Elevated red cell distribution width upon ICU admission independently predicts mortality in young patients with sepsis-associated encephalopathy: A propensity score-matched retrospective cohort study using MIMIC-IV database

Keywords: Sepsis-associated encephalopathy, 30-day mortality in intensive care unit, Red cell distribution width, Predictor, Young patients, MIMIC-IV database

Sepsis-associated encephalopathy (SAE) represents a major medical complication. The occurrence rate ranges between 8% and over 70%, with variations based on how severe the sepsis condition is in patients requiring critical care [1]. The clinical manifestation of SAE is an acute disorder of consciousness with delirium, hallucinations and agitation [2]. Sepsis is often accompanied by inflammation and coagulation problems. Consequently, systemic inflammation induces SAE by activating endothelial cells and microglial cells, leading to increased blood-brain barrier penetrability and subsequent damage to brain tissue [1, 3, 4]. SAE has been demonstrated to be closely associated with increased short-term mortality and with impairment of cognitive function, anxiety and stress-related disorders; therefore, accurate and convenient assessment of outcomes is particularly crucial [2, 5]. For the clinical front line, a biomarker that is both simple and reliable and can be repeatedly measured at the bedside can help to identify high-risk patients early and guide subsequent treatment decisions. Most existing studies have focused on elderly patients with SAE, whereas younger patients, whose physiological reserve is greater and who are often regarded as a "low-risk population", are instead easily overlooked in research and in clinical alertness [6, 7]. In this population, once SAE occurs, the 30-day mortality rate can still be as high as 9% (data from the present study), and this figure in itself suggests that it is necessary to establish as soon as possible a dedicated early warning tool for young patients.

Red cell distribution width (RDW) is used to quantify the variation in red blood cell (RBC) volume; it reflects the balance between production and clearance of RBCs and can also serve as a simple indicator of acute and chronic systemic inflammatory burden [8]. When RBC homoeostasis is severely disrupted, RDW is usually increased, a phenomenon that is very common in patients with severe infection and can also be seen in various metabolic disorders such as inflammatory diseases, malnutrition and lipid metabolic disturbances [9-11]. An increasing number of studies have found that among populations at intensive care unit (ICU), RDW levels are closely related to patients’ survival outcomes and can be used to assess the risk of death in various diseases such as sepsis, diabetes, and hepatic and renal failure [9, 10, 12-15]. It is noteworthy that RDW has also gradually shown potential to reflect the prognosis of encephalopathy and brain injury; the studies by Peng et al. and Yang et al. support the view that RDW can serve as a predictive factor for septic encephalopathy [16, 17]. Additionally, a recent study has proposed a correlation between heightened RDW levels in individuals with traumatic brain injury and increased mortality rates, as well as unfavorable neurological outcomes [18]. Kenangil et al. discovered that Parkinson’s disease patients with neuroinflammation as a pathological feature had significantly higher RDW than healthy subjects [19].

While these initial findings propose a plausible association between RDW and brain disorders, further extensive investigations are warranted to authenticate and expand upon these preliminary observations. The predictive value of initial RDW measurements at ICU admission for SAE patient outcomes remains unclear. The goal is to find out how initial RDW relates to ICU prognosis in young patients with SAE, which could facilitate risk stratification. 

2.1 Data sources

The complete collection of patient health states and diseases was constructed utilizing information derived from the openly available Medical Information Mart for Intensive Care, version 4 (MIMIC-IV) database, which contains the Beth Israel Deaconess Medical Center's medical records from 2008 to 2019 [20, 21]. The MIMIC-IV database, a comprehensive healthcare repository featuring anonymized, high-quality clinical data, includes approximately 10,000 recorded ICU stays. The author, JH, was permitted to retrieve and analyze database information. The utilization of these data was endorsed by Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center’s Institutional Ethics Board, who also waived informed consent requirements (credential ID: 55826703).

2.2 Cohort selection

In this retrospective study, we examined SAE patients younger than 65 using MIMIC-IV database entries. SAE was defined by the occurrence of sepsis plus either reduced consciousness (Glasgow Coma Scale scores <15) or delirium symptoms. The identification of sepsis followed Sepsis-3 criteria, requiring suspected infection with an increase in sequential organ failure assessment (SOFA) score to 2 or above. Patient selection was conducted according to the following eligibility standards: (1) first entry into the ICU; (2) length of ICU admission greater than 24 hours. Patients were not eligible if they met any of the following exclusion criteria: (1) primary cerebral trauma; (2) pre-existing mental health conditions or neurologic disorders; (3) development of metabolic, hepatic, or hypertensive encephalopathy, or any hepatic/renal conditions affecting mental status; (4) presence of severe electrolyte or glucose abnormalities, specifically sodium levels below 120 mmol/L, blood glucose above 180 mg/dL, or below 54 mg/dL; (5) documented delirium related to dementia, alcohol withdrawal, or medication use; (6) age ≥65 years; (7) missing RDW data. The process diagram is illustrated in Supplementary Figure 1.

2.3 Data exaction

The original data was obtained from the MIMIC-IV dataset through Structured Query Language executed using Navicat Premium 15.0.12. Data integration was performed using Stata MP 17. The gathered data included: (1) population-related and hospitalization statistics: patient's age, sex, racial background, hospital length of stay (LOS), and LOS in ICU; (2) physical signs and measurements: heart rate, systolic blood pressure (SBP), diastolic blood pressure (DBP), mean blood pressure (MBP), respiratory rate, and temperature; (3) existing health conditions: myocardial infarction, congestive heart failure, chronic pulmonary disease, renal disease, liver disease, diabetes, and peripheral vascular disease; (4) patient assessment metrics: SOFA score, logistic organ dysfunction system score , and simplified acute physiology score II; (5) treatment procedures: use of vasoactive drugs and invasive mechanical ventilation; (6) laboratory tests: white blood cell count, hemoglobin concentration, mean corpuscular volume, platelet count, RBC count, anion gap, bicarbonate levels, electrolyte levels including sodium, potassium, calcium, and chloride, renal function indices including blood urea nitrogen (BUN) and creatinine, glucose, and blood clotting markers including international normalized ratio (INR), prothrombin time (PT), activated partial thromboplastin time (APTT), and RDW [20-22]. It was within the first 24 hours post-ICU admission that these baseline demographic data and assay findings were obtained. For parameters with multiple measurements within this initial 24-hour window, average values were computed, specifically for heart rate, SBP, DBP, MBP, respiratory rate, and body temperature.

2.4 Outcomes

In our research, the main outcome measure was mortality within 30 days following ICU admission. Secondary endpoints encompassed overall in-hospital death rate, duration of hospitalization, and time spent in intensive care.

2.5 Management of missing data

To reduce potential bias, we excluded from our examination any variables containing over 15% missing values. The small number of remaining missing data points was imputed using mean values.

2.6 Statistical analysis

According to the distribution characteristics of the data, continuous variables were expressed as mean ± standard deviation or median (interquartile range), and comparisons between groups were performed using the Mann–Whitney test or Student's test; categorical variables were expressed as percentages, and Fisher's exact test was used to assess differences. We plotted receiver operating characteristic (ROC) curves to evaluate the ability of RDW to predict 30-day mortality risk in the included patients, and, by means of a generalized additive model with restricted cubic splines, explored whether there was a non-linear association between continuous RDW and the primary outcome. The optimal cut-off value of RDW was determined on the basis of the ROC curve, after first including RDW as a continuous variable in the analysis; subsequently, survival curves were plotted by the Kaplan–Meier method, and survival differences between groups were compared using the log-rank test.

To minimize selection bias and balance baseline characteristics as far as possible, we conducted a one-to-one nearest-neighbour propensity score matching (PSM) analysis with a caliper width of 0.02. Each patient's propensity score was calculated by logistic regression, in which the baseline variables listed in Table 1 were included. After completing PSM, we compared the standardized mean differences between the two groups, taking 0.15 as the upper limit for acceptable balance.

In order to further evaluate the relationship between RDW and clinical outcomes, we constructed univariable and multivariable Cox regression models respectively. Variables with a P value less than 0.1 in the univariable analysis after PSM were included in the multivariable model to control potential confounders. After matching was completed, we tested the proportional hazards assumption of the Cox model to confirm that there were no time-varying covariates in the adjusted model. All statistical analyses were performed in the R 4.2.0 environment, and a difference was considered statistically significant when P<0.05.

3.1 Clinical characteristics

A total of 1,594 subjects were included in this study, with a mean age of 49.8±12.1 years, of whom 63.0% were male. Within 30 days after ICU admission, a total of 144 patients (9.0%) died, indicating that even in a relatively young population, short-term mortality risk cannot be ignored. As shown in Supplementary Figure 2, there were obvious differences in RDW levels between groups, and the RDW of patients who died was significantly higher than that of survivors (17.5±3.1 vs 14.9±2.4, P<0.001).

Based on ROC curve analysis, the optimal RDW cut-off value was 14.65, with a corresponding area under the curve (AUC) of 0.760 (95% confidence interval [CI]: 0.720-0.800) (Supplementary Figure 3), according to which the study population was divided into a low-RDW group (n=859) and a high-RDW group (n=735). Comparison of the clinical characteristics of young SAE patients (Table 1) showed that, compared with the low-RDW group, patients in the high-RDW group were older, had comorbidities more frequently (except for myocardial infarction and peripheral vascular disease), had higher heart rates and higher disease severity scores, while their blood pressure, body temperature and SpO2 were relatively lower. Furthermore, the elevated RDW category showed higher anion gap, BUN, creatinine, INR, PT, and APTT values (all P<0.05). In contrast, mean corpuscular volume, calcium, and potassium showed no significant difference between groups, while other parameters were significantly lower in the high-RDW group (all P<0.05).

3.2 PSM procedure

We conducted a 1:1 PSM analysis to equalize covariate distribution between subjects categorized by RDW thresholds, resulting in 352 matched pairs. Table 1 presents the clinical features of young SAE patients post-matching. Both groups showed comparable distributions of demographics, comorbidities, signs, scores, and laboratory values. The matching quality was evaluated through standardized mean difference calculations pre- and post-PSM (Supplementary Figure 4).

3.3 RDW levels and primary outcomes

The mortality rate at 30 days showed a significant elevation among matched subjects with elevated RDW values (9.9% compared to 3.4%, P<0.001) (Table 2). Increased RDW de-monstrates an independent correlation with elevated 30-day mortality among young ICU patients experiencing SAE (Supplementary Tables 1 and 2).

Through multivariable Cox proportional hazards analysis evaluating RDW as a continuous parameter, this biomarker demonstrated significant predictive capacity for 30-day ICU mortality in the matched cohort (HR, 1.3 [95% CI, 1.2-1.4], P<0.001). In comparison to individuals with low RDW levels (≤14.65), young ICU patients diagnosed with SAE demonstrated an increased hazard ratio (95% CI) of 2.7 (1.4-5.3) for overall mortality. Other parameters emerged as independent predictors including patient age, liver disease, and body temperature (Supplementary Tables 1 and 2). As a continuous variable, RDW maintained status as an independent predictor of elevated hospital mortality risk (HR, 1.2 [95% CI, 1.1-1.4], P<0.001) (Table 3). Following sequential adjustments for demographic factors, comorbid conditions, vital signs, clinical scores, therapeutic measures and pharmacological treatments, and laboratory indicators, the observed correlations in matched patients maintained statistical significance, as demonstrated in Table 3. Nevertheless, upon implementing adjustments in Model V, the association between RDW and hospital mortality risk lost statistical significance (all P>0.05).

Survival analysis indicated significantly poorer 30-day ICU survival in young SAE patients with RDW>14.64% (log-rank test, P=0.002; Figure 1).

It was through restricted cubic spline modeling that the nonlinear correlation between RDW and 30-day ICU mortality was shown, where incremental RDW elevations predicted worsening survival probabilities (Likelihood ratio test, P<0.001) (Figure 2).

This study shows that among young patients with SAE, an elevated RDW level (>14.65%) is an independent predictor of 30-day ICU mortality risk, and this association remains stable even after full adjustment for multiple covariates. It is worth emphasizing that we focused on a group that is often overlooked yet very crucial, namely young SAE patients aged 18-65 years with relatively good physiological reserve, whereas previous studies on SAE prognosis and RDW have mostly been based on elderly or mixed-age cohorts. Young patients usually have fewer comorbidities and stronger organ reserve, and the driving factors of their adverse outcomes may be essentially different from those in elderly populations; even so, this study still observed a 30-day mortality rate close to 9%, and RDW was consistently a robust independent predictor, a result that strongly challenges the ingrained notion that "being young equals low risk". At the same time, in young patients, an elevated RDW is more likely to reflect the acute blow to RBC homoeostasis caused by the current acute infectious event such as an intense inflammatory storm or microcirculatory dysfunction, rather than being driven mainly by long-term chronic diseases or confounding factors such as malnutrition; this relationship still exists after adjustment for multidimensional covariates, including comorbidities. Consequently, the RDW threshold (14.65%) and its predictive efficacy (AUC=0.760) identified in this study may be more specific to young SAE patients, and its clinical early-warning value in this population may have been underestimated in previous research.

Previously, studies documented RDW's role in detecting anemia [22, 23]. More recent investigations have demonstrated RDW's connections to numerous infectious and inflammatory conditions [24-27]. In individuals diagnosed with COVID-19, RDW levels were significantly higher among those who developed severe disease or died, as opposed to those who exhibited mild symptoms or recovered from the infection [28-30]. Yeşil et al. revealed that patients with inflammatory bowel disease in the active phase showed higher RDW levels than those in remission, indicating its possible role as a precise and responsive biomarker for identifying active Crohn’s disease [24]. Additionally, findings from a retrospective analysis indicated that elevated RDW correlated with increased hospitalization rates and higher 30-day mortality among younger patients suffering from community-acquired pneumonia [25]. To our knowledge, this research represents the first investigation exploring RDW’s correlation with short-term clinical outcomes in young SAE patients through analysis of population-level real-world evidence.

Alterations in blood coagulation mechanisms could account for the RDW-prognosis correlation among younger individuals with SAE. During septic conditions, substantial disruptions in coagulation processes occur. Under normal physiological conditions, the activation of blood clotting is regulated by intrinsic anticoagulation systems, differing from normal homeostasis maintained through tissue factor inhibition, protein C activation, and antithrombotic regulation [3]. Sepsis interferes with all three coagulation regulatory mechanisms. Research demonstrated that RDW showed positive associations with various prognostic indicators in multiple hepatic disorders, including serum bilirubin and creatinine concentrations, as well as PT measurements [31]. Elevated RDW is strongly associated with venous thrombosis and early death from acute pulmonary embolism [26, 32]. It is also possible for renal insufficiency to affect RBC production and platelet dysfunction [33, 34]. Table 1 data revealed a correlation between rising RDW levels and increased incidence of liver-related conditions in this study, which was associated with elevated values of INR, PT, APTT, as well as increased BUN and creatinine levels. This, to some extent, reflects coagulation disorder, hepatic and renal dysfunction in these patients. Coagulation disorders in young SAE patients may be one of the potential mechanisms by which high level of RDW leads to poor outcomes.

An elevated RDW may indicate ineffective erythropoiesis or delayed clearance of erythrocytes, which may be influenced by disease, as it is an early indication of physiological stress [30, 35, 36]. Anisocytosis may indicate the presence of an underlying inflammatory condition, with cytokine release potentially influencing bone marrow functionality [37]. The RDW showed a robust and autonomous positive correlation with traditional markers of inflammation [38]. Inflammatory biomarkers have been used to diagnose, treat, and prognosticate sepsis patients [39, 40]. In septic encephalopathy, infection is accompanied by inflammation and stress. Reduced blood pressure and an elevated heart rate indicate the cardiac stress condition in individuals with high RDW, while the increased heart rate serves as a compensatory mechanism in response to sympathetic nervous system activation. Additionally, patients with high RDW had more extreme differences in white blood cell count.

Previous studies have already suggested that RDW is closely related to systemic inflammation and the prognosis of sepsis, and on this basis the present study approached the topic from a new angle, focusing on the early warning needs of young SAE populations. Most previous work on RDW and sepsis has included elderly or all-age cohorts, whereas young patients, who have stronger physiological reserve and are often labelled as low risk, are often neglected; however, we found that in this population the mortality rate still reached as high as 9%, and the predictive performance of RDW (AUC=0.760) even exceeded previously reported values in elderly sepsis cohorts (AUC=0.710) [6, 7, 41]. This means that, in young SAE patients, RDW may capture acute injury signals that are more difficult for traditional risk models to identify, such as a more intense coagulation–inflammation cascade manifested as elevated INR, prolonged PT and impaired hepatic and renal function or microcirculatory dysfunction, rather than merely chronic organ failure, which is more common in older populations. At the same time, the independent value of RDW observed in this study challenges the traditional view that RDW is only a surrogate marker of inflammation. Even after strict PSM and comprehensive adjustment for covariates, RDW still maintained a significant predictive role (HR=1.2, P<0.001; Table 3), suggesting that the information it integrates also includes microcirculatory damage caused by abnormal RBC morphology, among which anisocytosis, that is, abnormal variation in RBC size, may disturb passage through cerebral capillaries and aggravate SAE-related ischemic and hypoxic injury.

On this basis, we further propose a more exploratory hypothesis: an elevated RDW may not only be a bystander of disease severity, but may also directly participate, through specific pathways, in neurological injury and adverse outcomes in SAE. Anisocytosis reduces the deformability of RBCs and their ability to pass through microvessels, particularly affecting cerebral microcirculation; against the background of sepsis, in which there is already uneven perfusion, stagnant blood flow and tissue hypoxia, an elevated RDW may further limit the delivery of oxygen and nutrients, thereby aggravating ischemic and hypoxic brain injury. The oxygen-carrying and oxygen-releasing functions of morphologically abnormal RBCs may also be impaired, and, in the context of increased metabolic demands of brain tissue during SAE, such functional defects may contribute to relative tissue hypoxia and drive the continuous progression of neurological dysfunction. It is also noteworthy that there is a bidirectional interaction between neuroinflammation in the brain and systemic inflammation and the RBC system: the core mechanisms of SAE include activation of microglia and the release of pro-inflammatory cytokines such as IL-6 and TNF-α, which can inhibit bone marrow hematopoiesis or shorten RBC lifespan; conversely, the intense systemic inflammation and potential oxidative stress reflected by elevated RDW may in turn, by disrupting the blood–brain barrier or through humoral signaling pathways, aggravate neuroinflammation within the brain and form a self-reinforcing vicious cycle [42, 43]. The characteristics and significance of this central–peripheral interaction in young patients with SAE still need to be elucidated systematically.

Compared with many complex or expensive tests, measurement of RDW is simple, rapid and easy to repeat; several studies have linked it with inflammatory diseases and thrombosis, suggesting that RDW may be closely related to the occurrence and prognosis of SAE [24-27]. Because it is affected by the lifespan of RBCs of about 3-4 months, RDW can integrate disease progression information over a relatively long time scale; this characteristic gives it a certain irreplaceability and also makes it promising in improving the accuracy of existing prognostic models for sepsis and SAE. The present study adopted a strict one-to-one PSM design, effectively balancing baseline differences, so that the significant association observed in the matched cohort between elevated RDW and 30-day mortality (HR=2.7, 95% CI: 1.4-5.3, P=0.003), as well as the obvious differences in survival rate between the high- and low-RDW groups (9.9% vs 3.4%, P<0.001), can more reasonably be attributed to RDW itself or to the pathophysiological states that it represents but have not yet been fully quantified, rather than to confounding factors. Relying on the large-scale, high-quality real-world database MIMIC-IV, the results of this study appropriately reflect the predictive performance of RDW in actual clinical settings, suggesting that this easily obtainable, low-cost routine test index has high potential for clinical translation, especially in settings with limited resources or without advanced bedside monitoring methods, where it can be used for rapid risk stratification and optimization of resource allocation.

Although in young SAE cases we observed a robust association between RDW levels and outcomes, this study still inevitably has limitations. The retrospective design based on the MIMIC-IV database may have introduced selection bias; even though PSM was used to reduce baseline differences as far as possible, residual confounding is still difficult to eliminate completely. Some variables in the database were missing, and even though appropriate handling has been carried out, their potential impact on the results still cannot be completely ruled out; at the same time, this study relied on a single database and lacked external cohort validation, so that the generalizability and applicability of the conclusions still need further testing. In addition, this study only evaluated the relationship between RDW and 30-day mortality, and the impact on more long-term outcomes such as long-term survival and quality of life has not yet been systematically assessed. The enrollment of SAE patients in our center is inherently challenging, primarily due to the limited number of eligible cases encountered in routine clinical practice. Therefore, we chose to use a public database for our initial SAE research.

Future research can build on the findings and limitations of this study and continue to expand. On the one hand, it is necessary to carry out large-scale prospective cohort studies covering multiple regions and multiple medical centers, in order to externally validate the results of this study and to compare the predictive performance of RDW for outcomes in young SAE patients under different clinical environments and treatment strategies; on the other hand, the intrinsic links between RDW, coagulopathy and inflammatory responses should be explored in depth, to clarify the specific biological pathways by which high RDW affects SAE prognosis and to provide clues for potential therapeutic targets. Combining RDW with other novel biomarkers and clinical indicators is expected to construct a more comprehensive prognostic evaluation model for SAE and improve predictive accuracy; for young SAE patients with elevated RDW, strengthening follow-up and dynamic monitoring, particularly the assessment of cognitive function, mental health and overall quality of life, will also help to optimize clinical care and rehabilitation strategies.

On the basis of the new perspective proposed in this study, subsequent work can also focus on several directions with greater application prospects. At the clinical level, multicenter external validation studies are needed to systematically assess the generalizability, sensitivity and specificity of the RDW cut-off value (14.65%) determined in this study in independent and diverse cohorts of young SAE, and to further explore whether this threshold needs to be modified according to the source of infection such as Gram-positive or Gram-negative bacteria or fungi, pathogen virulence or the host genetic background. At the same time, the value of dynamic RDW trajectories should be a focus, for example changes at 24 h, 48 h, 72 h and even later time points after ICU admission, in order to observe whether they have earlier and more precise predictive ability than a single measurement and whether they can indicate disease progression trends, treatment responses such as the effects of anti-infective or anti-inflammatory therapies and the potential for neurological recovery. Another important goal is to construct integrated predictive models that combine RDW especially dynamic RDW with existing organ function scores such as SOFA and APACHE II, novel biomarkers such as the aforementioned markers of neuronal injury and specific inflammatory cytokines and feasible bedside microcirculatory or physiological function assessments, so as to promote the real implementation of precision medicine in critically ill young patients. To verify whether intensified treatment or refined monitoring strategies based on RDW risk stratification can improve short-term survival, neurological recovery and long-term outcomes, rigorously designed prospective randomized controlled trials are still needed, with comprehensive assessment of their risk–benefit ratio, cost-effectiveness and impact on medical resource allocation; this will be one of the key steps in elevating RDW from a simple prognostic marker to a tool for guiding treatment decisions.

Taken together, using a strict PSM method and large-scale real-world data, this study clearly delineated the level of short-term mortality in young patients with SAE and confirmed that elevated RDW at ICU admission is independently associated with 30-day mortality risk in this population. In doing so, it not only undermines the ingrained perception that young patients are naturally at lower risk, but also highlights the unique value of RDW in prognostic assessment among young SAE patients. Young SAE patients with high RDW levels should therefore receive special attention and more active management in clinical practice.

Author contributions

Yalin Zhu, Zhengyu Jiang and Wangzheqi Zhang wrote the manuscript and performed the data analysis. Jie Huang, Haoling Zhang, Haiwen Wang and Wen Xu checked the data and the results. Yalin Zhu provided the funding of this study. Jiafeng Wang designed the study. Jiafeng Wang and Wen Xu revised the manuscript and supervised research. All authors have reviewed and approved the final version.

Funding

This study is supported by Zhoushan Medical and Health Technology Program (2023JYB01).

Data availability

The complete dataset referenced throughout our publications is accessible via the public MIMIC-IV database at https://mimic.mit.edu/iv/.

Ethics approval and consent to participate

This investigation adhered to the ethical standards of the Helsinki Declaration in 1964 and its subsequent amendments, or equivalent ethical guidelines, utilizing de-identified data obtained from the publicly accessible MIMIC-IV database. Access authorization was secured through certification ID 55826703 by researcher JH. Ethics approval was granted under a consent waiver protocol by the IRBs of Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Acknowledgements

Not applicable.

SUPPLEMENTARY METERIAL

[1] Mazeraud A, Righy C, Bouchereau E, Benghanem S, Bozza FA, Sharshar T. Septic-associated encephalopathy: A compre-hensive review. Neurotherapeutics. 2020 Apr;17(2):392-403. https://doi.org/10.1007/s13311-020-00862-1

[2] Gofton TE, Young GB. Sepsis-associated encephalopathy. Nat Rev Neurol. 2012 Oct;8(10):557-566. https://doi.org/10.1038/nrneurol.2012.183

[3] Levi M, Poll Tv. Coagulation in patients with severe sepsis. Semin Thromb Hemost. 2015 Feb;41(1):9-15. https://doi.org/10.1055/s-0034-1398376

[4] Tauber SC, Djukic M, Gossner J, Eiffert H, Brück W, Nau R. Sepsis-associated encephalopathy and septic encephalitis: An update. Expert Rev Anti Infect Ther. 2021 Feb;19(2):215-231. https://doi.org/10.1080/14787210.2020.1812384

[5] Iwashyna TJ, Ely EW, Smith DM, Langa KM. Long-term cognitive impairment and functional disability among survivors of severe sepsis. JAMA. 2010 Oct 27;304(16):1787-1794. https://doi.org/10.1001/jama.2010.1553

[6] Jin G, Wang S, Chen J, Hu W, Zhu Y, Xi S. Identification of sepsis-associated encephalopathy risk factors in elderly patients: A retrospective observational cohort study. Turk J Med Sci. 2022 Oct;52(5):1513-1522. https://doi.org/10.55730/1300-0144.5491

[7] Zhang Z, Guo L, Jia L, Duo H, Shen L, Zhao H. Factors contributing to sepsis-associated encephalopathy: A comprehensive systematic review and meta-analysis. Front Med (Lausanne). 2024 May 21;11:1379019. https://doi.org/10.3389/fmed.2024.1379019

[8] Förhécz Z, Gombos T, Borgulya G, Pozsonyi Z, Prohászka Z, Jánoskuti L. Red cell distribution width in heart failure: Prediction of clinical events and relationship with markers of ineffective erythropoiesis, inflammation, renal function, and nutritional state. Am Heart J. 2009 Oct;158(4):659-666. https://doi.org/10.1016/j.ahj.2009.07.024

[9] Kim CH, Park JT, Kim EJ, Han JH, Han JS, Choi JY, et al. An increase in red blood cell distribution width from baseline predicts mortality in patients with severe sepsis or septic shock. Crit Care. 2013 Dec 9;17(6):R282. https://doi.org/10.1186/cc13145

[10] Jo YH, Kim K, Lee JH, Kang C, Kim T, Park HM, et al. Red cell distribution width is a prognostic factor in severe sepsis and septic shock. Am J Emerg Med. 2013 Mar;31(3):545-548. https://doi.org/10.1016/j.ajem.2012.10.017

[11] Salvagno GL, Sanchis-Gomar F, Picanza A, Lippi G. Red blood cell distribution width: A simple parameter with multiple clinical applications. Crit Rev Clin Lab Sci. 2015;52(2):86-105. https://doi.org/10.3109/10408363.2014.992064

[12] Malandrino N, Wu WC, Taveira TH, Whitlatch HB, Smith RJ. Association between red blood cell distribution width and macrovascular and microvascular complications in diabetes. Diabetologia. 2012 Jan;55(1):226-235. https://doi.org/10.1007/s00125-011-2331-1

[13] Ferreira JP, Lamiral Z, Bakris G, Mehta C, White WB, Zannad F. Red cell distribution width in patients with diabetes and myocardial infarction: An analysis from the EXAMINE trial. Diabetes Obes Metab. 2021 Jul;23(7):1580-1587. https://doi.org/10.1111/dom.14371

[14] Wu J, Zhang X, Liu H, Guo N, Pan Q, Wang Y. RDW, NLR and RLR in predicting liver failure and prognosis in patients with hepatitis E virus infection. Clin Biochem. 2019 Jan;63:24-31. https://doi.org/10.1016/j.clinbiochem.2018.11.012

[15] Hua R, Liu X, Yuan E. Red blood cell distribution width at admission predicts outcome in critically ill patients with kidney failure: A retrospective cohort study based on the MIMIC-IV database. Ren Fail. 2022 Dec;44(1):1182-1191. https://doi.org/10.1080/0886022X.2022.2098766

[16] Peng L, Peng C, Yang F, Wang J, Zuo W, Cheng C, et al. Machine learning approach for the prediction of 30-day mortality in patients with sepsis-associated encephalopathy. BMC Med Res Methodol. 2022 Jul 4;22(1):183. https://doi.org/10.1186/s12874-022-01664-z

[17] Yang Y, Liang S, Geng J, Wang Q, Wang P, Cao Y, et al. Development of a nomogram to predict 30-day mortality of patients with sepsis-associated encephalopathy: A retrospective cohort study. J Intensive Care. 2020 Jul 2;8:45. https://doi.org/10.1186/s40560-020-00459-y

[18] Sadaka F, Doctors N, Pearson T, Snyders B, O’Brien J. Does red cell distribution width predict outcome in traumatic brain injury: Comparison to corticosteroid randomization after significant head injury. J Clin Med Res. 2018 Jan;10(1):9-12. https://doi.org/10.14740/jocmr3173w

[19] Kenangil G, Ari BC, Kaya FA, Demir M, Domac FM. Red cell distribution width levels in Parkinson’s disease patients. Acta Neurol Belg. 2020 Oct;120(5):1147-1150. https://doi.org/10.1007/s13760-019-01197-1

[20] Wu WT, Li YJ, Feng AZ, Li L, Huang T, Xu AD, et al. Data mining in clinical big data: The frequently used databases, steps, and methodological models. Mil Med Res. 2021 Aug 11;8(1):44. https://doi.org/10.1186/s40779-021-00338-z

[21] Zhou S, Zeng Z, Wei H, Sha T, An S. Early combination of albumin with crystalloids administration might be beneficial for the survival of septic patients: a retrospective analysis from MIMIC-IV database. Ann Intensive Care. 2021 Mar 10;11(1):42. https://doi.org/10.1186/s13613-021-00830-8

[22] Lam AP, Gundabolu K, Sridharan A, Jain R, Msaouel P, Chrysofakis G, et al. Multiplicative interaction between mean corpuscular volume and red cell distribution width in predicting mortality of elderly patients with and without anemia. Am J Hematol. 2013 Nov;88(11):E245-E249. https://doi.org/10.1002/ajh.23529

[23] Lippi G, Mattiuzzi C, Cervellin G. Learning more and spending less with neglected laboratory parameters: The paradigmatic case of red blood cell distribution width. Acta Biomed. 2016 Jan 16;87(3):323-328.

[24] Yeşil A, Senateş E, Bayoğlu IV, Erdem ED, Demirtunç R, Kurdaş Övünç AO. Red cell distribution width: A novel marker of activity in inflammatory bowel disease. Gut Liver. 2011 Dec;5(4):460-467. https://doi.org/10.5009/gnl.2011.5.4.460

[25] Braun E, Domany E, Kenig Y, Mazor Y, Makhoul BF, Azzam ZS. Elevated red cell distribution width predicts poor outcome in young patients with community acquired pneumonia. Crit Care. 2011 Aug 11;15(4):R194. https://doi.org/10.1186/cc10355

[26] Rezende SM, Lijfering WM, Rosendaal FR, Cannegieter SC. Hematologic variables and venous thrombosis: Red cell distribution width and blood monocyte count are associated with an increased risk. Haematologica. 2014 Jan;99(1):194-200. https://doi.org/10.3324/haematol.2013.083840

[27] Şenol K, Saylam B, Kocaay F, Tez M. Red cell distribution width as a predictor of mortality in acute pancreatitis. Am J Emerg Med. 2013 Apr;31(4):687-689. https://doi.org/10.1016/j.ajem.2012.12.015

[28] Gama S. RDW shows prognostic potential in hospitalized patients with COVID-19. J Med Virol. 2022 Aug;94(8):3498-3500. https://doi.org/10.1002/jmv.27764

[29] Lee JJ, Montazerin SM, Jamil A, Jamil U, Marszalek J, Chuang ML, et al. Association between red blood cell distribution width and mortality and severity among patients with COVID-19: A systematic review and meta-analysis. J Med Virol. 2021 Apr;93(4):2513-2522. https://doi.org/10.1002/jmv.26797

[30] Rapp JL, Tremblay D, Alpert N, Lieberman-Cribbin W, Mascarenhas J, Taioli E, et al. Red cell distribution width is associated with mortality in non-anemic patients with COVID-19. J Med Virol. 2021 Jul;93(7):4130-4132. https://doi.org/10.1002/jmv.27011

[31] Hu Z, Sun Y, Wang Q, Han Z, Huang Y, Liu X, et al. Red blood cell distribution width is a potential prognostic index for liver disease. Clin Chem Lab Med. 2013 Jul;51(7):1403-1408. https://doi.org/10.1515/cclm-2012-0704

[32] Zorlu A, Bektasoglu G, Guven FM, Dogan OT, Gucuk E, Ege MR, et al. Usefulness of admission red cell distribution width as a predictor of early mortality in patients with acute pulmonary embolism. Am J Cardiol. 2012 Jan 1;109(1):128-134. https://doi.org/10.1016/j.amjcard.2011.08.015

[33] Lipkin GW, Kendall RG, Russon LJ, Turney JH, Norfolk DR, Brownjohn AM. Erythropoietin deficiency in acute renal failure. Nephrol Dial Transplant. 1990;5(11):920-922. https://doi.org/10.1093/ndt/5.11.920

[34] Jensen JLS, Hviid CVB, Hvas CL, Christensen S, Hvas AM, Larsen JB. Platelet function in acute kidney injury: A Systematic review and a cohort study. Semin Thromb Hemost. 2023 Jul;49(5):507-522. https://doi.org/10.1055/s-0042-1757167

[35] Patel HH, Patel HR, Higgins JM. Modulation of red blood cell population dynamics is a fundamental homeostatic response to disease. Am J Hematol. 2015 May;90(5):422-428. https://doi.org/10.1002/ajh.23982

[36] Rogiers P, Zhang H, Leeman M, Nagler J, Neels H, Mélot C, et al. Erythropoietin response is blunted in critically ill patients. Intensive Care Med. 1997 Feb;23(2):159-162. https://doi.org/10.1007/s001340050310

[37] Jelkmann WE, Fandrey J, Frede S, Pagel H. Inhibition of erythropoietin production by cytokines. Implications for the anemia involved in inflammatory states. Ann N Y Acad Sci. 1994 Apr 15;718:300-309; discussion 309-311.

[38] Lippi G, Targher G, Montagnana M, Salvagno GL, Zoppini G, Guidi GC. Relation between red blood cell distribution width and inflammatory biomarkers in a large cohort of unselected outpatients. Arch Pathol Lab Med. 2009 Apr;133(4):628-632. https://doi.org/10.5858/133.4.628

[39] Huang Z, Fu Z, Huang W, Huang K. Prognostic value of neutrophil-to-lymphocyte ratio in sepsis: A meta-analysis. Am J Emerg Med. 2020 Mar;38(3):641-647. https://doi.org/10.1016/j.ajem.2019.10.023

[40] Zhang S, Luan X, Zhang W, Jin Z. Platelet-to-lymphocyte and neutrophil-to-lymphocyte ratio as predictive biomarkers for early-onset neonatal sepsis. J Coll Physicians Surg Pak. 2021 Jul;31(7):821-824. https://doi.org/10.29271/jcpsp.2021.07.821

[41] Wang TH, Hsu YC. Red cell distribution width as a prognostic factor and its comparison with lactate in patients with sepsis. Diagnostics (Basel). 2021 Aug 14;11(8):1474. https://doi.org/10.3390/diagnostics11081474

[42] Cluzeau T, McGraw KL, Irvine B, Masala E, Ades L, Basiorka AA, et al. Pro-inflammatory proteins S100A9 and tumor necrosis factor-α suppress erythropoietin elaboration in myelodysplastic syndromes. Haematologica. 2017 Dec;102(12):2015-2020. https://doi.org/10.3324/haematol.2016.158857

[43] Madeddu C, Gramignano G, Astara G, Demontis R, Sanna E, Atzeni V, et al. Pathogenesis and treatment options of cancer related anemia: Perspective for a targeted mechanism-based approach. Front Physiol. 2018 Sep 20;9:1294. https://doi.org/10.3389/fphys.2018.01294