Abstract
Background
The prognostic significance of the hemoglobin-to-red blood cell distribution width ratio (HRR) for 28-day all-cause mortality remains incompletely characterized in patients with acute kidney injury.
Methods
This cross-sectional study included 27,670 patients with acute kidney injury from the Medical Information Mart for Intensive Care-IV database (2008–2022). Multivariable Cox regression, Kaplan–Meier survival analysis, and receiver operating characteristic curve analysis were performed.
Results
Patients were stratified by HRR tertiles: T1 (1.84–6.4), T2 (6.4–8.37), and T3 (8.37–15.5). Mortality rates differed significantly across tertiles (25.2% vs. 14% vs. 8.9%, p < 0.001). Kaplan–Meier analysis showed a significantly higher mortality risk in the lower HRR tertiles (log-rank p < 0.0001). After full adjustment for potential confounders, the hazard ratios were 0.68 (95% confidence interval: 0.63–0.74) for T2 and 0.58 (95% confidence interval: 0.52–0.65) for T3, compared with T1. Receiver operating characteristic curve analysis identified an optimal HRR cutoff value of 6.53 (area under the curve: 0.657, 95% confidence interval: 0.648–0.666) for risk stratification.
Conclusion
Lower HRR values were independently associated with an increased risk of 28-day mortality in critically ill patients, with a practical cutoff of 6.53 for clinical application.
Keywords: Hemoglobin-to-red blood cell distribution width ratio, acute kidney injury, intensive care unit patients, all-cause mortality, critical care
Introduction
Acute kidney injury (AKI) is a common and life-threatening clinical condition characterized by a sudden decline in kidney function, often leading to systemic organ failure and increased mortality. 1 This condition affects approximately 20% of hospitalized patients worldwide, constituting a major contributor to in-hospital deaths. 2 In critical care settings, AKI prevalence is concerning, with reported incidence rates ranging between 33% and 67% among critically ill adults and mortality reaching up to 28%.3,4 Its clinical course is often complicated by subsequent development of chronic kidney disease (CKD), resulting in substantial long-term health implications. 5 Despite advances in intensive care and renal replacement therapies, effective treatment options for AKI remain limited. 6 Continuous renal replacement therapy (CRRT) has become an important treatment modality for critically ill patients with AKI, offering advantages through hemodynamic stabilization and sustained blood purification processes. 7 However, mortality rates among patients receiving CRRT remain concerning at approximately 63%, underscoring the need for improved biomarkers and predictive methodologies to optimize outcomes. 8 The International Society of Nephrology’s “0 by 25” initiative highlights the global commitment to eliminating preventable AKI-related deaths by 2025. 2 Overall, AKI continues to pose a significant clinical challenge in intensive care settings, highlighting the importance of early identification, novel therapeutic approaches, and enhanced patient care protocols to reduce associated morbidity and mortality.
The hemoglobin-to-red blood cell distribution width ratio (HRR) is a composite biomarker calculated by dividing hemoglobin levels (g/L) by red blood cell distribution width (RDW; %). 9 This parameter integrates prognostic data from hemoglobin levels, which reflect anemic states, and RDW values, an established inflammatory biomarker.10,11 Scientific evidence has established the prognostic value of RDW in inflammation-driven cardiovascular pathologies, including atherosclerosis 12 and ischemic heart disease, 13 and its utility in forecasting severity and prognosis of acute ischemic stroke (AIS). 14 First reported by Sun et al., 15 HRR has emerged as a prognostic indicator across diverse clinical settings, including mortality prediction in patients with heart failure, 13 cardiovascular readmissions, 13 and oncologic survival outcomes. 9 Furthermore, HRR has shown potential in predicting adverse events in patients with sepsis and those with atrial fibrillation (AF). 16 The practical advantages of HRR in routine clinical practice, including its accessibility and economic feasibility, position it as a valuable marker for chronic inflammatory processes and prognostic assessment. Although HRR demonstrates extensive clinical applicability and significant associations with numerous pathological conditions, its relationship with all-cause mortality among intensive care unit (ICU) patients with AKI remains unexplored.
Based on current evidence, we conducted this study to evaluate the predictive value of HRR for 28-day all-cause mortality in ICU patients with AKI through comprehensive analysis of the Medical Information Mart for Intensive Care (MIMIC)-IV database. We examined the relationship between reduced admission HRR levels and unfavorable clinical outcomes to determine whether HRR can serve as a clinically useful predictor for identifying patients at elevated risk.
Methods
Study population
We conducted a cross-sectional study using data from the MIMIC-IV database, covering the period from 2008 to 2022 (https://physionet.org/content/mimiciv/3.0/).17–19 The MIMIC-IV database contains clinical records from more than 90,000 adult ICU admissions across the United States and is publicly available for research purposes. The Institutional Review Board of Harbin Medical University Cancer Hospital approved the study and waived the requirement for written informed consent due to its retrospective design and use of publicly available data. All methodological procedures adhered to established guidelines and regulatory standards. Clinical data were obtained from the ICUs of Beth Israel Deaconess Medical Center and included physiological measurements, laboratory findings, therapeutic interventions, and medication records. The author, Yangang Zhu, successfully completed Collaborative Institutional Training Initiative (CITI) program requirements (certification number: 64180628) and received authorization to use the database for research purposes. All investigators involved in data assessment completed mandatory human participants research training and executed data use agreements. The study was conducted in accordance with the Declaration of Helsinki 1975, as revised in 2024. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
The inclusion criteria were as follows: (a) age ≥18 years; (b) ICU length of stay >1 day; (c) first ICU admission during the index hospitalization; (d) fulfillment of AKI diagnostic criteria, defined as an increase in serum creatinine levels to 1.5 times the baseline within 7 days, an increase of ≥0.3 mg/dL within 48 h, or persistent oliguria (urine output < 0.5 mL/kg/h) lasting ≥6 h; and (e) complete data on covariates and survival outcomes. A total of 27,670 patients met these criteria and were included in the final analysis (Figure 1). AKI was diagnosed using Kidney Disease: Improving Global Outcomes (KDIGO) criteria throughout the study period.
Figure 1.
Flow chart of patient’s selection.
Data collection
We used PostgreSQL (version 16) with Structured Query Language (SQL) commands to extract comprehensive clinical data from the database. The acquisition process encompassed the following: (a) Baseline characteristics. Age (years), sex (male or female), ethnicity (White, Black, or Other), and marital status (single, married, divorced, or widowed); (b) Biochemical and physiological measurements. White blood cell (WBC) count (K/µL), red blood cell (RBC) count (M/µL), hemoglobin concentration (g/dL), RDW (%), serum sodium concentration (mEq/L), serum potassium concentration (mEq/L), blood urea nitrogen levels (BUN; mg/dL), serum creatinine levels (mg/dL), heart rate (beats/min), systolic blood pressure (mmHg), diastolic blood pressure (mmHg), respiratory rate (breaths/min), oxygen saturation (SpO2, %), and body temperature (°C); (c) Pre-existing medical conditions. Arterial hypertension, diabetes mellitus, chronic obstructive pulmonary disease (COPD), myocardial infarction, malignant neoplasms, and cerebrovascular disease.
The primary endpoint was 28-day ICU mortality, defined as death from any cause occurring within the ICU during hospitalization. Patient surveillance continued throughout the entire ICU stay, with termination points established as either in-ICU death or ICU discharge. The median observation period measured 4.72 days (SD, 5.88 days).
HRR definition
HRR was calculated as hemoglobin concentration (g/L) divided by RDW (%): HRR = hemoglobin concentration (g/L) / RDW (%) Study participants were stratified into three groups according to HRR tertile distributions: (a) Lower tertile (T1). HRR values ranging from 1.84 to 6.4; (b) Middle tertile (T2). HRR values spanning 6.4 to 8.37; and (c) Upper tertile (T3). HRR values ranging from 8.37 to 15.5.
Statistical analysis
Quantitative variables were expressed as mean ± SD, whereas qualitative variables were presented as frequencies and percentages. Comparative analysis of continuous variables were performed using Student’s t-test, one-way analysis of variance (ANOVA), or the Kruskal–Wallis nonparametric test, as appropriate. Categorical variables were compared using the chi-square test or Fisher’s exact test when indicated.
We constructed sequential Cox proportional hazards regression models to evaluate the association between HRR tertiles and 28-day all-cause mortality. Results were reported as hazard ratios (HRs) with corresponding 95% confidence intervals (CIs). To mitigate potential model overfitting arising from multicollinearity, we calculated variance inflation factors (VIF) and eliminated variables with VIF ≥3 (Table S1). Our modeling strategy comprised the following: Model 1, crude analysis without adjustment; Model 2, adjusted for sociodemographic characteristics, including age, sex, ethnicity, and marital status; Model 3, further adjusted for biochemical and physiological variables, including WBC, RBC, sodium, potassium, glucose, BUN, creatinine, heart rate, systolic and diastolic blood pressure, respiratory rate, SpO2, and body temperature; and Model 4, additionally adjusted for all Model 3 covariates along with comorbidities, including arterial hypertension, diabetes mellitus, myocardial infarction, malignant neoplasms, cerebrovascular disease, and COPD. Linear trend assessments across tertiles were conducted using ordinal variables. The discriminative ability of models was evaluated using Harrell’s concordance index (C-index), with values >0.7 indicating good predictive performance. The proportional hazards assumption was tested using Schoenfeld residuals; p > 0.05 indicated that the assumption was met. Kaplan–Meier survival probability estimates were generated to illustrate temporal mortality patterns, and between-group comparisons were conducted using log-rank test.
To minimize selection bias and confounding, we conducted propensity score matching (PSM) between the lowest (T1) and highest (T3) HRR tertiles as a sensitivity analysis. Propensity scores were estimated using logistic regression, incorporating demographic variables (age, sex, race, and marital status), laboratory parameters (WBC, RBC, electrolytes, glucose, BUN, and creatinine), vital signs (heart rate, blood pressure, respiratory rate, SpO2, and body temperature), and comorbidities (hypertension, diabetes, myocardial infarction, malignancy, stroke, and COPD). One-to-one nearest-neighbor matching without replacement was performed using a caliper width of 0.2 SDs. Covariate balance was evaluated using standardized mean differences (SMD), with SMD <0.1 indicating adequate balance. Cox regression models were applied to the matched cohort with progressive adjustment as described previously.
Subgroup analyses were conducted to evaluate the prognostic consistency of HRR across predefined subgroups stratified by sex, ethnicity, arterial hypertension, diabetes mellitus, myocardial infarction, and malignant neoplasms. Effect modification between HRR and stratification variables was assessed through likelihood ratio tests.
To establish a clinically applicable threshold for risk stratification based on HRR values, we performed receiver operating characteristic (ROC) curve analysis. The area under the curve (AUC) was calculated to evaluate discriminative ability of HRR for predicting 28-day all-cause mortality. The optimal cutoff value was determined using the Youden index, which maximizes the sum of sensitivity and specificity. Statistical significance was established at α value of 0.05 using two-tailed hypothesis testing. All computational analyses were performed using R statistical software (version 4.4.0; https://cran.r-project.org/).
Results
Baseline characteristics
Our analytical cohort comprised data from 27,670 patients with AKI identified from the MIMIC-IV database. The study population had a mean age of 67.3 ± 15.9 years. Sex distribution included 15,688 males (56.7%) and 11,982 females (43.3%). The ethnic composition was 71.6% White, 10.3% Black, and 18.1% categorized as Other ethnicities. The mean HRR was 7.43 ± 2.03. Throughout the ICU hospitalization period, 4441 patients (16%) died. Table 1 presents the baseline characteristics stratified by HRR tertiles: T1 (5.18 ± 0.849), T2 (7.39 ± 0.565), and T3 (9.72 ± 1.04). In terms of marital status, 28.5% were single, 49.3% were married, 8% were divorced, and 14.2% were widowed. Patients in the highest HRR tertile had a lower prevalence of diabetes mellitus, malignant neoplasms, and COPD than those in the lowest tertile. Mortality rates demonstrated significant inter-tertile variation (25.2% vs. 14% vs. 8.9%, p < 0.001), suggesting an inverse association between HRR and 28-day all-cause mortality.
Table 1.
Baseline characteristics of patients according to HRR tertiles.
| Characteristics | Tertiles |
Total (n = 27,670) | p-value | ||
|---|---|---|---|---|---|
| T1 (n = 9234) | T2 (n = 9220) | T3 (n = 9216) | |||
| Age, years | 68.3 (15.4) | 69.6 (15.2) | 64.1 (16.5) | 67.3 (15.9) | <0.001 |
| Sex, n (%) | |||||
| Male | 4733 (51.3%) | 4799 (52.0%) | 6156 (66.8%) | 15,688 (56.7%) | <0.001 |
| Female | 4501 (48.7%) | 4421 (48.0%) | 3060 (33.2%) | 11,982 (43.3%) | |
| Ethnicity, n (%) | |||||
| White | 6464 (70.0%) | 6640 (72.0%) | 6702 (72.7%) | 19,806 (71.6%) | <0.001 |
| Black | 1161 (12.6%) | 940 (10.2%) | 763 (8.3%) | 2864 (10.3%) | |
| Other | 1609 (17.4%) | 1640 (17.8%) | 1751 (19.0%) | 5000 (18.1%) | |
| Marital status, n (%) | |||||
| Single | 2697 (29.2%) | 2426 (26.3%) | 2771 (30.1%) | 7894 (28.5%) | <0.001 |
| Married | 4286 (46.4%) | 4537 (49.2%) | 4814 (52.2%) | 13,637 (49.3%) | |
| Divorced | 780 (8.4%) | 756 (8.2%) | 672 (7.3%) | 2208 (8.0%) | |
| Widowed | 1471 (15.9%) | 1501 (16.3%) | 959 (10.4%) | 3931 (14.2%) | |
| WBC, k/μL | 12.7 (12.4) | 12.2 (7.86) | 12.0 (5.45) | 12.3 (9.07) | <0.001 |
| RBC, m/μL | 3.02 (0.550) | 3.59 (0.499) | 4.25 (0.524) | 3.62 (0.729) | <0.001 |
| HB, g/dL | 8.74 (1.11) | 10.7 (1.05) | 13.0 (1.34) | 10.8 (2.10) | <0.001 |
| RDW, % | 17.1 (2.55) | 14.5 (1.30) | 13.4 (0.927) | 15.0 (2.34) | <0.001 |
| Serum sodium, mEq/L | 138 (5.42) | 138 (4.85) | 138 (4.51) | 138 (4.95) | <0.001 |
| Serum potassium, mEq/L | 4.29 (0.652) | 4.20 (0.559) | 4.13 (0.515) | 4.21 (0.581) | <0.001 |
| Glucose, mmol/dL | 7.81 (3.11) | 7.84 (2.89) | 7.86 (2.96) | 7.84 (2.99) | 0.015 |
| Cr, mg/dL | 1.98 (2.12) | 1.42 (1.48) | 1.08 (0.957) | 1.49 (1.64) | <0.001 |
| BUN, mg/dL | 36.0 (27.4) | 25.9 (19.6) | 19.5 (13.2) | 27.1 (22.0) | <0.001 |
| HR, bpm | 87.4 (16.8) | 85.1 (16.0) | 83.4 (16.5) | 85.3 (16.5) | <0.001 |
| SBP, mmHg | 115 (17.4) | 117 (17.5) | 122 (17.6) | 118 (17.7) | <0.001 |
| DBP, mmHg | 61.6 (11.7) | 63.5 (11.6) | 69.6 (11.8) | 64.9 (12.2) | <0.001 |
| RR, insp/min | 19.7 (4.25) | 19.4 (3.93) | 19.2 (3.75) | 19.4 (3.99) | <0.001 |
| SpO2, % | 96.9 (2.05) | 96.7 (2.09) | 96.4 (1.97) | 96.7 (2.05) | <0.001 |
| Temperature, °C | 36.8 (1.32) | 36.8 (1.48) | 36.8 (1.42) | 36.8 (1.41) | <0.001 |
| Hypertension, n (%) | 3059 (33.1%) | 4302 (46.7%) | 4799 (52.1%) | 12,160 (43.9%) | <0.001 |
| Diabetes, n (%) | 3263 (35.3%) | 2964 (32.1%) | 2194 (23.8%) | 8421 (30.4%) | <0.001 |
| Myocardial infarction, n (%) | 726 (7.9%) | 720 (7.8%) | 736 (8.0%) | 2182 (7.9%) | 0.976 |
| Malignant tumor, n (%) | 2038 (22.1%) | 1663 (18.0%) | 1236 (13.4%) | 4937 (17.8%) | <0.001 |
| Stroke, n (%) | 718 (7.8%) | 865 (9.4%) | 965 (10.5%) | 2548 (9.2%) | <0.001 |
| COPD, n (%) | 867 (9.4%) | 829 (9.0%) | 591 (6.4%) | 2287 (8.3%) | <0.001 |
| HRR, mean (SD) | 5.18 (0.849) | 7.39 (0.565) | 9.72 (1.04) | 7.43 (2.03) | <0.001 |
| 28 days all-cause mortality, n (%) | 2327 (25.2%) | 1290 (14.0%) | 824 (8.9%) | 4441 (16.0%) | <0.001 |
BUN: blood urea nitrogen; COPD: chronic obstructive pulmonary disease; Cr: creatinine; DBP: diastolic blood pressure; HB: hemoglobin; HR: heart rate; HRR: hemoglobin-to-red cell distribution width ratio; RBC: red blood cell; RDW: red blood cell distribution width; RR: respiration rate; SBP: systolic blood pressure; SpO2: peripheral capillary oxygen saturation; WBC: white blood cell.
Association between HRR and 28-day all-cause mortality
Figure 2 presents the Kaplan–Meier survival probability curves for 28-day all-cause mortality across HRR tertiles. The survival analysis demonstrated that patients in the lower HRR tertiles experienced substantially elevated mortality rates (log-rank p < 0.0001).
Figure 2.
Kaplan–Meier survival curves for 28-day all-cause mortality.
The Cox proportional hazards regression analyses are summarized in Table 2. In the crude model, the HRs for 28-day all-cause mortality were 0.59 (95% CI: 0.55–0.64) for T2 and 0.36 (95% CI: 0.33–0.39) for T3, compared with the reference tertile. After adjustment for sociodemographic factors, including age, sex, ethnicity, and marital status, HRs remained statistically significant at 0.57 (95% CI: 0.53–0.61) for T2 and 0.39 (95% CI: 0.36–0.43) for T3, indicating the protective effect associated with higher HRR levels (Table 2; Model 2). Further adjustment for biochemical and physiological variables, including WBC count, RBC count, electrolyte levels, renal function indices, vital signs, and oxygenation parameters, did not materially alter the significant association between HRR and mortality outcomes. The adjusted HRs were 0.68 (95% CI: 0.63–0.73) for T2 and 0.58 (95% CI: 0.52–0.65) for T3, supporting an inverse relationship between higher HRR levels and mortality risk (Table 2; Model 3). In the fully adjusted model, which additionally adjusted for comorbidities such as hypertension, diabetes mellitus, myocardial infarction, malignant neoplasms, cerebrovascular disease, and COPD, the association between HRR and mortality remained significant. The final adjusted HRs were 0.68 (95% CI: 0.63–0.74) for T2 and 0.58 (95% CI: 0.52–0.65) for T3, consistently demonstrating lower mortality risk with higher HRR levels (Table 2; Model 4). Model 4 demonstrated good discriminative ability, with a C-index of 0.74, indicating satisfactory predictive performance for 28-day mortality. The global test yielded p value of 0.172, confirming that the proportional hazards assumption was satisfied for the overall model, thereby supporting the suitability of the Cox proportional hazards model for this analysis.
Table 2.
Cox proportional hazard ratios for 28-day all-cause mortality.
| Tertiles |
|||
|---|---|---|---|
| T1 (n = 9234) | T2 (n = 9220) | T3 (n = 9216) | |
| Model 1 | |||
| HR (95% CI) | Ref | 0.59 (0.55, 0.64) | 0.36 (0.33, 0.39) |
| P-value | p < 0.001 | p < 0.001 | |
| p for trend | p < 0.001 | ||
| Model 2 | |||
| HR (95% CI) | Ref | 0.57 (0.53, 0.61) | 0.39 (0.36, 0.43) |
| p-value | p < 0.001 | p < 0.001 | |
| p for trend | p < 0.001 | ||
| Model 3 | |||
| HR (95% CI) | Ref | 0.68 (0.63, 0.73) | 0.58 (0.52, 0.65) |
| p-value | p < 0.001 | p < 0.001 | |
| p for trend | p < 0.001 | ||
| Model 4 | |||
| HR (95% CI) | Ref | 0.68 (0.63, 0.74) | 0.58 (0.52, 0.65) |
| p-value | p < 0.001 | p < 0.001 | |
| p for trend | p < 0.001 | ||
Model 1: unadjusted; Model 2: adjusted for age, sex, race, and marital status; Model 3: adjusted for variables in Model 2 along with WBC, RBC, sodium, potassium, glucose, BUN, Cr, HR, SBP, DBP, RR, SpO2, and body temperature; Model 4: adjusted for variables in Model 3 along with hypertension, diabetes, myocardial infarction, malignant tumor, stroke, and COPD.
BUN: blood urea nitrogen; CI: confidence interval; COPD: chronic obstructive pulmonary disease; Cr: creatinine; DBP: diastolic blood pressure; HR: heart rate; RBC: red blood cell; Ref: reference; RR: respiration rate; SBP: systolic blood pressure; SpO2: peripheral capillary oxygen saturation; WBC: white blood cell.
Subgroup analysis
We conducted comprehensive subgroup analyses to evaluate the prognostic value of HRR for 28-day all-cause mortality across diverse patient subgroups stratified by sex, ethnicity, arterial hypertension, diabetes mellitus, myocardial infarction, and malignant neoplasms (Figure 3). HRR demonstrated consistent protective associations with mortality reduction across all examined strata: male (HR (95% CI): 0.82 (0.80–0.85)) and female (HR (95% CI): 0.84 (0.81–0.87)); White (HR (95% CI): 0.84 (0.81–0.86)), Black (HR (95% CI): 0.78 (0.72–0.85)), and Other ethnicities (HR (95% CI): 0.82 (0.78–0.86)); patients with hypertension (HR (95% CI): 0.86 (0.83–0.89)) and those without hypertension (HR (95% CI): 0.82 (0.79–0.84)); patients with diabetes (HR (95% CI): 0.84 (0.80–0.88)) and those without diabetes (HR (95% CI): 0.83 (0.81–0.86)); patients with a history of myocardial infarction (HR (95% CI): 0.80 (0.74–0.86)) and those without myocardial infarction (HR (95% CI): 0.83 (0.81–0.86)); and patients with malignant neoplasms (HR (95% CI): 0.79 (0.75–0.82)) and those without malignant neoplasms (HR (95% CI): 0.83 (0.81–0.86)). Furthermore, the absence of statistically significant interaction effects (all p-values >0.05) in subgroup analyses indicates that the prognostic performance of HRR remained robust and uniform across heterogeneous patient populations, supporting its applicability as a universal prognostic biomarker for mortality prediction.
Figure 3.
Subgroup analysis of the association between HRR and 28-day all-cause mortality. HRR: hemoglobin-to-red cell distribution width ratio.
Optimal cutoff value for clinical risk stratification
ROC curve analysis was performed to determine an optimal HRR threshold for clinical risk stratification (Figure 4). The analysis yielded an AUC of 0.657 (95% CI: 0.648–0.666), indicating acceptable discriminative ability of HRR as a single predictor of 28-day mortality. Based on the Youden index, the optimal cutoff value for HRR was identified as 6.53
Figure 4.
Receiver operating characteristic (ROC) curve of HRR for predicting 28-day all-cause mortality in patients with AKI. AKI: acute kidney injury; HRR: hemoglobin-to-red cell distribution width ratio.
PSM analysis
After 1:1 PSM, 1799 matched pairs were obtained. Baseline characteristics were well-balanced between the groups, with most covariates achieving an SMD <0.1 (Table S2). The 28-day mortality rate remained significantly lower in the high HRR group (T3: 10.6% vs. T1: 18.0%, p < 0.001). Kaplan–Meier analysis demonstrated significantly improved survival in the T3 group (log-rank p < 0.0001; Figure S1). In the matched cohort, Cox regression showed consistent protective associations across all models (fully adjusted HR: 0.60, 95% CI: 0.50–0.73, p < 0.001) (Table S3), corroborating our primary findings.
Discussion
This study provides the first comprehensive evaluation of the relationship between HRR and 28-day all-cause mortality among ICU patients with AKI, using multiple methodologies. By analyzing clinical data from 27,670 patients with AKI, we identified a strong association between HRR and the risk of 28-day all-cause mortality, which remained significant after adjustment for potential confounders. Kaplan–Meier survival analysis further demonstrated that patients in the lower HRR tertiles had significantly poorer outcomes in terms of 28-day all-cause mortality. The ROC analysis identified an optimal HRR cutoff value of 6.53 for risk stratification, with an AUC of 0.657. Although this reflects moderate discriminative ability as a single marker, it compared favorably with other commonly used biomarkers for AKI prognostication. Importantly, HRR offers distinct advantages over complex scoring systems. It requires no additional laboratory tests beyond routine complete blood count (CBC), incurs no additional cost, and can be calculated immediately at the bedside. The cutoff value of 6.53 offers a practical, actionable threshold for clinicians to identify patients who require intensified monitoring or intervention. Moreover, HRR demonstrated independent prognostic value even after adjustment for multiple clinical and laboratory variables (C-index: 0.74 in multivariable model), supporting its clinical utility as part of a comprehensive risk assessment strategy. Overall, HRR proved to be a significant predictor of prognosis for patients with AKI in critical care settings.
AKI poses a significant challenge in the ICU setting because of its high incidence and severe impact on patient outcomes. 20 This condition often results in a rapid decline in kidney function, leading to complications across multiple organ systems and elevating the mortality risk among ICU patients. 21 Despite advancements in technology and improvements in critical care protocols, the management of AKI remains complex, and outcomes are often unfavorable. Identifying high-risk factors for AKI is essential to improve prognosis. Traditional markers, including serum creatinine and BUN, may not provide a comprehensive assessment, particularly in ICU patients with complex clinical conditions. 22 Previous studies have shown that biomarkers such as serum ferritin, which is associated with with inflammation and disease severity, may improve risk stratification. 23 Elevated serum ferritin levels have been associated with worse outcomes, likely due to the inflammatory response and impaired iron metabolism in severe cases. 23 Our study underscores the importance of HRR as a predictive marker for 28-day all-cause mortality in ICU patients with AKI. We observed a strong association between lower HRR levels and higher mortality, which remained significant after adjustment for potential confounders. These findings support the robustness of HRR as a prognostic indicator and suggest that incorporating HRR into routine clinical assessments can significantly improve early identification of high-risk patients with AKI, thereby facilitating timely and targeted interventions. Furthermore, the etiologies of AKI vary across regions, ranging from infections and trauma in less developed areas to sepsis and nephrotoxic drug exposure in more developed settings. This complicates clinical management and underscores the need to identify better high-risk factors to predict AKI outcomes.24,25
HRR has emerged as a novel and informative biomarker that integrates prognostic information from hemoglobin and RDW. 9 By combining anemia-related parameters with markers of systemic inflammation, HRR provides a comprehensive view of a patient’s health status. 26 Substantial evidence has demonstrated the prognostic value of RDW in various cardiovascular diseases,27–29 sepsis, 30 and other critical illnesses. 31 Hemoglobin level, as an indicator of anemia, has also been consistently associated with outcomes in various medical conditions, including stroke 32 and heart disease. 33 Recent studies have highlighted the potential of HRR as a predictor for adverse outcomes across multiple diseases, including cardiovascular diseases, 34 malignancies, 35 and sepsis. 36 Lower HRR levels have been associated with increased mortality and poorer prognosis, supporting its potential role in clinical risk stratification. Despite these promising findings, the application of HRR in predicting outcomes among patients with AKI, particularly in the ICU setting, has not been fully investigated.
One of the primary advantages of HRR is its cost-effectiveness and accessibility, as it can be derived from routine CBC tests. This makes it suitable for broad clinical application without requiring additional resources. 34 However, its predictive performance and clinical utility in specific conditions such as AKI require further validation. Our study underscores the potential role of HRR as a prognostic marker for 28-day all-cause mortality in ICU patients with AKI. The robust association between lower HRR levels and higher mortality, which persisted after adjustment for confounders, supports its potential clinical value. These findings indicate the need for further studies to validate the role of HRR in AKI and to evaluate its integration into clinical practice to improve outcomes in ICU patients.
This study has several limitations that should be acknowledged. First, as a single-center retrospective cohort study based on the MIMIC-IV 3.0 database, the findings may not be generalizable to clinical settings in other countries or regions, necessitating further research for broader validation. Second, the study analyzed HRR only at the time of admission. Future studies should consider multiple time-point measurements of HRR to better evaluate its predictive value for AKI risk over time. Third, important biomarkers, including erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP), were excluded because of substantial missing data, and cystatin C, an established kidney injury marker, was not available in the database. Future studies should aim to incorporate these biomarkers to provide a more comprehensive evaluation. Fourth, the 14-year study period may have encompassed evolving clinical practices. However, the robustness of our findings in both primary and PSM analyses suggests minimal impact on our conclusions. Fifth, the data were derived from a single center in the United States with a predominantly Caucasian population (71.6%), which may limit generalizability to other ethnic groups and healthcare systems. Therefore, external validation in diverse populations is warranted. Finally, important biomarkers such as CRP and cystatin C were unavailable in the MIMIC-IV database. Nevertheless, we adjusted for established prognostic factors including renal function markers (creatinine and BUN), inflammatory indicators (WBC), and relevant clinical variables. Future studies incorporating these biomarkers would allow a more comprehensive assessment.
Conclusion
Our study demonstrated a robust inverse relationship between lower HRR levels and increased 28-day all-cause mortality among critically ill patients with AKI in the intensive care settings. The RDW ratio represents a cost-effective and clinically accessible prognostic biomarker for mortality risk stratification. ROC curve analysis identified an optimal HRR cutoff value of 6.53 (AUC: 0.657), providing a practical threshold for identifying high-risk patients. Incorporating HRR into clinical assessment may improve risk prediction in AKI management and support individualized monitoring and targeted therapeutic strategies for high-risk populations.
Supplemental Material
Supplemental material, sj-pdf-1-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Supplemental material, sj-xlsx-2-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Supplemental material, sj-xlsx-3-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Supplemental material, sj-xlsx-4-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Acknowledgments
The authors appreciate the researchers at the MIT Laboratory for Computational Physiology for publicly sharing of the MIMIC-IV clinical database.
Author contributions: The project was designed by Fei Xu and Man Luo. Material preparations were conducted by Fei Xu and Yangang Zhu. Data collection and analysis were performed by Yichi Yang, Fei Xu, and Yangang Zhu. The first draft of the manuscript was written by Yichi Yang, Yaxin Wang, and Fei Xu and was critically revised by Man Luo and Fei Xu. All the revisions were finished by Wei Li. All authors reviewed and approved the final manuscript.
Funding: Not applicable.
ORCID iD: Man Luo https://orcid.org/0009-0004-0777-7194
Data availability statement
All data were from the public database (https://physionet.org/content/mimiciv/3.0/); no ethical approval was needed.
Declaration of conflicting interests
The authors have no conflicts of interest to disclose.
Ethics approval and consent to participate
The collection of patient information and creation of the research resource was reviewed by the Institutional Review Board at the Beth Israel Deaconess Medical Center, who granted a waiver of informed consent and approved the data sharing initiative.
Supplemental material
Supplemental material for this article is available online.
References
- 1.Tsao CW, Aday AW, Almarzooq ZI, et al. Heart disease and stroke statistics-2022 update: a report from the American heart association. Circulation 2022; 145: e153–e639. [DOI] [PubMed] [Google Scholar]
- 2.Raimann JG, Riella MC, Levin NW. International Society of Nephrology's 0by25 initiative (zero preventable deaths from acute kidney injury by 2025): focus on diagnosis of acute kidney injury in low-income countries. Clin Kidney J 2018; 11: 12–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hoste EAJ, Kellum JA, Selby NM, et al. Global epidemiology and outcomes of acute kidney injury. Nat Rev Nephrol 2018; 14: 607–625. [DOI] [PubMed] [Google Scholar]
- 4.Fujarski M, Porschen C, Plagwitz L, et al. Prediction of acute kidney injury in the intensive care unit: preliminary findings in a European open access database. Stud Health Technol Inform 2022; 294: 139–140. [DOI] [PubMed] [Google Scholar]
- 5.GBD 2019 Diseases and Injuries Collaborators. Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 2020; 396: 1204–1222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Baeg SI, Jeon J, Kang D, et al. Impact of protocolized fluid management on electrolyte stability in patients undergoing continuous renal replacement therapy. Front Med (Lausanne) 2022; 9: 915072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Husain-Syed F, Takeuchi T, Neyra JA, et al. Acute kidney injury in neurocritical care. Crit Care 2023; 27: 341. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Prasad B, Urbanski M, Ferguson TW, et al. Early mortality on continuous renal replacement therapy (CRRT): the prairie CRRT study. Can J Kidney Health Dis 2016; 3: 36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yılmaz H, Yılmaz A, Demirağ G. Prognostic significance of hemoglobin-to-red cell distribution width ratio in patients with metastatic renal cancer. Future Oncol 2021; 17: 3853–3864. [DOI] [PubMed] [Google Scholar]
- 10.Danese E, Lippi G, Montagnana M. Red blood cell distribution width and cardiovascular diseases. J Thorac Dis 2015; 7: E402–E411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Yeşil A, Senateş E, Bayoğlu IV, et al. Red cell distribution width: a novel marker of activity in inflammatory bowel disease. Gut Liver 2011; 5: 460–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Tonelli M, Sacks F, Arnold M; for the Cholesterol and Recurrent Events (CARE) Trial Investigators et al. Relation between red blood cell distribution width and cardiovascular event rate in people with coronary disease. Circulation 2008; 117: 163–168. [DOI] [PubMed] [Google Scholar]
- 13.Wang J, Chen Z, Yang H, et al. Relationship between the hemoglobin-to-red cell distribution width ratio and all-cause mortality in septic patients with atrial fibrillation: based on propensity score matching method. J Cardiovasc Dev Dis 2022; 9: 400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Krongsut S, Srikaew S, Anusasnee N. Prognostic value of combining 24-hour ASPECTS and hemoglobin to red cell distribution width ratio to the THRIVE score in predicting in-hospital mortality among ischemic stroke patients treated with intravenous thrombolysis. PLOS One 2024; 19: e0304765. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Sun P, Zhang F, Chen C, et al. The ratio of hemoglobin to red cell distribution width as a novel prognostic parameter in esophageal squamous cell carcinoma: a retrospective study from southern China. Oncotarget 2016; 7: 42650–42660. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wan H, Yang Y, Zhu J, et al. The relationship between elevated red cell distribution width and long-term outcomes among patients with atrial fibrillation. Clin Biochem 2015; 48: 762–767. [DOI] [PubMed] [Google Scholar]
- 17.Johnson A, Bulgarelli L, Pollard T, et al. MIMIC-IV (version 3.0). PhysioNet 2024. [Google Scholar]
- 18.Johnson AEW, Bulgarelli L, Shen L, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data 2023; 10: 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Goldberger A, Amaral L, Glass L, et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 2000; 101: e215–e220. [DOI] [PubMed] [Google Scholar]
- 20.Bellomo R, Kellum JA, Ronco C. Acute kidney injury. Lancet 2012; 380: 756–766. [DOI] [PubMed] [Google Scholar]
- 21.Levey AS, James MT. Acute kidney injury. Ann Intern Med 2017; 167: ITC66–ITC80. [DOI] [PubMed] [Google Scholar]
- 22.Sileanu FE, Murugan R, Lucko N, et al. AKI in low-risk versus high-risk patients in intensive care. Clin J Am Soc Nephrol 2015; 10: 187–196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Góes MA, Iizuka IJ, Quinto BM, et al. Serum soluble-Fas, inflammation, and anemia in acute kidney injury. Artif Organs 2018; 42: E283–E289. [DOI] [PubMed] [Google Scholar]
- 24.McClave SA, Taylor BE, Martindale RG; American Society for Parenteral and Enteral Nutrition et al. Guidelines for the provision and assessment of nutrition support therapy in the adult critically ill patient: Society of Critical Care Medicine (SCCM) and American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.). JPEN J Parenter Enteral Nutr 2016; 40: 159–211. [DOI] [PubMed] [Google Scholar]
- 25.Singer P, Blaser AR, Berger MM, et al. ESPEN practical and partially revised guideline: clinical nutrition in the intensive care unit. Clin Nutr 2023; 42: 1671–1689. [DOI] [PubMed] [Google Scholar]
- 26.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: 628–632. [DOI] [PubMed] [Google Scholar]
- 27.Liang L, Huang L, Zhao X, et al. Prognostic value of RDW alone and in combination with NT-proBNP in patients with heart failure. Clin Cardiol 2022; 45: 802–813. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Guan YZ, Yin RX, Zheng PF, et al. Association of RDW, NLR, and PLR with atrial fibrillation in critical care patients: a retrospective study based on propensity score matching. Dis Markers 2022; 2022: 2694499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lappegård J, Ellingsen TS, Vik A, et al. Red cell distribution width and carotid atherosclerosis progression. The Tromsø study. Thromb Haemost 2015; 113: 649–654. [DOI] [PubMed] [Google Scholar]
- 30.Huang Y, Jiang S, Li W, et al. Establishment and effectiveness evaluation of a scoring system-RAAS (RDW, AGE, APACHE II, SOFA) for sepsis by a retrospective analysis. J Inflamm Res 2022; 15: 465–474. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Li G, Jia P, Zhao J, et al. Usefulness of RBC distribution width and C-reactive protein to predict mortality in pediatric non-cardiac critical illness. Am J Emerg Med 2019; 37: 2143–2150. [DOI] [PubMed] [Google Scholar]
- 32.Chang JY, Lee JS, Kim BJ, et al. Influence of hemoglobin concentration on stroke recurrence and composite vascular events. Stroke 2020; 51: 1309–1312. [DOI] [PubMed] [Google Scholar]
- 33.Filippatos G, Ponikowski P, Farmakis D; AFFIRM-AHF Investigators et al. Association between hemoglobin levels and efficacy of intravenous ferric carboxymaltose in patients with acute heart failure and iron deficiency: an AFFIRM-AHF subgroup analysis. Circulation 2023; 147: 1640–1653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Qu J, Zhou T, Xue M, et al. Correlation analysis of hemoglobin-to-red blood cell distribution width ratio and frailty in elderly patients with coronary heart disease. Front Cardiovasc Med 2021; 8: 728800. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Chen JL, Wu JN, Lv XD, et al. The value of red blood cell distribution width, neutrophil-to-lymphocyte ratio, and hemoglobin-to-red blood cell distribution width ratio in the progression of non-small cell lung cancer. PLOS One 2020; 15: e0237947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Huang X, Yuan S, Ling Y, et al. The hemoglobin-to-red cell distribution width ratio to predict all-cause mortality in patients with sepsis-associated encephalopathy in the MIMIC-IV database. Int J Clin Pract 2022; 2022: 7141216. [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
Supplemental material, sj-pdf-1-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Supplemental material, sj-xlsx-2-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Supplemental material, sj-xlsx-3-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Supplemental material, sj-xlsx-4-imr-10.1177_03000605261427144 for Association between hemoglobin-to-red blood cell distribution width ratio and 28-day all-cause mortality in patients with acute kidney injury: A cross-sectional study by Fei Xu, Yichi Yang, Yaxin Wang, Yangang Zhu, Wei Li and Man Luo in Journal of International Medical Research
Data Availability Statement
All data were from the public database (https://physionet.org/content/mimiciv/3.0/); no ethical approval was needed.