including covariate adjustment, subgroup multivariable analysis, and sex-stratified RCS analysis. Meanwhile, it should be mentioned that there are still several limitations in this study. First, given the retrospective nature, the study is unable to conclusively infer causality. Meanwhile, the results might be impacted by unmeasured factors, especially the potential sex-specific confounders. Nevertheless, we adjusted for the risk factors as much as the sample size would allow, and performed subgroup analysis to verify the robustness of our conclusions. Second, there are some missing values for key parameters due to inherent limitations of the database, so the sample size was impacted by our strict exclusion criteria. However, we believe that this exclusion process was essential to draw convincing conclusions. Third, we primarily focused on the baseline RDW levels at ICU admission, but the dynamic changes in RDW during the ICU stay or time-varying confounders that could influence both RDW levels and mortality risk were not included in our study. Meanwhile, the diagnosis of cholangitis was determined by the ICD codes, which cannot reflect the severity of the patient’s condition. Fourth, the study sample was selected from one U.S. center, which might have limited the generalization of the results. Future randomized controlled trials are needed to further assess the potential impact of RDW on the short-time outcomes of critical AC in the multicenter.
Conclusion
Higher RDW is linked to an increased risk of 30-day mortality in a US critically ill population with AC. RDW may serve as a practical tool for AC risk stratification. Moreover, male patients demonstrated a relatively high risk in the RDW-mortality association, indicating clinicians should pay more attention to this subgroup.
Supplementary Information
Acknowledgements
We are grateful to all the BIDMC and Massachusetts Institute of Technology researchers and staff who contributed to the MIMIC-IV database.
Authors’ contributions
Jie Liu conceived the study idea, collected the data, and wrote the first draft; Yang Song analyzed the results; Dongxin Zhang assisted with manuscript editing and revising; and Jiajun Ji suggested critical revisions and supervised the study. Both authors have read and approved the final version of the manuscript.
Funding
None.
Data availability
The MIMIC IV database (v3.1) at https://www.physionet.org/content/mimiciv/3.1/contains the data that are used to support the study’s conclusions. The corresponding author can provide detailed data upon reasonable request, provided that researchers had access to the MIMIC database.
Declarations
Ethics approval and consent to participate
Not applicable. The Medical Information Mart for Intensive Care-IV (MIMIC-IV) is a publicly available research resource that offers comprehensive critical care data from the Beth Israel Deaconess Medical Center (BIDMC). BIDMC’s institutional review board granted ethical approval while waiving the need for individual consent. Author Jie Liu obtained access to the database through PhysioNet credentialing following completion of mandatory research ethics training (Record ID: 61737483, 14 March, 2024).
Consent for publication
Not applicable.
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.
References
- 1.Lee JM, Lee SH, Chung KH, et al. Risk factors of organ failure in cholangitis with bacteriobilia. World J Gastroenterol. 2015;21:7506–13. 10.3748/wjg.v21.i24.7506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Miura F, Okamoto K, Takada T, et al. Tokyo guidelines 2018: initial management of acute biliary infection and flowchart for acute cholangitis. J Hepato-Biliary-Pancreat Sci. 2018;25:31–40. 10.1002/jhbp.509. [DOI] [PubMed] [Google Scholar]
- 3.Rumsey S, Winders J, MacCormick AD. Diagnostic accuracy of charcot’s triad: A systematic review. ANZ J Surg. 2017;87:232–8. 10.1111/ans.13907. [DOI] [PubMed] [Google Scholar]
- 4.Sperna Weiland CJ, Busch CBE, Bhalla A, et al. Performance of diagnostic tools for acute cholangitis in patients with suspected biliary obstruction. J Hepato-Biliary-Pancreat Sci. 2022;29:479–86. 10.1002/jhbp.1096. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Chen Y, Chen Y, Ming L, et al. Red blood cell distribution width as a risk factor for 30/90-day mortality in patients with Gastrointestinal bleeding: analysis of the MIMIC-IV database. Dig Dis Sci. 2024;69:1740–54. 10.1007/s10620-024-08295-y. [DOI] [PubMed] [Google Scholar]
- 6.Conway R, Byrne D, O’Riordan D, et al. Red cell distribution width as a prognostic indicator in acute medical admissions. J Clin Med. 2023;12:5424. 10.3390/jcm12165424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lichtman MA. Red cell distribution width as a bellwether of prognosis. Blood Cells Mol Dis. 2024;109:102884. 10.1016/j.bcmd.2024.102884. [DOI] [PubMed] [Google Scholar]
- 8.Wu Y-C, Chen H-H, Chao W-C. Association between red blood cell distribution width and 30-day mortality in critically ill septic patients: a propensity score-matched study. J Intensive Care. 2024;12:34. 10.1186/s40560-024-00747-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Song Q, Wu X, Taheri FA, et al. Association of red blood cell distribution width with short- and long-term all-cause mortality in patients with acute pancreatitis and sepsis. BMC Gastroenterol. 2025;25:539. 10.1186/s12876-025-04167-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Qin K, Su Y, Ding N. Red cell distribution width and clinical outcomes in sepsis patients infected with Escherichia coli using data from MIMIC-IV. Eur J Med Res. 2025;30:580. 10.1186/s40001-025-02756-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Li X, He Y, Lu X, et al. The predictive value of platelet parameters and hemorheological indicators for elderly sepsis patients. BMC Anesthesiol. 2025;25:391. 10.1186/s12871-025-03222-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gao Y, Chen Y, Gao L. Evaluation of sepsis severity using combined high-density lipoprotein and red cell distribution width indicators. Br J Hosp Med. 2024;85:1–12. 10.12968/hmed.2024.0473. [DOI] [PubMed] [Google Scholar]
- 13.Zalawadiya SK, Veeranna V, Panaich SS, et al. Gender and ethnic differences in red cell distribution width and its association with mortality among low risk healthy united state adults. Am J Cardiol. 2012;109:1664–70. 10.1016/j.amjcard.2012.01.396. [DOI] [PubMed] [Google Scholar]
- 14.Aslam H, Oza F, Ahmed K, et al. The role of red cell distribution width as a prognostic marker in chronic liver disease: a literature review. Int J Mol Sci. 2023;24:3487. 10.3390/ijms24043487. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kalairajan S, K K K PG. Red cell distribution width in chronic liver disease: an observational study. Cureus. 2023;15:e40158. 10.7759/cureus.40158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Tan M, Liu B, You R, et al. Red blood cell distribution width as a potential valuable survival predictor in hepatitis B virus-related hepatocellular carcinoma. Int J Med Sci. 2023;20:976–84. 10.7150/ijms.79619. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Meng J, Xu H, Liu X, et al. Increased red cell width distribution to lymphocyte ratio is a predictor of histologic severity in primary biliary cholangitis. Med (Baltim). 2018;97:e13431. 10.1097/MD.0000000000013431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Tu Z, Wang Y, Wang Y, et al. TR score: a noninvasive model to predict histological stages in patients with primary biliary cholangitis. Front Immunol. 2023;14:1152294. 10.3389/fimmu.2023.1152294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.van Buuren S, Groothuis-Oudshoorn K. Mice: multivariate imputation by chained equations in R. J Stat Softw. 2011;45:1–67. 10.18637/jss.v045.i03. [Google Scholar]
- 20.Huang Y, Wang J, Chen S, et al. Red cell distribution width related to the severity of the disease in patients with AECOPD. Biomarkers Med. 2025;19:177–86. 10.1080/17520363.2025.2468152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Azzam SZ, Ghersin I, Fischman M, et al. The association between red blood cell distribution width at admission and early mortality in acute pancreatitis. Isr Med Assoc J: IMAJ. 2025;27:87–91. [PubMed] [Google Scholar]
- 22.Krongsut S, Na-Ek N, Khongthon N. Admission red blood cell distribution width as a prognostic biomarker of stroke-associated pneumonia and mortality in acute ischemic stroke patients treated with thrombolysis. J Stroke Cerebrovasc Dis: Off J Natl Stroke Assoc. 2025;34:108254. 10.1016/j.jstrokecerebrovasdis.2025.108254. [DOI] [PubMed] [Google Scholar]
- 23.Yıldız BD, Özden S, Saylam B, et al. Simplified scoring system for prediction of mortality in acute suppurative cholangitis. Kaohsiung J Med Sci. 2018;34:415–9. 10.1016/j.kjms.2017.12.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wu Q, Liao L, Deng Q. Association between red blood cell distribution width-to-albumin ratio at admission and all-cause mortality in patients with acute pancreatitis based on the MIMIC-III database. PLoS ONE. 2025;20:e0318873. 10.1371/journal.pone.0318873. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wang Y, Chen D, Zhang C, et al. Unveiling the prognostic power of HRR in ICU-admitted COPD patients: a MIMIC-IV database study. Int J Chronic Obstr Pulm Dis. 2025;20:11–21. 10.2147/COPD.S482344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Acehan F, Camli H, Kalkan C, et al. Red cell distribution width-to-albumin ratio is a simple promising prognostic marker in acute cholangitis requiring biliary drainage. Hepatobiliary Pancreat Dis Int. 2024;23:487–94. 10.1016/j.hbpd.2023.08.005. [DOI] [PubMed] [Google Scholar]
- 27.Targoński R, Sadowski J, Starek-Stelmaszczyk M, et al. Prognostic significance of red cell distribution width and its relation to increased pulmonary pressure and inflammation in acute heart failure. Cardiol J. 2020;27:394–403. 10.5603/CJ.a2018.0103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lippi G, Targher G, Montagnana M, et al. Relation between red blood cell distribution width and inflammatory biomarkers in a large cohort of unselected outpatients. Arch Pathol Lab Med. 2009;133:628–32. 10.5858/133.4.628. [DOI] [PubMed] [Google Scholar]
- 29.González-Sierra M, Romo-Cordero A, Quevedo-Abeledo JC, et al. Red cell distribution width association with subclinical cardiovascular disease in patients with rheumatoid arthritis. J Clin Med. 2023;12:6497. 10.3390/jcm12206497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Vidili G, Zinellu A, Mangoni AA, et al. Red cell distribution width as a predictor of survival in patients with hepatocellular carcinoma. Med (Kaunas Lith). 2024;60:391. 10.3390/medicina60030391. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Jing J-S, Fu X-L, Zhao W, et al. Red cell distribution width as a prognostic factor in patients with hepatocellular carcinoma. Clin Lab. 2020;66. 10.7754/Clin.Lab.2019.191027. [DOI] [PubMed]
- 32.Li N, Zhou H, Tang Q. Red blood cell distribution width: a novel predictive indicator for cardiovascular and cerebrovascular diseases. Dis Markers. 2017;2017:7089493. 10.1155/2017/7089493. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Salvagno GL, Sanchis-Gomar et al. Fabian, Picanza, Alessandra,. Red blood cell distribution width: a simple parameter with multiple clinical applications. Crit Rev Clin Lab Sci. 2015;52:86–105. 10.3109/10408363.2014.992064 [DOI] [PubMed]
- 34.Lee S-Y, Wong L-T, Chao W-C. The association between week-one red blood cell distribution width and one-year survival in critically ill patients: propensity score-based multicenter analysis. Eur J Med Res. 2025;30:551. 10.1186/s40001-025-02839-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Popotas A, Casimir GJ, Corazza F, et al. Sex-related immunity: could toll-like receptors be the answer in acute inflammatory response? Front Immunol. 2024;15. 10.3389/fimmu.2024.1379754. [DOI] [PMC free article] [PubMed]
- 36.Borrás C, Sastre J, García-Sala D, et al. Mitochondria from females exhibit higher antioxidant gene expression and lower oxidative damage than males. Free Radical Biol Med. 2003;34:546–52. 10.1016/s0891-5849(02)01356-4. [DOI] [PubMed] [Google Scholar]
- 37.Vanden Berghe T. Oestrogen defends against kidney damage caused by iron-dependent cell death. Nature. 2025;645:862–4. 10.1038/d41586-025-02422-z. [DOI] [PubMed] [Google Scholar]
- 38.Li Q, Chen X, Han B. Red blood cell distribution width is associated with frailty in older inpatients in china: sex differences in a cross-sectional study. Exp Gerontol. 2021;150:111392. 10.1016/j.exger.2021.111392. [DOI] [PubMed] [Google Scholar]
- 39.Ainiwaer A, Kadier K, Abulizi A, et al. Association of red cell distribution width (RDW) and the RDW to platelet count ratio with cardiovascular disease among US adults: a cross-sectional study based on the National health and nutrition examination survey 1999–2020. BMJ Open. 2023;13:e068148. 10.1136/bmjopen-2022-068148. [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 MIMIC IV database (v3.1) at https://www.physionet.org/content/mimiciv/3.1/contains the data that are used to support the study’s conclusions. The corresponding author can provide detailed data upon reasonable request, provided that researchers had access to the MIMIC database.