On admission hemoglobin and albumin, as the two novel factors associated with thrombosis in COVID-19 pneumonia

3-0.86) were independently correlated with thrombosis due to COVID-19. Other factors, including demographic, infection severity, laboratory and in-hospital characteristics, were not significantly associated with thrombotic events. Conclusion: Based on this study’s findings, hemoglobin and albumin levels were the independent factors associated with the thrombotic events in COVID-19 patients.


Introduction
The unrelenting storm of coronavirus disease of 2019 (COVID- 19), which has emerged since December 2019, has turned into a crucial health matter of the globe.This infection has involved over 90 million people worldwide and led to death in approximately two million cases.This pandemic's exact pathophysiology is unknown still and scientists are searching for efficient management approaches to this infection (1).
COVID-19 is an acute complex systemic disorder which presentation varies from mild influenza-like symptoms to catastrophic conditions by interstitial pneumonia progressing to acute respiratory distress syndrome (ARDS), sepsis and multi-organ failure (2).Increasing evidence is going on regarding hypercoagulable states among those infected with COVID-19, also known as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2).The epidemiological studies have stated a wide range of 7.7%-49% for thrombotic events incidence in COVID-19 infected patients, including deep vein thrombosis (DVT), pulmonary thromboembolism (PTE), ischemic cerebrovascular accident (CVA), acute myocardial infarction (AMI) and arterial thrombosis (3)(4)(5).An increase in the inflammatory factors, endothelial dysfunction, and thromboinflammation propagated by angiotensin-converting enzyme-2 are the potential factors presumed to play a role in coagulopathy pathogenesis (6).
Although numerous studies have been conducted to assess the prevalence, etiology and features of thromboembolic events due to SARS-CoV-2, the knowledge in this regard is limited (7).

Objectives
The current study aims to clarify the correlation between thromboembolic events incidence in COVID-19 and the patient, infection, and in-hospital related characteristics.

Study design
The current case-control study has been conducted on 243 patients admitted at Amin and Alzahra hospitals (affiliated to Isfahan University of Medical Sciences) due to SARS-CoV-2 from May to June 2020.The included patient consisted of 83 cases with thrombotic events and 160 individuals as controls with no thrombotic event, matched according to presence of COVID-19 pneumonia.
First, the study protocol was explained to the patients, or their legal guardians; then, the participants became aware of the study and signed the written consent accordingly.
The case group was selected from the patients with any thrombotic event, like DVT, ischemic CVA, PTE or AMI whose COVID-19 infection was approved by a positive polymerase chain reaction (PCR) test and also presence of signs of COVID-19 pneumonia in the imaging.The inclusion of the patients was conducted employing a convenience sampling.
Additionally, all patients in the control group had positive results for PCR test and radiographic signs of COVID-19 pneumonia, but with no thrombotic event.The control group was matched to the cases with a 2:1 proportion.
The previous history of coagulopathies and a recent history of thrombotic events (DVT, PTE, CVA or AMI) within a month prior to the hospital admission were determined as the exclusion criteria.

Diagnosis of thrombotic events
DVT diagnosis was made according to the patients' clinical manifestations (8) and confirmed through Doppler ultrasonography.To make a PTE diagnosis, computed tomographic pulmonary angiography was conducted for those referred with the symptoms compatible with PTE (9).Acute MI (myocardial infarction) was determined as STsegment elevation myocardial infarction (STEMI) or non-STEMI.Typical chest pain for cardiac ischemia plus an elevation in highly-sensitive troponin levels as a sensitive and specific cardiac biomarker was defined as acute MI.STEMI in two or more echocardiogram leads indicating the involvement of a particular epicardial territory or newonset left bundle branch block; otherwise, assumed as the non-STEMI.Hemiplegia, facial hemiparesis or dysarthria with a CT (computed tomography) scan compatible with an ischemic CVA was the CVA determinants.
All patients received the required COVID-19 treatment and anticoagulant according to Iran's national guidelines.

Data collection
The demographic data including age, gender and smoking and full medical history were imported into the study checklist.
Anticoagulation in the studied group was classified as no anticoagulation, prophylactic, intermediate and therapeutic dose.Type of anticoagulation prior to thrombotic event was considered in the case group.Prophylactic doses were 5000 IU of subcutaneous unfractionated heparin (UFH) (twice daily) or 40 mg of subcutaneous low molecular weight heparin (LMWH) (once daily) or 10 mg of oral rivaroxaban (daily).Intermediate doses included 7500 IU of subcutaneous UFH (twice daily) or 60 mg of subcutaneous LMWH (daily) or 40 mg of subcutaneous LMWH (twice daily).The therapeutic doses were determined as 80 IU/kg of IV UFH followed by infusion of 18 IU/kg/h or 1 mg/kg of subcutaneous LMWH (twice daily) or 15 mg of oral rivaroxaban (twice daily).
The hospitalization-related characteristics were gathered too.These data were mobility during the hospitalization [complete bed rest (CBR) or relative bed rest (RBR)], type of ventilation (non-invasive ventilation/ intubation), intensive care unit (ICU) admission requirement, length of hospitalization, the time between hospitalization to ICU admission and death incidence.

Data analysis
The obtained data were entered into the Statistical Package for Social Sciences (SPSS; version 14.0, SPSS Inc., Chicago, IL, USA).The descriptive data were presented in mean, standard deviation, median and range for the continuous variable and also absolute numbers and percentages for categorical variables.The chi-square test or Fisher's exact test was utilized to compare the categorical variables between the groups.The continuous variables were compared using a t-test.Binary logistic regression analysis was applied to estimate the odds ratio and find the association between the assessed factors and thrombotic events in the crude model and adjusted model for age and gender.A P value of less than 0.05 was considered as a significant level.
The thrombotic events occurred in 51 patients (61.5%) on admission or within the first two days of hospitalization.The mean period between COVID-19 symptoms onset and a thrombotic event incidence was 8.2 ± 8.6 days (median; 7 days).Forty-three (51.8%) of the cases required ICU admission, among which thrombosis occurred in 26 cases (60.5%).Further information is presented in Table 1.
The gender distribution of thrombotic events revealed insignificant differences according to the type of the events; however, the comparison of the cases with controls revealed a significantly higher prevalence of females among the case group (P = 0.025).None of the comorbid conditions were associated with thrombotic events (P > 0.05).The patients' mobility and anticoagulation status were remarkably different between the cases and control (P < 0.0001).It is also worth noting that all the cases with CVA and arterial thrombosis had not received any form of anticoagulation (P < 0.001).Detailed information in terms of the comparison between cases and controls is demonstrated in Table 2.
As illustrated in Table 3, the intubation requirement rate was significantly more in CVA cases than other groups of patients with thrombotic events (P = 0.048).The most prolonged period of hospitalization belonged to the cases with arterial thrombosis, whereas the shortest period between hospitalization to ICU admission was found in CVA cases (P = 0.044).
Table 4 demonstrates the hemodynamic and laboratory characteristics of the cases versus controls.The mean oxygen saturation in the case group was remarkably lower than controls (P = 0.031); however, the comparison among the groups regarding the categorized oxygen saturation revealed an insignificant difference (P = 0.768).
The laboratory evaluations revealed significant differences between the groups in terms of PMN count, Hb level, PLT count, D-dimer, albumin, and LDH levels (P < 0.05).
The comparison of troponin level regardless of the thrombotic event type was accompanied by significantly higher levels in the case group (P = 0.027); however, after

Sadeghi S et al
eliminating those with AMI, the two groups were similar (P = 0.607).According to Table 5, oxygen saturation, Hb and albumin levels were the only markers inversely associated with thrombotic events.

Discussion
In the current case-control study, we tried to assess the clinical, laboratory and in-hospital characteristics of the patients with thrombotic events following a SARS-CoV-2 infection and find the predisposing factors that make a person prone to thrombotic events.Over 60% of the cases admitted in ICUs, experienced the thrombotic event after ICU admission.There are several studies representing the high incidence of these events in critically ill patients too (3,5).Nevertheless, 61.5% of the COVID-19 infected patients had a thrombotic event on-admission or presented it within the first two days.A paucity of knowledge is available regarding the characteristics of COVID-19 for increasingly developing these events.
The studies on general populations insisted on the role of medical conditions including hypertension, diabetes mellitus, chronic pulmonary disorders, chronic renal disorders, and active malignancy on thrombotic events incidence (10), which has been notified in patients with SARS-CoV-2 (3).However, none of the comorbid conditions was remarkably different between cases and controls in this study.We assume that better insight may be provided in larger populations.Among the hemodynamic parameters, oxygen saturation was found the only independent factor associated with the thrombotic events.SARS-CoV-2 pneumonia can lead to improper oxygenation due to acute respiratory distress itself.In addition, inappropriate respiration disables a person from exertion and is causative for immobilization, as well (11).It should be noted that approximately 63% of the cases were admitted due to PTE, a condition accompanied by a decrease in oxygen saturation and CVA, which required intubation due to a similar condition.Therefore, a two-sided association should be considered between the oxygen saturation and thrombotic events.
We found on-admission Hb level as an inverse factor associated with thrombotic events, which has not been noted previously.Furthermore, most of the studies represented an insignificant difference between those with thromboembolic events and the control groups (10)(11)(12).It is hypothesized that low Hb level, which represents low blood viscosity, reduces stress formation on the endothelial bed of the vessels and, in turn, inhibits the function of the anti-thrombotic mechanism (13).Therefore, patients with lower Hb levels are prone to thrombotic events due to impaired endothelial function as one of the Virchow triad (14).Recent investigations on COVID-19 have notified increased risk of thromboembolism development due to diverse factors such as the imbalance between oxidative and anti-oxidative processes, accumulation of proinflammatory and pro-thrombotic factors and endothelial dysfunction (15).It seems that mentioned factors accompanying low levels of Hb reinforce the cascade of thrombosis formation.
Contrary to most of the investigations insisting on the standalone role of low-Hb levels on thrombosis inhospital and long-term adverse events (16)(17)(18), studies in the literature have rarely evaluated the predictive value of Hb level for thrombotic events developing.Yamashita et al reported exceptional lower levels of Hb among the patients with venous thromboembolism and in accordance with other presentations, found a predictive role for this index for in-hospital outcomes, long-term adverse events and all-cause mortality (19).A similar reverse association was presented by Can et al (13).On the other hand, Kalra et al hesitate to confirm whether low Hb has a role in cardiovascular events and/or is a marker of comorbidities (17).This poor correlation was noted for thromboembolism accordingly (20).Therefore, the attitude about Hb's role in the development of thrombotic events is not unanimous and requires further investigations.D-dimer was shown to be directly correlated with thrombotic events in our study.It is traditionally known as a marker of inflammation, coagulation activation and hyperfibrinolysis.The studies assessing these events in COVID-19 patients insisted on the elevated levels of this marker among those who experience a thrombotic event during a course of COVID-19 infection (12,21).In addition, Lodigiani et al represented a rapid increase in D-dimer levels among non-survivors of thrombosis in COVID-19 patients (4).Helms and colleagues noted that a rapid rising D-dimer level despite anticoagulation is a reflection of clot formation and a probable thrombotic event.They even recommended imaging assessments for patients with a sudden increase in D-dimer along with the deterioration of the clinical course of the disease (22).The significance of D-dimer is to the extent that guidelines for anticoagulant therapy in this infection have mentioned elevated D-dimer as a factor for high dose anticoagulant treatment (23).
Plasma albumin is the other factor found inversely associated with thrombosis in COVID-19 infection.To the best of our knowledge, despite all the notifications regarding the hypercoagulable state in SARS-CoV-2 infection, albumin level has not been well studied.Zhang et al determined the proportion of fibrinogen-to-albumin as a determinant for the risk of thrombosis development (24).
The increased probability of thrombosis in hypoalbuminemia has been well-established in chronic conditions such as nephrotic syndrome or cirrhosis (25); however, most of our patients were gathered from the general population infected by COVID-19.Nevertheless, studies on the general population have stated a significant association between albumin levels with venous thromboembolism incidence (26) and its severity (27).Further studies on other thrombotic events presented similar correlations (28,29).
To evaluate the role of low albumin levels as a risk factor for thrombosis formation, Mirsaeidi et al concluded that it acts as a representative of inflammation status (30).This theory has been confirmed by presenting an inverse correlation between serum albumin level with CRP and estimated erythrocyte sedimentation (28,27).In other studies, an antagonistic role of oxidation, stagnant, thrombosis and leukocyte adhesion has been considered for albumin (31).Further investigations, particularly in the critical group of COVID-19 patients, are required.
According to the findings of this study in terms of a significant association between anticoagulant prophylaxis and thrombosis formation and due to the recommendations of the other investigations of SARS-CoV-2 infected patients, prophylactic anticoagulation is strongly recommended, particularly in critical patients (32).It should also be noted that high number of cases in "no anticoagulation" group may be due to this fact that thrombotic events occurred on admission and no

Table 1 .
The characteristics of the thrombotic cases (n=83) * Thrombosis diagnosis within less than two days after admission (n=5) was defined as on admission thrombosis.

Table 2 .
The comparison of demographic and medical characteristics between cases and controls

Table 3 .
The comparison of hospitalization characteristics between cases and controls PTE, pulmonary thromboembolism; DVT, deep vein thrombosis; AMI, acute myocardial infarction; CVA, cerebrovascular accident; NIV, Non-invasive ventilation; ICU, intensive care unit.χ 2 test for categorical variable and independent t test for a continuous variable were significant if P < 0.05.

Table 4 .
The comparison of on admission maximum-values of hemodynamic and laboratory parameters between the cases and controls χ 2 test for categorical variable and independent t test (or median test a ) for a continuous variable were significant if P value < 0.05.
* b Cases without MI (n = 66) were compared with control.

Table 5 .
Logistic regression assessments of related factor to thrombosis incidence