Red cell distribution width is a potent prognostic parameter for in-hospital and post-discharge mortality in hospitalized coronavirus disease 2019 patients: a registry-based cohort study on 3941 patients

Aim To investigate clinical and prognostic associations of red cell distribution width (RDW) in hospitalized coronavirus disease 2019 (COVID-19) patients. Methods We retrospectively analyzed the records of 3941 consecutive COVID-19 patients admitted to a tertiary-level institution from March 2020 to March 2021 who had available RDW on admission. Results The median age was 74 years. The median Charlson comorbidity index (CCI) was 4. The majority of patients (84.1%) on admission presented with severe or critical COVID-19. Patients with higher RDW were significantly more likely to be older and female, to present earlier during infection, and to have higher comorbidity burden, worse functional status, and critical presentation of COVID-19 on admission. RDW was not significantly associated with C-reactive protein, occurrence of pneumonia, or need for oxygen supplementation on admission. During hospital stay, patients with higher RDW were significantly more likely to require high-flow oxygen therapy, mechanical ventilation, intensive care unit, and to experience prolonged immobilization, venous thromboembolism, bleeding, and bacterial sepsis. Thirty-day and post-hospital discharge mortality gradually increased with each rising RDW percent-point. In a series of multivariate Cox-regression models, RDW demonstrated robust prognostic properties at >14% cut-off level. This cut-off was associated with inferior 30-day and post-discharge survival independently of COVID-19 severity, age, and CCI; and with 30-day survival independently of COVID severity and established prognostic scores (CURB-65, 4C-mortality, COVID-gram and VACO-index). Conclusion RDW has a complex relationship with COVID-19-associated inflammatory state and is affected by prior comorbidities. RDW can improve the prognostication in hospitalized COVID-19 patients.

Coronavirus disease 2019 , caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), is a systemic infectious disease usually presenting with fever and respiratory symptoms (1). Although the most frequent serious manifestation of COVID-19 is pneumonia, the disease has been associated with cardiovascular, neurological, and gastrointestinal symptoms (2). Systemic inflammatory response mediated by high interleukin-6 concentrations induced by SARS-CoV-2 infection is associated with more severe clinical presentation, respiratory deterioration, and death (3,4). The presence of prior chronic comorbidities substantially affects the survival of COVID-19 patients (1).
Anisocytosis, ie, unequal red blood cells (RBC) size, is a sensitive marker of distress in erythropoiesis or RBC destruction. It can be induced by various metabolic and inflammatory stimuli, nutrient deficiencies, infections, spleen disorders, and specific drugs interfering with RBC production (5). Anisocytosis can be quantified as a coefficient of variation of mean cell volume termed red blood cell distribution width (RDW), which is obtained by automatic cell counters. Higher RDW levels have recently gained attention as they are uniformly associated with unfavorable presentation and inferior outcomes in many chronic metabolic and malignant diseases (6)(7)(8)(9)(10)(11)(12). More severe clinical presentation and higher mortality rates were also found in COVID-19 patients with higher RDW levels (13)(14)(15)(16). However, an association of RDW with other clinical outcomes in hospitalized COVID-19 patients, as well as the relationship with increased mortality in the context of other established prognostic scores, are not well defined. Thus, we aimed to investigate clinical and prognostic significance of RDW in a large cohort of hospitalized COVID-19 patients from our institution. We hypothesized that RDW was associated with more severe COVID-19 on admission and higher death rate.

PAtIENtS AND MEtHODS
We retrospectively analyzed the records of 3941 patients admitted to University Hospital Dubrava because of acute COVID-19 infection from March 2020 to March 2021. We included only patients who had available data on RDW on admission. During the pandemic, our institution was repurposed into a tertiary COVID-19 regional center treating patients with the most serious COVID-19 clinical presentation and COVID-19 patients with comorbidities who required immediate hospital level of care. All patients had a positive polymerase chain reaction or rapid antigen COVID-19 test before hospital admission. Patients were treated according to the contemporary guidelines. Only index hospital ad-missions for acute COVID-19 were investigated. Data were obtained from the hospital registry thorough an analysis of electronic and written medical records of 4014 COVID-19 patients. Seventy-three patients were excluded due to lack of available RDW data on admission. The study was approved by the University Hospital Dubrava Review Board (2021/2503-04).
RDW on admission was expressed as a coefficient of variation (%) of mean corpuscular volume (MCV) as reported by Advia 2120i automated cell counter (Siemens Medical Solutions Diagnostics Pte Ltd, Swords, Ireland). COVID-19 severity at admission was graded according the World Health Organization (WHO) recommendations and national guidelines (17,18) as mild, moderate, severe, and critical. Comorbidities, assessed as individual entities, were summarized by using the Charlson comorbidity index. Mortality and other clinical outcomes were assessed from the start of hospital stay.
Modified early warning score (MEWS) was used to quantify COVID-19 symptoms. Eastern Cooperative Oncology Group (ECOG) classification was used to estimate functional status on admission. Confusion, urea, respiratory rate, blood pressure, and 65 years of age (CURB-65); International Severe Acute Respiratory and emerging Infections Consortium 4C mortality score; COVID-gram; and Veterans Health Administration COVID-19 (VACO) index were used as prognostic risk scores. Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula was used for the calculation of estimated glomerular filtration rate (eGFR) .
The primary objective was to assess the clinical associations of RDW in the context of acute COVID-19 and its prognostic properties for investigated clinical outcomes. The secondary objectives were to assess the relationship with baseline clinical characteristics (both demographic and COVID-19-related), laboratory and radiologic features, and to establish appropriate optimized RDW cut-off levels to discriminate patient groups according to the risk of unwanted outcomes (30-day mortality, need for high flow oxygen therapy, need for mechanical ventilation, need for intensive care unit, arterial thromboses, venous thromboses, bleeding complications, bacterial sepsis, six-month post hospital discharge mortality, hospital-readmission).

Statistical methods
The normality of distribution of numerical variables was assessed with the Shapiro-Wilk test. The results 2). The optimal cut-off level for in-hospital and post-discharge survival discrimination defined by the ROC curve analysis was >14%. Patients' characteristics stratified at this level are presented in Table 1. Patients with higher RDW were significantly more likely to be older, to be female, to present earlier in the disease course, and to have a worse functional status on admission. Patients with higher RDW were more likely to be admitted to hospital due to an acute medical or surgical condition as a leading reason and were less likely to be admitted due to pneumonia itself. Although patients with higher and lower RDW had a similar frequency of radiological pneumonia and need for oxygen supplementation, patients with higher RDW were more likely to have critical severity of symptoms and to have other infection on admission (P < 0.05 for all analyses).
Charlson comorbidity index was significantly higher in patients with higher RDW, with higher RDW being associated with arterial hypertension, diabetes mellitus, hyperlipoproteinemia, chronic heart failure, atrial fibrillation, coronary artery disease, history of cerebrovascular insult, history of myocardial infarction, chronic kidney disease, chronic obstructive pulmonary disease, chronic liver disease, liver cirrhosis, active malignancy, metastatic malignancy and dementia (P < 0.05 for all analyses).
RDW was not significantly associated with C-reactive protein but was significantly associated with higher interleukin-6, which gradually increased with rising RDW percent points (P < 0.001 for trend of increase and P = 0.003 for overall difference in interleukin-6 between the categories) (Figure 1). Higher RDW was significantly associated with higher white blood cells, lower hemoglobin, lower MCV, lower mean corpuscular hemoglobin concentration, higher reticulocyte count, lower platelets, higher procalcitonin, lower ferritin, higher D-dimer, lower eGFR, lower lactate dehydrogenase, lower aspartate aminotransferase, lower alanine aminotransferase, lower gamma-glutamyl aminotransferase, higher alkaline phosphatase, higher total bilirubin, lower albumin, and lower prothrombin time (P < 0.05 for all analyses).
The associations of RDW with clinical outcomes during and after hospital stay Associations of RDW with clinical outcomes are presented in Table 2. The median length of hospital stay was 10 days and did not significantly differ between patients with lower and higher RDW. Although patients with RDW>14% and ≤14% (cut-off optimized for survival discrimination) had similar frequencies of intensive care unit use, high flow oxygen therapy and mechanical ventilation use, arterial and venous thrombotic events, and bacteriemia, we estab-   gories HR = 1.7, P < 0.001; COVID gram categories HR = 2.67, P < 0.001; VACO index categories HR = 3.51, P = 0.002).

DISCuSSION
Our study, the largest single-institution cohort of hospitalized COVID-19 patients investigated so far, showed a gradual worsening of 30-day and post-hospital discharge survival with each rising percent point of RDW on admission.
In addition, our study was first to investigate the associations of RDW with various clinical outcomes and to assess the prognostic role of RDW in the context of established COVID-19 prognostic scores. We demonstrated that RDW possessed independent prognostic properties and had a good potential for improvement of prognostication of hospitalized COVID-19 patients.
The main cause of RDW elevation varies from patient to patient with the same disease. This cause is hard to determine since it can be an impaired function of a number of organ systems (5,19). Nevertheless, RDW seems to be a universal marker of biological fitness as it reflects unwanted clinical outcomes in all diseases where it was investigated. Our data show that RDW is profoundly affected by the presence of prior comorbidities and age, with patients with higher RDW being more likely to have chronic metabolic comorbidities, more unfavorable cardiovascular risk profile, chronic renal disease, chronic obstructive pulmonary disease, liver cirrhosis, active and metastatic malignancy, and dementia. Interestingly, patients with higher RDW significantly grouped only among patients with critical COVID-19 on admission, and no similar association of RDW was observed with either radiological presence of pneumonia or the need for oxygen supplementation. Thus, the mentioned chronic comorbidities might predispose to higher functional impairment and critical clinical presentation observed in our study. Despite similar frequency of pneumonia and respiratory insufficiency at baseline, patients with higher RDW were significantly more likely to develop respiratory deterioration during hospital stay and to need intensive care and more advanced respiratory support.
RDW has a complex relationship with COVID-19-associated inflammatory state. Despite a significant association with higher interleukin-6 concentrations, which increased with an increase in RDW percent points, RDW was not significantly associated with CRP concentration on admission.
In addition, RDW was significantly associated with higher WBC, higher procalcitonin, higher D-dimer, lower albumin, lower prothrombin time, and higher bilirubin on admission, but also with lower liver transaminases, lower ferritin, and lower lactate dehydrogenase. This ambiguous inflammatory profile associated with higher RDW further implies that RDW elevation is affected by mechanisms other than acute inflammatory state associated with COVID-19. The magnitude of RDW elevation occurring in acute COVID-19 infection is unknown, and high RDW, and consequently higher hazard of death, might represent a higher degree of chronic inflammation, a higher comorbidity burden, and a predisposition to worse clinical outcomes predating acute COVID-19. It was speculated whether patients with higher RDW might present later during disease course and thus represent more advanced inflammation (20). Our data suggest the opposite -patients with higher RDW were more likely to present earlier during disease course, and RDW might be elevated because of acutization of chronic medical conditions other than pneumonia due to active COVID-19 infection.
A number of studies and several large meta-analyses have confirmed potent prognostic properties of RDW for the survival of COVID-19 patients (13)(14)(15)(16). Our study provides important novelty as it comprehensively assesses several important clinical outcomes and investigates how different levels of RDW increase reflect on survival, both in univariate and multivariate context controlling for clinically relevant parameters. RDW on admission not only predicts short-and long-term survival but also predicts a wide range of unwanted clinical outcomes, such as the need for more intensive respiratory support, prolonged immobilization, deep venous thrombosis, bacterial sepsis, and bleeding. The ability of RDW to discriminate between different outcomes changes when using different cut-off levels. Lower cut-off values optimally predicted complications of infectious diseases (bacterial sepsis and tendency for COVID-19 deterioration), whereas higher values optimally predicted venous thrombotic events. Although pronounced anisocytosis reflects a worse prognosis for thrombotic and septic outcomes in COVID-19 patients, the precise mechanism underlying these processes is not completely clear (compared with the intuitive role of some other blood components such as platelets in the same events [21]). Considering survival as the most important outcome, long-term and short-term prognostic properties of RDW are independent from COVID-19 severity on admission, age, and comorbidities. Additionally, RDW covers different parts of prognostic spectrum than WHO-defined COVID-19 severity categories: CURB-65, 4C mortality, COVID gram, and VACO index. Therefore, RDW might be very useful in assessing the prognosis of hospitalized COVID-19 patients. Our results need to be validated by future studies in independent samples and by using RDW to further extend prognostic scores.
Limitations of our study are single-center experience, retrospective study design, and the lack of longitudinal assessment of RDW values. Our results are representative of tertiary level institution, with a majority of patients being elderly, having severe or critical COVID-19, and having high comorbidity burden. The main strength of our study is a large sample of patients who were uniformly exposed to standardized diagnostic and therapeutic procedures. This gives the study high statistical power to assess clinical and prognostic associations of RDW in the context of hospitalized COVID-19 patients.
In conclusion, RDW has a complex relationship with COVID-19-associated inflammatory state and is profoundly affected by prior comorbidities. It has a good potential for improvement of prognostication of hospitalized COVID-19 patients.