Association between Red Blood Cell Distribution Width and In-Hospital Mortality among Congestive Heart Failure Patients with Diabetes among Patients in the Intensive Care Unit: A Retrospective Cohort Study

Background Elevated red blood cell distribution width (RDW) levels are strongly associated with an increased risk of mortality in patients with congestive heart failure (CHF). Additionally, heart failure has been closely linked to diabetes. Nevertheless, the relationship between RDW and in-hospital mortality in the intensive care unit (ICU) among patients with both congestive heart failure (CHF) and diabetes mellitus (DM) remains uncertain. Methods This retrospective study utilized data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, a comprehensive critical care repository. RDW was assessed as both continuous and categorical variables. The primary outcome of the study was in-hospital mortality at the time of hospital discharge. We examined the association between RDW on ICU admission and in-hospital mortality using multivariable logistic regression models, restricted cubic spline analysis, and subgroup analysis. Results The cohort consisted of 7,063 patients with both DM and CHF (3,135 females and 3,928 males). After adjusting for potential confounders, we found an association between a 9% increase in mortality rate and a 1 g/L increase in RDW level (OR = 1.09; 95% CI, 1.05∼1.13), which was associated with 11 and 58% increases in mortality rates in Q2 (OR = 1.11, 95% CI: 0.87∼1.43) and Q3 (OR = 1.58, 95% CI: 1.22∼2.04), respectively, compared with that in Q1. Moreover, we observed a significant linear association between RDW and in-hospital mortality, along with strong stratified analyses to support the findings. Conclusions Our findings establish a positive association between RDW and in-hospital mortality in patients with DM and CHF.


Introduction
Diabetes mellitus (DM) is a metabolic disorder characterized by chronic hyperglycemia and impaired metabolism of carbohydrates, lipids, and proteins due to defects in insulin secretion, insulin action, or both [1].As of 2019, diabetes globally afected approximately 9.3% of the population (463 million individuals), and these numbers are anticipated to rise to 10.2% (578 million) by 2030 and 10.9% (700 million) by 2045 [2].Te correlation between diabetes and heart failure is well-documented, as diabetes is associated with elevated rates of heart failure [3].Conversely, heart failure contributes to a higher likelihood of new-onset diabetes [4,5].In individuals with heart failure, the prevalence of diabetes and prediabetic dysglycemia is substantial, heightening the risk of cardiovascular death and heart failure-related hospitalization for those with coexisting diabetes or prediabetes [6].
Despite the growing recognition of heart failure as a frequent and potentially fatal complication of diabetes, a defnitive and straightforward index to predict mortality among patients with both congestive heart failure (CHF) and diabetes remains lacking.
Several studies have highlighted the association between the progression of diabetes-related heart failure and the escalation of oxidative stress and infammation within the body [7,8].Moreover, research indicates that red blood cell distribution width (RDW), a commonly used hematological indicator, correlates with levels of oxidative stress and infammation, potentially signifying its role as a biomarker [9][10][11][12].RDW, a quantitative measure of erythrocyte size variability primarily used in diagnosing anemia, has emerged as a potential marker [13].Clinical investigations have consistently suggested that elevated RDW levels may serve as biomarkers for various cardiovascular and cerebrovascular conditions, such as heart failure (HF), coronary artery disease (CAD), brain death, and pulmonary arterial hypertension [14][15][16].Furthermore, elevated RDW levels upon admission have exhibited signifcant associations with the occurrence and prognosis of complications in chronic heart failure (CHF) [17].Tese elevated RDW levels are strongly linked to adverse outcomes in patients with CHF and CAD, regardless of other hematological variables [18][19][20].
However, despite the established link between diabetes and heart failure, limited research has explored the relationship between RDW levels in individuals with both diabetes mellitus (DM) and congestive heart failure (CHF), and their clinical outcomes.Building upon previous fndings, this study hypothesizes that higher RDW levels are associated with in-hospital mortality within a substantial cohort of American adults concurrently admitted to the intensive care unit with congestive heart failure and diabetes.Terefore, this research aims to investigate the connection between RDW and in-hospital mortality within this cohort.[21,22].Te database spans 2008-2019, comprising diverse information such as demographics, vital signs, lab results, fuid balance, vital status, ICD-9 codes, hourly physiologic data, and radiologic evaluations [23].To access the MIM-IC-IV database, we completed a training course on the National Institutes of Health (NIH) website and passed the "Protecting Human Research Participants" exam (author certifcation number: 11639604).Additionally, the database received approval from the institutional review boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center (Boston, MA, United States) [24].Our study also obtained approval from the Beth Israel Deaconess Medical Center Institutional Review Board, which waived the need for patient consent due to the retrospective and deidentifed nature of the data [25].Tis study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement to ensure comprehensive and transparent reporting practices.

Study Population.
Te study enrolled 7,063 patients diagnosed with congestive heart failure (CHF) who subsequently developed diabetes (DM) during their ICU admission.Diagnosis adhered to the International Classifcation of Diseases (ICD) standards, which have demonstrated efcacy across numerous prior investigations [26,27].Diagnostic data were extracted from the "diag-noses_icd" and "d_icd_diagnoses" tables in the database.ICD codes for CHF and diabetes are detailed in Table S1 of the Supporting Information.Exclusion criteria encompassed individuals under 18, those lacking outcome data, and those devoid of diabetes.Figure 1 delineates the selection process.

Expose and Outcome.
Data retrieval from the Post-greSQL database (version 13) involved structured query language (SQL).While multiple RDW measurements were available for selected patients, only the initial measurement upon hospital admission was retained and treated as a continuous variable.Patients were subsequently stratifed into three groups based on RDW tertiles on the frst day of ICU admission: Q1 (≤14.6%),Q2 (14.6%-16.4%),and Q3 (>16.4%).Te primary endpoint of this study was inhospital mortality, determined by the patient's survival status upon discharge.Tis methodology has been validated through multiple corroborative investigations [28,29].

Data Retrieval.
Using PostgreSQL (version 9.6) and the Structured Query Language (SQL), we conducted data extraction from the MIMIC-IV database.Data was collected within 24 hours of ICU admission and encompassed several data points, including demographic variables (sex, age, race), comorbidities (chronic obstructive pulmonary disease (COPD), acute myocardial infarction (AMI), colonic melanosis (MC), hepatic failure (HepF)), medical procedures (vent, intubated), medication administration (norepinephrine, dopamine, epinephrine, phenylephrine, and vasopressin), basic vital signs (temperature, respiratory rate, heart rate, systolic blood pressure (SBP)), blood biochemical indicators (anion gap (AG), urea nitrogen (BUN), chloride, creatinine, hemoglobin (Hb), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), mean corpuscular volume (MCV), platelet, potassium, sodium, red blood cell (RBC), white blood cell (WBC)), as well as the severity of illness at ICU admission, assessed by the Sequential Organ Failure Assessment (SOFA) and simplifed acute physiology score (SAPS) II.Demographic characteristics and vital signs within the initial 24 hours of admission were recorded, with the initial measurement upon admission serving as the laboratory examination indicators.To investigate the association between RDW and hospital mortality, multivariate logistic regression analyses were performed, calculating odds ratios (ORs) and 95% confdence intervals (CIs) to evaluate the efect.RDW was treated as both a continuous and categorical variable (tertiles).Five adjustment models were utilized, gradually including additional covariates.Model 1 remained unadjusted, while Model 2 was adjusted for demographic variables (sex, age, and race).Model 3 included demographic variables and concomitant diseases (COPD, AMI, MC, and HepF).Model 4 was adjusted for demographic variables, complicating diseases, medical procedures, medication situation, basic vital signs, and blood biochemical indicators.Finally, Model 5 included demographic variables, complicating diseases, medical procedures, medication situation, basic vital signs, blood biochemical indicators, APSIII, and SOFA.
Te trend of RDW was analyzed using the chi-square trend test (Cochran-Armitage trend test).In the second step, a restricted cubic curve spline analysis and smooth curve ftting (penalized spline method) were employed to estimate the relationship between RDW and hospital mortality.Additionally, stratifed linear regression models and likelihood ratio tests were applied to identify modifcations and interactions in subgroups based on various factors.
All statistical analyses were conducted using the R software package, version 4.1.1(R Foundation for Statistical Computing, Vienna, Austria), and Free Statistics software, version 1.7.A signifcance level of P < 0.05 (two-sided) was considered statistically signifcant.Te reporting of this cross-sectional study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.

Baseline Characteristics of Selected Participants.
Table 1 presents the baseline characteristics of 7063 patients with concurrent diabetes and congestive heart failure (CHF), categorized into three groups based on their red cell distribution width (RDW) levels: Q1 (≤14.6%),Q2 (14.6%-16.4%),and Q3 (>16.4%).Patients with higher RDW levels exhibited the following trends: higher proportion of female participants, increased usage of vasopressin and no epinephrine, reduced rates of intubation, higher incidences of chronic obstructive pulmonary disease (COPD), acute myocardial infarction (AMI), hepatic failure (HepF), acute gastroenteritis (AG), blood urea nitrogen (BUN), serum creatinine, Sequential Organ Failure Assessment (SOFA) score, Acute Physiology and Chronic Health Evaluation (APACHE) III score, and mortality rate.In contrast, this group demonstrated lower proportions of white individuals, lower body temperature, lower systolic blood pressure (SBP), lower hemoglobin (Hb), lower mean corpuscular hemoglobin (MCH), lower mean corpuscular hemoglobin concentration (MCHC), lower mean corpuscular volume (MCV), lower red blood cell (RBC) count, and lower blood chloride levels.

Association between RDW and In-Hospital Mortality.
Table 2 presents the relationship between red cell distribution width (RDW) and in-hospital mortality in patients with congestive heart failure (CHF) and diabetes.Odds ratios (ORs) and their corresponding 95% confdence intervals (CIs) were calculated.In our study, we reported a 9% increase in in-hospital mortality risk for each unit increase in RDW, after adjustment for confounders (OR � 1.09; 95% CI, 1.05∼1.13;P < 0.001).  2 presents the distributions of variables (grey histograms), the smoothing curve ft (solid grey curve) between the variables, and the 95% confdence interval of the curve ft (grey zone).Upon adjusting for all covariates in the restricted cubic curve spline analysis, a linear relationship was observed between the continuous variable RDW and in-hospital mortality, with a nonlinear P value of 0.447 (Figure 2).

Subgroup Analysis.
To explore potential modifcations of this association by confounding factors, subgroup analyses were conducted, evaluating the efects of various stratifcation variables such as age, sex, norepinephrine, dopamine, epinephrine, phenylephrine, vasopressin, vent, chronic obstructive pulmonary disease, acute myocardial infarction, and hepatic failure on the relationship between RDW and mortality.Te results, along with their corresponding interactions, are provided in the Figure 3. Notably, no statistically signifcant associations were observed in any of the subgroups (P > 0.05).

Discussion
In this retrospective cohort analysis, a signifcant association between red cell distribution width (RDW) and mortality risk was discerned in patients diagnosed with both congestive heart failure and diabetes.Tis observed correlation retained its statistical signifcance even after adjusting for potential confounding variables, indicating an independent linear connection between RDW levels and mortality within this patient cohort.Furthermore, consistent stratifed assessments reinforced the robustness of the relationship between RDW and in-hospital mortality.Numerous prior inquiries have explored the impact of RDW on the risk of in-hospital mortality.Consistent with our outcomes, Pascual-Figal et al. [30] documented that an elevated RDW level at discharge was linked to unfavorable long-term outcomes, irrespective of anemia presence and hemoglobin concentrations.Similarly, Muhlestein et al. [31] suggested that higher RDW levels upon initial hospitalization for heart failure (HF) were associated with 30-day allcause readmissions, prolonged hospital stays, and 30-day mortality, indicating the potential utility of early-stage RDW levels for tailored intervention and prognosis enhancement.
A recently published meta-analysis [32] established a substantial connection between elevated baseline RDW and heightened mortality in ischemic stroke, encompassing extended-term mortality (one year or more).Furthermore, other investigations have also identifed a noteworthy correlation between RDW and mortality linked to sepsis [33][34][35][36][37]. Importantly, preceding studies have predominantly encompassed patients with diferent medical conditions, setting our study apart as the sole one simultaneously exploring congestive heart failure (CHF) and diabetes mellitus (DM).
One important fnding of this study is the strong positive linear relationship between RDW (red cell distribution width) and in-hospital mortality among American patients with congestive heart failure and diabetes.Subsequent subgroup analysis was performed to validate the consistency of the primary fndings.Te outcomes validate that an escalation in RDW is tied to an elevated risk of in-hospital mortality.Tese results underscore the signifcance of integrating RDW measurements into clinical protocols.Furthermore, RDW testing proves to be economical, swift, and easily attainable, making it feasible to combine with other predictive markers to achieve more precise risk stratifcation and enable early intervention for these patients [38][39][40].Given the simplicity, afordability, and widespread accessibility of RDW assessment, these conclusions bear considerable clinical ramifcations for prognostic evaluation in acute HF patients.
Plausible rationales for the escalated mortality risk associated with red cell distribution width (RDW) are discernible, notwithstanding an incomplete comprehension of the precise underlying mechanism.One potential mechanism involves the induction of proinfammatory cytokines, recognized to be linked with heart failure (HF).Te elevation of interleukin-1β, TNFα, and interleukin 6 could impede erythropoietin-driven erythrocyte maturation, leading to an elevation in RDW [41].Furthermore, numerous infammatory markers connected to heart failure, such as erythrocyte sedimentation rate, high-sensitivity C-reactive protein levels, and white blood cell count, display a robust correlation with RDW [42] An additional factor contributing to the heightened mortality risk may stem from prevalent erythropoiesis irregularities.It is broadly acknowledged that individuals with HF are more predisposed to developing anemia, a condition associated with an    [43].Suggested causes of anemia in HF encompass chronic ailments, disordered iron metabolism, renal dysfunction, and hemodilution.Moreover, discernible fuctuations in circulating erythropoietin concentrations are observed among anemic HF patients, particularly those with renal impairment.Te notion of bone marrow resistance to erythropoietin's efects has been posited as a plausible mechanism for anemia and heightened mortality in selected patients [44] Considering the pivotal role of erythropoietin in averting apoptosis of erythrocyte progenitor cells and invigorating their proliferation, maturation, and terminal diferentiation [45], the raised erythropoietin levels may ofer an explicable account for elevated RDW levels in HF patients with a more adverse prognosis.To fathom the underlying biology of the correlation between elevated RDW and unfavorable outcomes, as well as the conceivable link between RDW and HF-related anemia, future research pursuits are warranted.
Our study ofers several notable strengths.Firstly, as far as we know, this is the frst study investigating the association between RDW and in-hospital mortality in patients with congestive heart failure (CHF) and diabetes.Te fndings of our study reveal a linear association between RDW and in-hospital mortality, providing a solid theoretical basis for establishing targeted strategies for RDW control in these patients.Our data from the MIMIC dataset further suggest a potential broad clinical applicability.To adjust for the infuence of possible confounding factors, we conducted multiple logistic regression analysis and conducted relevant subgroup analyses.
Despite these strengths, we must acknowledge the limitations of our study.Its retrospective nature may introduce inherent biases, despite our eforts to adjust for relevant variables to ensure result accuracy.While RDW was chosen as the parameter of interest due to its ease of measurement and clinical convenience, other relevant factors may exist.Additionally, the severity of heart failure and diabetes was not investigated due to limitations of the public database; however, we intend to address this in future research using our own database.Furthermore, our study only considered the frst RDW measurement after ICU admission, without tracking its dynamics over time.Nevertheless, this initial measurement may better refect the RDW at the beginning of hospitalization.It is important to exercise caution when generalizing our fndings to other nations or ICU institutions, as the research was conducted in a single ICU facility in the USA.However, the substantial and representative sample size enhances the reliability of our fndings.For future validation and broader applicability, we recommend conducting multicenter prospective studies.

Conclusions
Te data from this population-based observational study revealed a conspicuous probability of linear association between RDW and in-hospital mortality in patients with diabetes mellitus (DM) and congestive heart failure (CHF).Tis study derives its clinical signifcance from analyzing data from MIMIC to assess the relationship between RDW and in-hospital mortality.However, future research should clarify these results through prospective study designs in similar populations.Critical Care Research and Practice

Figure 3 :
Figure3: Stratifed analyses of the association between RDW with in-hospital mortality rate.Note.Te P value for interaction represents the likelihood of interaction between the RDW with intrahospital mortality rate.OR, odd ratio; CI, confdence interval.

8
Te study employed data from the MIMIC-IV database, an extensive open-access repository collaboratively published by the Computational Physiology Laboratory of Massachusetts Institute of Technology, Beth Israel Deacon Medical Center, and Philips Healthcare 2.1.Data Source.

Table 2
3.3.Dose-ResponseRelationships.Tis study employed a logistic regression model incorporating a cubic spline function to investigate the association between RDW and in-

Table 2 :
Multivariable logistic regression to assess the association of RDW with in-hospital mortality rate.