Association Between Red Cell Distribution Width and Hospital Mortality in Patients with Sepsis

Abstract

Objective

Sepsis is the leading cause of death in patients admitted to adult intensive care units (ICUs). We aimed to determine the predictive value of red blood cell distribution width (RDW) in patients with sepsis in a large cohort.

Methods

This retrospective observational study used data from the eICU Collaborative Research Database. The prognostic value of RDW was investigated using the receiver operating characteristic (ROC) curve, multiple logistic regression model, integrated discriminatory index (IDI), and net reclassification index (NRI).

Results

In total, 9743 patients were included. The area under the ROC curve of the RDW for predicting hospital mortality was 0.631 (95% confidence interval [CI]: 0.616–0.645). Based on the multiple logistic regression model, an RDW of ≥14.5% was correlated with hospital mortality, regardless of Sequential Organ Failure Assessment (SOFA) and Acute Physiology and Chronic Health Evaluation IV (APACHE IV) scores (odds ratio [OR]: 1.838, 95% CI: 1.598–2.119). Using SOFA and APACHE IV scores as reference, the IDI and continuous NRI of RDW for hospital mortality was about 0.3 and 0.014, respectively.

Conclusions

The RDW may be useful in predicting hospital mortality in patients with sepsis, offering extra prognostic value beyond SOFA and APACHE IV scores.

Keywords

Red cell distribution width eICU Collaborative Research Database sepsis hospital mortality intensive care unit Sequential Organ Failure Assessment Acute Physiology and Chronic Health Evaluation IV

Introduction

Red blood cell distribution width is a measure of the spectrum of discrepancy in red blood cell volume and is commonly assessed as part of a complete blood count. The normal reference range for red cell distribution width in adults is 11.5% to 14.5%.¹ An increase in red cell distribution width suggests that some red blood cells are either too large or too small, which is indicative of several abnormalities, such as oxidative stress, hypertension, poor nutritional status, inflammation, erythrocyte fragmentation, dyslipidemia, and alteration of erythropoietin function.² This measure can also be used as a predictor of poor outcomes in several diseases, such as cardiovascular diseases,^3–5 peripheral artery disease,⁶ cerebrovascular disease,^7,8 pulmonary embolism,⁹ cancers,^10,11 kidney disease,¹² and chronic obstructive pulmonary disease.¹³ Red cell distribution width is a widely used laboratory indicator and is increasingly valued.

Sepsis is a potentially life-threatening condition caused by the body’s response to an infection. Sepsis is the most expensive condition treated in hospitals in the United States, accounting for more than $23 billion (6.2%) of the total hospital costs in that country in 2013.¹⁴ One report showed that no significant changes were observed in the sepsis mortality rate from 2002 to 2016, after controlling for severity.¹⁵ The severity of sepsis can range from mild to fatal. Hence, a reliable method for clinical practitioners to predict mortality in patients with sepsis is urgently needed. In addition to several scoring systems, such as the quick Sequential Organ Failure Assessment (qSOFA),¹⁶ and New York Sepsis Severity Score,¹⁷ several markers have been found to be effective in predicting the prognosis of sepsis, including procalcitonin, presepsin, galectin-3, soluble suppression of tumorigenicity-2, fibroblast growth factor 21, interleukin-8, and soluble tumor necrosis factor-alpha receptor 1.^18–20 However, the tools required to measure some of these biomarkers are not available in most hospitals. Even the use of lactate level, as a classical predictor of sepsis, remains controversial.²¹ Hence, a new and simple predictor with satisfactory outcomes must be identified.

Several authors have explored the relationship between red cell distribution width and mortality in patients with sepsis. However, the participants in previous studies were neonates^22,23 or older adults,²⁴ and some studies were limited by a small sample size.^25–27 Thus, in the present study, we aimed to determine the predictive value of red cell distribution width in patients with sepsis using a large cohort.

Materials and Methods

Database

Philips Healthcare has developed a remote health system called the eICU Program, which uses a large amount of data to support the management of critically ill patients. These data are archived and then transformed into a structured query language database, to be used for further research by the eICU Research Institute (eRI).²⁸ The Laboratory for Computational Physiology at the Massachusetts Institute of Technology (LCP-MIT) has partnered with the eRI to establish the eICU Collaborative Research Database (eICU-CRD), which is a large, multicenter intensive care unit (ICU) database with high-granular data on over 200,000 ICU admissions monitored by eICU Programs across the United States in 2014 and 2015. The database comprises 208 hospitals, including 19 teaching hospitals; 70 of these hospitals are located in the Midwestern United States, 56 in the South, and 43 in the West. The data are deidentified and include information on, for example, vital signs, diagnosis, disease severity measures, treatment information, and care plan documentation. Data on the clinical outcomes are available until patients are discharged from the hospital. The database is constantly updated, and the current version is v2.0 (17 May 2018). In this study, we used the current version of the database, and PostgreSQL v11.2 was used to query and manage the data (http://www.postgresql.org/).

After completing a training course of the Collaborative Institutional Training Initiative (CITI Program), one author (YL) was authorized to access the data from the eICU-CRD for medical research purposes. The database was released under the Health Insurance Portability and Accountability Act (HIPAA) safe harbor provision. The reidentification risk was certified as meeting safe harbor standards by Privacert (Cambridge, MA) (HIPAA certification no. 1031219-2). Because this study was a retrospective analysis of a third-party anonymized publicly available database with pre-existing institutional review board (IRB) approval, IRB approval from our institution and patient consent were not required.

Study population

The inclusion criteria were as follows:

Patients with a primary diagnosis of sepsis upon admission to the ICU

Patients aged older than 16 years and younger than 89 years

The exclusion criteria were as follows:

Patients who had been repeatedly admitted in the ICU

Patients whose length of ICU stay was <4 hours

Patients whose SOFA score was <2

Patients whose key information was unavailable (sex, Acute Physiology and Chronic Health Evaluation IV [APACHE IV] score, mortality, and red cell distribution width assessed within the first 24 hours in the ICU)

The eICU-CRD includes patients who were admitted to the ICU from 2014 to 2015. The inclusion of patients was carried out before publication of the third definition of sepsis.²⁹ Thus, we included SOFA scores in our inclusion criteria.

Data, including demographic characteristics, severity-of-disease scores, mortality, red cell distribution width, other laboratory results, cause of ICU admission, length of ICU stay, and vital signs, were obtained from the eICU-CRD. If additional laboratory results were recorded for the same patient within the first 24 hours of ICU admission, only the first set of laboratory results was included in the study. In addition, the initial vital signs upon ICU admission were included. Severity-of-disease scores included SOFA and APACHE IV scores. The SOFA source code was shared by Alistair Johnson.³⁰

Statistical analysis

The distribution of continuous variables was assessed using the Shapiro–Wilk test, and none of these variables showed a normal distribution. Data were expressed as either frequency and percentage for categorical variables or median and interquartile range (IQR) for continuous variables. The Kruskal–Wallis H test was used to compare continuous data. Categorical variables were compared using the Fisher’s exact test. The receiver operating characteristic (ROC) curve and two multivariable logistic regression models were used to investigate the possible association between red cell distribution width and hospital mortality. Approximately <12.5% of the covariates were missing; hence, multiple imputation by chained equations was adopted to replace the missing values. The net reclassification index (NRI) and integrated discriminatory index (IDI) were used to measure improvement in the predictive performance of red blood cell distribution when added to the classical severity-of-disease scores.^31,32 We performed a subgroup analysis for each etiology of sepsis. All significance tests were two-sided, and a p value < 0.05 was considered statistically significant. All analyses were conducted using R 3.6.0 with the mice, pROC,³³ PredictABEL, and ggplot2 packages (The R Project for Statistical Computing, Vienna, Austria).

Sensitivity analyses

Sensitivity analyses were performed to evaluate the effect of missing data in multivariable logistic regression model 2. Ten different random numbers between 1 and 10,000, generated using the R command, were used as seed values in the multiple imputations. We also performed a sensitivity analysis using complete case analysis.

Results

Participants

Figure 1 shows the process of selecting participants who were eligible for the study. Finally, 9743 patients from 177 hospitals were included in the study.

Figure 1.

Patient selection process.

Baseline characteristics of participants

The baseline characteristics of the study population are shown in Table 1. The median age of the 9743 study participants was 65 years. Of these, 48% were women. Participants were categorized into two identical groups according to the normal reference values for red cell distribution width (normal: ≤14.5% and high: >14.5%). The group with a higher red cell distribution width had higher APACHE IV scores and higher hospital mortality rates (p < 0.001). It also had older age and a higher percentage of women (p < 0.001).

Table 1.

Baseline characteristics of the study population.

	Overall	RDW
	Overall	Normal (≤14.5%)	High (>14.5%)	p ^#
N	9743	3218	6525
Age (years), median [IQR]	67.00 [56.00, 77.00]	65.00 [54.00, 76.00]	68.00 [58.00, 78.00]	<0.001
Type of sepsis, n (%)				0.019
Sepsis, cutaneous/soft tissue	797 (8.2)	242 (7.5)	555 (8.5)
Sepsis, GI	1137 (11.7)	385 (12.0)	752 (11.5)
Sepsis, gynecologic	26 (0.3)	12 (0.4)	14 (0.2)
Sepsis, other	647 (6.6)	182 (5.7)	465 (7.1)
Sepsis, pulmonary	3535 (36.3)	1208 (37.5)	2327 (35.7)
Sepsis, renal/UTI	2424 (24.9)	814 (25.3)	1610 (24.7)
Sepsis, unknown	1177 (12.1)	375 (11.7)	802 (12.3)
Female sex, n (%)	4680 (48.0)	1476 (45.9)	3204 (49.1)	0.003
Ethnicity, n (%) *				<0.001
African American	1084 (11.2)	240 (7.5)	844 (13.0)
Asian	141 (1.5)	47 (1.5)	94 (1.5)
White	7565 (78.2)	2564 (80.0)	5001 (77.2)
Hispanic	373 (3.9)	167 (5.2)	206 (3.2)
Native American	57 (0.6)	16 (0.5)	41 (0.6)
Other/Unknown	460 (4.8)	169 (5.3)	291 (4.5)
Death in ICU, n (%)	1218 (12.5)	240 (7.5)	978 (15.0)	<0.001
Length of ICU stay (hours), median [IQR]	55.00 [30.00, 104.00]	51.00 [29.00, 97.00]	58.00 [31.00, 107.00]	0.002
Death in hospital, n(%)	1623 (16.7)	319 (9.9)	1304 (20.0)	<0.001
SOFA score, median [IQR]	5.00 [3.00, 8.00]	5.00 [3.00, 7.00]	5.00 [3.00, 8.00]	<0.001
APACHE IV score, median [IQR]	67.00 [53.00, 85.00]	63.00 [49.00, 79.00]	70.00 [55.00, 88.00]	<0.001
BMI (kg/m²), median [IQR]	27.12 [22.90, 33.15]	26.88 [22.97, 32.26]	27.29 [22.86, 33.59]	0.027
MAP (mmHg), median [IQR]	74.30 [64.30, 86.70]	75.70 [65.70, 88.70]	73.70 [64.00, 86.00]	<0.001
Heart rate, median [IQR]	97.00 [82.00, 112.00]	98.00 [84.00, 113.00]	96.00 [82.00, 112.00]	0.004
Temperature (°C), median [IQR]	36.90 [36.50, 37.40]	36.90 [36.60, 37.60]	36.80 [36.40, 37.40]	<0.001
RDW (%), median [IQR]	15.40 [14.10, 17.30]	13.70 [13.20, 14.10]	16.50 [15.40, 18.20]	<0.001
MCH (pg), median [IQR]	29.90 [28.10, 31.40]	30.70 [29.40, 31.90]	29.40 [27.50, 31.10]	<0.001
MCHC (g/dL), median [IQR]	32.80 [31.70, 33.70]	33.40 [32.60, 34.20]	32.40 [31.40, 33.30]	<0.001
Albumin (g/dL), median [IQR]	2.90 [2.40, 3.40]	3.20 [2.70, 3.60]	2.80 [2.30, 3.30]	<0.001
Creatinine (mg/dL), median [IQR]	1.50 [0.99, 2.56]	1.38 [0.94, 2.20]	1.60 [1.00, 2.76]	<0.001
Glucose (mg/dL), median [IQR]	131.00 [104.00, 177.00]	135.00 [109.00, 187.00]	128.00 [102.00, 173.00]	<0.001
WBC (×10⁹/L), median [IQR]	13.60 [8.70, 19.30]	13.70 [9.00, 19.10]	13.58 [8.60, 19.48]	0.674
Hematocrit (%), median [IQR]	34.90 [29.70, 39.80]	37.60 [33.23, 42.00]	33.30 [28.50, 38.50]	<0.001
Hemoglobin (g/dL), median [IQR]	11.40 [9.60, 13.10]	12.60 [11.10, 14.00]	10.70 [9.10, 12.50]	<0.001
Platelet (×10⁹/L), median [IQR]	202.00 [139.00, 283.00]	198.00 [145.00, 267.25]	205.00 [136.00, 292.00]	0.023

*Owing to missing data, the sum of patients was not equal to the total cohort number.

^#Continuous data were compared using the Kruskal–Wallis H test. Categorical variables were compared using the Fisher’s exact test.

GI, gastrointestinal; UTI, urinary tract infection; ICU, intensive care unit; BMI, body mass index; MAP, mean arterial pressure; RDW, red blood distribution width; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration; WBC, white blood cell; IQR, interquartile range; SOFA, Sequential Organ Failure Assessment; APACHE IV, Acute Physiology and Chronic Health Evaluation IV.

Association between red blood cell distribution width and hospital mortality

We performed multiple logistic regression analysis to explore the association between red cell distribution width and hospital mortality. In model 1, after adjusting for SOFA and APACHE IV scores, a significant correlation was still observed between a high red cell distribution width and hospital mortality, with an odds ratio (OR) of 1.838 (95% confidence interval [CI]: 1.598–2.119). There were no missing values in the variables used in this model. We ran an alternative model to determine the relationship between red cell distribution width and hospital mortality, in which the covariates were adjusted for SOFA and APACHE IV scores; age; albumin, creatinine, glucose, hematocrit, hemoglobin, platelet, and potassium levels; white blood cell count; mean arterial pressure; and body mass index. Approximately <12.5% of the covariate data were missing. These missing covariates were then imputed by multiple imputation using the mice package in R. Moreover, a high red cell distribution width was significantly associated with hospital mortality, after adjusting for potential confounders (OR: 1.608, 95% CI: 1.455–1.762) (Table 2).

Table 2.

Multivariable logistic regression analysis of the relationship between RDW and hospital mortality.

	Model 1		Model 2
	OR (95% CI)	p	OR (95% CI)	p
RDW (high group)	1.838 (1.598–2.119)	<0.001	1.608 (1.455–1.762)	<0.001
SOFA (per 1 score)	1.109 (1.083–1.135)	<0.001	1.140 (1.112–1.168)	<0.001
APACHE IV (per 1 score)	1.030 (1.027–1.033)	<0.001	1.025 (1.021–1.028)	<0.001

Model 1 adjusted for SOFA and APACHE IV scores.

Model 2 adjusted for SOFA and APACHE IV scores; age; albumin, creatinine, glucose, hematocrit, hemoglobin, platelet, and potassium levels; white blood cell count; mean arterial pressure; and body mass index.

RDW, red blood distribution width; OR, odds ratio; CI, confidence interval; SOFA, Sequential Organ Failure Assessment; APACHE IV, Acute Physiology and Chronic Health Evaluation IV.

ROC analysis was performed to investigate the relationship between red cell distribution width and hospital mortality. The ROC curves for red cell distribution width and classical severity-of-disease scores are illustrated in Figure 2, and the area under the ROC curve (AUC) values are presented in Table 3.

Figure 2.

Receiver operating characteristic curves of red blood cell distribution (RDW) and classical severity scores for hospital mortality.

Table 3.

Area under the receiver operating characteristic curve (AUC) for red blood cell distribution width and severity scores.

Name	AUC (95% CI)
RDW	0.631 (0.616–0.645)
SOFA	0.727 (0.712–0.741)
APACHE IV	0.769 (0.757–0.782)

RDW, red blood distribution width; SOFA, Sequential Organ Failure Assessment; APACHE IV, Acute Physiology and Chronic Health Evaluation IV scores; CI, confidence interval.

The AUCs of the three curves were > 0.6, and the AUC of the red cell distribution width was lower than that of the SOFA and APACHE IV scores. To determine whether red cell distribution width had a higher prognostic value than the classical severity-of-disease scores, we analyzed the data using both the IDI and NRI, as presented in Table 4. Red cell distribution width significantly improved the predictive efficacy of the classical severity-of-disease scores (p < 0.001).

Table 4.

Extra predictive value of RDW over classical severity scores.

	Continuous NRI		IDI
	Estimates (95% CI)	p	Estimates (95% CI)	p
SOFA	0.308 (0.256–0.361)	<0.001	0.015 (0.012–0.019)	<0.001
APACHE IV	0.305 (0.252–0.358)	<0.001	0.013 (0.010–0.017)	<0.001

IDI, integrated discriminatory index; NRI, net reclassification index; SOFA, Sequential Organ Failure Assessment; APACHE IV, Acute Physiology and Chronic Health Evaluation IV scores; CI, confidence interval.

Subgroup analyses

We conducted a series of subgroup analyses to examine the effect of red cell distribution width across various subgroups, which were classified according to the cause of sepsis. The distribution of red cell distribution width in each subgroup is shown in Figure 3. Logistic regression analysis model 1 was used in each group, except for the gynecologic sepsis group, which only had 26 participants. The ORs and 95% CIs are summarized in Table 5. Subgroup analyses also revealed a significant association between high red cell distribution width and hospital mortality.

Figure 3.

Red blood cell distribution width (RDW) across all subgroups, classified by source of sepsis.

Table 5.

Multivariable logistic regression analysis of the relationship between RDW and hospital mortality in each sepsis subgroup.

	Cutaneous (n = 797)	Gastrointestinal (n = 1137)	Other (n = 647)	Pulmonary (n = 3535)	Urinary (n = 2424)	Unknown (n = 1177)
	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
RDW(High group)	2.415**(1.298–4.862)	1.883***(1.301–2.767)	2.946***(1.631–5.648)	1.538***(1.255–1.890)	2.630***(1.829–3.880)	1.886**(1.246–2.912)
SOFA(per 1 score)	1.117*(1.014–1.229)	1.102**(1.038–1.170)	1.073(0.974–1.182)	1.080***(1.042–1.121)	1.140***(1.076–1.210)	1.140***(1.060–1.228)
APACHE IV(per 1 score)	1.028***(1.016–1.040)	1.032***(1.024–1.041)	1.037***(1.025–1.050)	1.028***(1.023–1.033)	1.026***(1.019–1.034)	1.036***(1.027–1.046)

*p < 0.05, **p < 0.01, ***p < 0.001.

RDW, red blood distribution width; OR, odds ratio; CI, confidence interval; SOFA, Sequential Organ Failure Assessment; APACHE IV, Acute Physiology and Chronic Health Evaluation IV.

Sensitivity analyses

Figure 4 summarizes the results of sensitivity analyses using various seeds in multiple imputation and complete case analysis. As shown in Figure 4, the OR for high red cell distribution width and hospital mortality was >1.5 for all random seeds and complete case analyses. The results of all analyses were statistically significant (p < 0.05).

Figure 4.

Sensitivity analyses of complete case analysis and various random seeds in multiple imputations

Discussion

This retrospective study showed that red cell distribution width was an independent predictor of mortality in critically ill patients with sepsis. Although the AUC for red cell distribution width was lower than that of APACHE IV and SOFA scores, it showed significant predictive performance. The group with a high red cell distribution width had a higher risk of death (OR: 1.838, 95% CI: 1.598–2.119). Han initially explored the relationship between red cell distribution width and mortality in patients with sepsis using the data obtained from the Medical Information Mart for Intensive Care III (MIMIC III), a single-center database, during 2001 to 2012,³⁴ In the present study, we used multicenter data, a contemporary definition of sepsis, and patients who were newly diagnosed with sepsis. In addition to the classical severity scores, the red blood cell distribution width can also yield a prognostic value. Using the classical severity-of-disease scores as covariates, the continuous NRI and IDI of the red cell distribution width for hospital mortality was approximately 0.3 and 0.014, respectively. This suggests that incorporation of the red cell distribution width into classical severity-of-disease scoring systems would provide more information about the patients’ prognosis. Subgroup analysis showed that high red cell distribution width had a significant predictive value in each subgroup with different sepsis etiologies, particularly those with cutaneous sepsis (OR: 2.415, 95% CI: 1.298–4.862) and other sources of sepsis (OR: 2.946, 95% CI: 1.631–5.648).

Recently, some novel biomarkers have been found to be effective in predicting mortality in critically ill patients with sepsis, such as apoptosis inhibitor of macrophage/CD5L,³⁵ adrenomedullin,³⁶ and acetylcarnitine.³⁷ However, these biomarkers can only be used in the laboratory setting. As red cell distribution width is a routine parameter assessed as part of a complete blood cell count, measurement of this parameter does not involve additional cost. As red cell distribution width was found to improve the predictive efficacy of the classical severity scores, its inclusion must be considered in the development of new scoring systems in the future.

As mentioned in the published literature, red cell distribution width can predict poor outcome in several diseases. However, the exact mechanism had not been validated. One hypothesis proposes that red cell distribution width can indicate other underlying abnormalities. A red cell distribution width higher than the normal range may demonstrate abrupt red blood cell damage or, more generally, ineffective erythropoiesis. Previous studies have shown a strong association of red cell distribution width with erythrocyte sedimentation rate and high-sensitivity C-reactive protein (CRP), independent of abundant confounding factors.³⁸ Inflammatory factors, such as interleukin-6,³⁹ white blood cell count, fibrinogen,⁴⁰ and CRP can alter erythrocyte homeostasis and may help improve red cell distribution width levels by impairing iron metabolism, shortening red blood cell survival, and inhibiting red blood cell production. Thus, these phenomena may explain the findings of the current study. Moreover, underlying metabolic abnormalities, including poor nutritional status, shortening of telomere length, dyslipidemia, oxidative stress, and hypertension, can cause an increase in red cell distribution width.² Sepsis is a condition that involves organ dysfunction triggered by a dysregulated host response to infection; thus, all the above abnormalities reflecting the severity of sepsis may lead to a poor outcome.

Our study has several strengths, which differ from other previous studies. First, unlike one recent study,⁴¹ we analyzed the data of a large sample (n = 9743 from 177 hospitals), making this the largest study investigating the relationship between red cell distribution width and outcome of sepsis. Compared with patients from the MIMIC III database, the data of those from the eICU-CRD are more recent (2001–2012 vs. 2014–2015). Second, we revised our inclusion criteria by incorporating the SOFA score based on the latest definition of sepsis, which has rarely been adopted in previous retrospective studies.^34,42

The present study also has several limitations, which should be considered when interpreting the findings. First, owing to the retrospective nature of this study, selection bias and misclassification or information bias might had be introduced. Only randomized controlled trials can identify a causal relationship. Second, the red cell distribution width was only assessed in patients upon ICU admission, and the changes in red cell distribution width during the ICU stay were not evaluated. Third, owing to the absence of post-discharge data in the eICU-CRD, we did not assess the relationship between red cell distribution width and long-term prognosis in critically ill patients with sepsis. Fourth, some factors, such as deficiencies of iron, vitamin B12, and folate, were not considered as secondary causes of red cell distribution width elevation owing to limited data available in the eICU-CRD.

Conclusion

This study showed that red cell distribution width can predict hospital mortality in patients with sepsis and has prognostic importance beyond the use of SOFA and APACHE IV scores. However, further prospective cohort studies must be conducted to validate the findings of this study.

Footnotes

Acknowledgements

We thank the team of the Laboratory for Computational Physiology from the Massachusetts Institute of Technology (LCP-MIT) who work to keep the MIMIC-III and eICU databases available. We also thank Alistair Johnson, who contributed a large amount of source code on GitHub.

Declaration of conflicting interest

The authors declare that there is no conflict of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

ORCID iDs

Yide Li

Liang Luo

References

Vajpayee

Graham

Bem

Basic examination of blood and bone marrow.

Henry’s clinical diagnosis and management by laboratory methods 2011; 22: 509–535.

Salvagno

Sanchis-Gomar

Picanza

, et al. Red blood cell distribution width: A simple parameter with multiple clinical applications. Crit Rev Clin Lab Sci 2015; 52: 86–105. DOI: 10.3109/10408363.2014.992064.

Uyarel

Ergelen

Cicek

, et al. Red cell distribution width as a novel prognostic marker in patients undergoing primary angioplasty for acute myocardial infarction. Coron Artery Dis 2011; 22: 138–144. DOI: 10.1097/MCA.0b013e328342c77b.

Lee

Park

Cho

, et al. Red cell distribution width as a novel predictor for clinical outcomes in patients with paroxysmal atrial fibrillation. Europace 2015; 17: ii83–ii88. DOI: 10.1093/europace/euv210.

Xanthopoulos

Giamouzis

Melidonis

, et al. Red blood cell distribution width as a prognostic marker in patients with heart failure and diabetes mellitus. Cardiovasc Diabetol 2017; 16: 81. DOI: 10.1186/s12933-017-0563-1.

Zalawadiya

Veeranna

Panaich

, et al. Red cell distribution width and risk of peripheral artery disease: analysis of National Health and Nutrition Examination Survey 1999-2004. Vasc Med 2012; 17: 155–163. DOI: 10.1177/1358863x12442443.

Kim

Song

, et al. Red blood cell distribution width is associated with poor clinical outcome in acute cerebral infarction. Thromb Haemost 2012; 108: 349–356. DOI: 10.1160/th12-03-0165.

Siegler

Marcaccio

Nawalinski

, et al. Elevated Red Cell Distribution Width is Associated with Cerebral Infarction in Aneurysmal Subarachnoid Hemorrhage. Neurocrit Care 2017; 26: 26–33. DOI: 10.1007/s12028-016-0306-2.

Abul

Ozsu

Korkmaz

, et al. Red cell distribution width: a new predictor for chronic thromboembolic pulmonary hypertension after pulmonary embolism. Chron Respir Dis 2014; 11: 73–81. DOI: 10.1177/1479972314525057.

10.

Hirahara

Tajima

Fujii

, et al. Comprehensive Analysis of Red Blood Cell Distribution Width as a Preoperative Prognostic Predictor in Gastric Cancer. Anticancer Res 2019; 39: 3121–3130. DOI: 10.21873/anticanres.13448.

11.

Prognostic role of RDW in hematological malignancies: a systematic review and meta-analysis.

Cancer Cell Int 2018; 18: 61. DOI: 10.1186/s12935-018-0558-3.

12.

Zhang

Lin

, et al. Association Between Red Blood Cell Distribution Width and All-cause Mortality in Chronic Kidney Disease Patients: A Systematic Review and Meta-analysis. Arch Med Res 2017; 48: 378–385.

13.

Epstein

Nasser

Mashiach

, et al. Increased red cell distribution width: A novel predictor of adverse outcome in patients hospitalized due to acute exacerbation of chronic obstructive pulmonary disease. Respir Med 2018; 136: 1–7.

14.

Torio

Moore

BJ.

National Inpatient Hospital Costs: The Most Expensive Conditions by Payer, 2013: Statistical Brief #204. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD): Agency for Healthcare Research and Quality (US), 2016.

15.

Luhr

Cao

Söderquist

, et al. Trends in sepsis mortality over time in randomised sepsis trials: a systematic literature review and meta-analysis of mortality in the control arm, 2002-2016. Crit Care 2019; 23: 241. DOI: 10.1186/s13054-019-2528-0.

16.

Fernando

Tran

Taljaard

, et al. Prognostic Accuracy of the Quick Sequential Organ Failure Assessment for Mortality in Patients With Suspected Infection: A Systematic Review and Meta-analysis. Ann Intern Med 2018; 168: 266–275. DOI: 10.7326/m17-2820.

17.

Sathaporn

Khwannimit

Validation the performance of New York Sepsis Severity Score compared with Sepsis Severity Score in predicting hospital mortality among sepsis patients.

J Crit Care 2019; 53: 155–161.

18.

Kim

Hur

Moon

, et al. Multi-marker approach using procalcitonin, presepsin, galectin-3, and soluble suppression of tumorigenicity 2 for the prediction of mortality in sepsis. Ann Intensive Care 2017; 7: 27. DOI: 10.1186/s13613-017-0252-y.

19.

Zhu

Zhou

, et al. Early increases in serum FGF21 levels predict mortality of septic patients. Cytokine 2018; 111: 428–433.

20.

Mikacenic

Price

Harju-Baker

, et al. A Two-Biomarker Model Predicts Mortality in the Critically Ill with Sepsis. Am J Respir Crit Care Med 2017; 196: 1004–1011. DOI: 10.1164/rccm.201611-2307OC.

21.

Jones

Puskarich

MA.

Is lactate the “Holy Grail” of biomarkers for sepsis prognosis?

Crit Care Med 2009; 37: 1812–1813. DOI: 10.1097/CCM.0b013e3181a09487.

22.

Ellahony

El-Mekkawy

Farag

MM.

A Study of Red Cell Distribution Width in Neonatal Sepsis. Pediatr Emerg Care 2017. DOI: 10.1097/pec.0000000000001319.

23.

Martin

Desai

Nanavati

, et al. Red cell distribution width and its association with mortality in neonatal sepsis. J Matern Fetal Neonatal Med 2019; 32: 1925–1930. DOI: 10.1080/14767058.2017.1421932.

24.

Wang

Kao

, et al. Red blood cell distribution width is associated with mortality in elderly patients with sepsis. Am J Emerg Med 2018; 36: 949–953.

25.

Huda

Karim

Mahmud

, et al. Use of Acute Physiology and Chronic Health Evaluation (APACHE)-II and Red Cell Distribution Width (RDW) for Assessment of Mortality of Patients with Sepsis in ICU. Mymensingh Med J 2017; 26: 585–591.

26.

Özdoğan

Karateke

Özyazıcı

, et al. The predictive value of red cell distribution width levels on mortality in intensive care patients with community-acquired intra-abdominal sepsis. Ulus Travma Acil Cerrahi Derg 2015; 21: 352–357. DOI: 10.5505/tjtes.2015.26737.

27.

Lorente

Martín

Abreu-González

, et al. Red blood cell distribution width during the first week is associated with severity and mortality in septic patients. PLoS One 2014; 9: e105436. DOI: 10.1371/journal.pone.0105436.

28.

McShea

Holl

Badawi

, et al. The eICU research institute-a collaboration between industry, health-care providers, and academia. IEEE Eng Med Biol Mag 2010; 29: 18–25.

29.

Singer

Deutschman

Seymour

, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016; 315: 801–810.

30.

Johnson

Deliberato

Neto

AS.

alistairewj/mechanical-power: Mechanical Power v1.0.0. 2019. DOI: 10.5281/zenodo.3263855.

31.

Pencina

D'Agostino

Sr D'Agostino

Jr , et al. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008; 27: 157–172.

32.

Pencina

D'Agostino

Sr and Steyerberg

EW.

Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers.

Stat Med 2011; 30: 11–21. DOI: 10.1002/sim.4085.

33.

Robin

Turck

Hainard

, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 2011; 12: 77.

34.

Han

Zhang

Yan

, et al. Red blood cell distribution width predicts long-term outcomes in sepsis patients admitted to the intensive care unit. Clin Chim Acta 2018; 487: 112–116. DOI: 10.1016/j.cca.2018.09.019.

35.

Gao

Liu

, et al. Assessment of Apoptosis Inhibitor of Macrophage/CD5L as a Biomarker to Predict Mortality in the Critically Ill With Sepsis. Chest 2019. DOI: 10.1016/j.chest.2019.04.134.

36.

Mebazaa

Geven

Hollinger

, et al. Circulating adrenomedullin estimates survival and reversibility of organ failure in sepsis: the prospective observational multinational Adrenomedullin and Outcome in Sepsis and Septic Shock-1 (AdrenOSS-1) study. Crit Care 2018; 22: 354. DOI: 10.1186/s13054-018-2243-2.

37.

Chung

Chen

Chuang

, et al. Increased Plasma Acetylcarnitine in Sepsis Is Associated With Multiple Organ Dysfunction and Mortality: A Multicenter Cohort Study. Crit Care Med 2019; 47: 210–218. DOI: 10.1097/ccm.0000000000003517.

38.

Lippi

Targher

Montagnana

, 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: 10.1043/1543-2165-133.4.628.

39.

De Gonzalo-Calvo

De Luxán-Delgado

Rodríguez-González

, et al. Interleukin 6, soluble tumor necrosis factor receptor I and red blood cell distribution width as biological markers of functional dependence in an elderly population: a translational approach. Cytokine 2012; 58: 193–198. DOI: 10.1016/j.cyto.2012.01.005.

40.

Fornal

Wizner

Cwynar

, et al. Association of red blood cell distribution width, inflammation markers and morphological as well as rheological erythrocyte parameters with target organ damage in hypertension. Clin Hemorheol Microcirc 2014; 56: 325–335. DOI: 10.3233/ch-131745.

41.

Ghimire

Shakya

Shrestha

, et al. The utility of red cell distribution width to predict mortality of septic patients in a tertiary hospital of Nepal. BMC Emerg Med 2020; 20: 43. DOI: 10.1186/s12873-020-00337-8.

42.

Uffen

Oomen

De Regt

, et al. The prognostic value of red blood cell distribution width in patients with suspected infection in the emergency department. BMC Emerg Med 2019; 19: 76. DOI: 10.1186/s12873-019-0293-7.