Clinical, Biochemical, and ATR-FTIR Spectroscopic Parameters Associated with Death or Survival in Patients with Severe COVID-19

Te wide range of symptoms of the coronavirus disease 2019 (COVID-19) makes it challenging to predict the disease evolution using a single parameter. Terefore, to describe the pathophysiological response to SARS-CoV-2 infection in hospitalized patients with severe COVID-19, we compared according to survival or death, the sociodemographic and clinical characteristics, the biochemical and immunological attenuated total refection-Fourier transform infrared (ATR-FTIR) spectra from saliva samples and their correlation with chemometric fndings. Herein, we demonstrate that ATR-FTIR spectroscopy allows the description of the events related to cell damage, such as lipids biogenesis and the secondary structure of proteins associated with lactate dehydrogenase and albumin levels. Moreover, humoral (IgM) and cellular (IFN-c , TNF-α , IL-10, and IL-6) responses were also increased in patients who died from COVID-19.


Introduction
Coronavirus disease 2019 (COVID- 19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is still an urgent public health concern due to the consistent RNA modifcations across the viral genome, provoking resistance to the vaccine, and infection-induced immunity.Moreover, acute infection is not the only outcome of COVID-19 but there is also the possibility of long-term COVID infection [1,2].
Te wide range of symptoms, complications, and outcomes generate many eforts to predict the evolution of the disease at an early stage, allow more time management and resource allocation, increase the cure rate, and decrease the case fatality rate [3][4][5].
Current knowledge of COVID-19 pathogenesis includes the direct cytopathic efects, ischemic injury, and host characteristics such as age, comorbidities, and aberrant immune responses; this might explain diferent disease presentations and severity [2,6].SARS-CoV-2 infects the host cells through the angiotensin-converting enzyme 2 (ACE2) receptor, which is on the X chromosome.It is expressed in cells and vessels of several organs, including the lung, heart, kidney, and intestine [6].Te spike protein (S) has a receptor-binding domain (RBD) in the S1 protein; this RBD interacts with the metallopeptidase domain of ACE2, triggering the endocytosis of the SARS-CoV-2 virion and exposing it to endosomal proteases.Te S2 protein leads to membrane fusion and releases the viral package into the host cytoplasm [7].CoVs display an envelope formed by a lipid bilayer derived from the host cell membranes due to the intracellular membrane, playing a pivotal role in coronavirus replications [6].
Coronaviruses can produce their proteins by attaching to the host ribosomes and generating single-stranded, negative-sense RNA (ssRNA−) that is used by the RNA polymerase to make SSRNA+, with the Golgi apparatus ssRNA+, which is packaged in the nucleocapsids to create new virion particles, which are then released by exocytosis.In this sense, infection, replication, and cell damage induce an immune response [8].
SARS-CoV-2 is recognized by several receptors such as Toll-like receptors (TLR) 3, 7, 8, and 9, RIG-I, and MDA5, which induce the type I interferon (IFN) response, provoking the formation of reactive oxidative species, calcium fux from cytoplasmic storages, protein aggregation, and the release of danger-associated patterns; at the same time, this contributes to the activation of the NLRP3 infammasome.NLRP3 infammasome induces caspase-1-dependent cleavage and release of key proinfammatory cytokines interleukin-1β (IL-1β) and IL-18 and triggers gasdermin Dmediated pyroptotic cell death.As a result of pyroptotic cell death, the enzyme lactate dehydrogenase (LDH) is released, and levels of this enzyme correlate with disease severity.Also, SARS-CoV-2 can inhibit the type I IFN responses in infected cells, leading to an imbalanced immune response [2].
Besides, SARS-CoV-2 infection and the destruction of lung cells release cytokines and prime adaptive T and B cell immune responses.A wave of proinfammatory cytokines and chemokines is related to the increase of IL-6, IFN-c, MCP1, and IP-10, which are indicators of T helper 1 (TH1) cell polarized response [7].Tese responses increase vascular permeability and adhesion molecule expression and recruit more immune cells, including neutrophils and monocytes.IL-8 recruits neutrophils, causing fuid leakage into the interstitial space and alveoli, resulting in interstitial and pulmonary edema.Tis can lead to dyspnea, impaired oxygenation, or hypoxemia.Hypoxemia stimulates chemoreceptors in the cardiopulmonary center of the brain, resulting in the development of tachypnea and tachycardia [8].
White blood cells (WBCs) and damaged endothelial cells release other infammatory mediators, including arachidonic acid metabolites, leukotrienes, and prostaglandins, which provoke fever.
Moreover, specifc humoral immunity against this virus plays an essential role in viral clearance; cytotoxic lymphocytes can eliminate infected cells, and specifc antibody responses have the potential to neutralize the SARS-CoV-2 or help cytotoxic T cells, eliminating the virus from the infected cells to control disease progression.SARS-CoV-2 antigens induce B cells from germinal centers to proliferate and diferentiate into plasma cells, triggering and releasing specifc antibodies [9].
COVID-19 patients produce one or more specifc antibodies against SARS-CoV-2; the main relevance of this response is related to serological diagnosis and prognosis prediction, such as levels of serum SARS-CoV-2-specifc IgM, IgG, and IgA.Zheng et al. showed that elevated anti-SARS-CoV-2 IgG titers correlate with the length of hospitalization and are also associated with disease severity in a time-dependent manner, as demonstrated by the delayed IgG response observed in deceased patients [9,10].
Te antibody level refects the dependence on the amide and lipid metabolisms; symptomatic patients have decreased levels of sphingosine consistent with the presence of acid ceramidase protein compared to asymptomatic patients [11].
In recent years, Fourier transform infrared (FTIR) spectroscopy has been used to diferentiate between sick and healthy people by employing tissues, cells, and body fuids, as well as in the stratifcation of diferent diseases.In this sense, FTIR spectroscopy is a promising point-of-care technology for the classifcation of COVID-19 severity or measurement of periodic changes in COVID-19 antibody levels on serum or saliva samples [11][12][13][14].
With all this information, nowadays, there are many markers of prognosis and severity such as sex, age, comorbidities, WBC count, lymphocyte count, creatinine and bilirubin levels, as well as D-dimer, and C-reactive protein (PCR), lactate dehydrogenase (LDH), and viral load.Furthermore, many of these biomarkers change during the hospitalization stay, predicting the progression to severe disease [4,10,15].
Once FTIR spectroscopy changes could be related to these biomarkers, this technique seems helpful at any time point of patient hospital admission because it shows the biochemical and immunological status.
Terefore, our work compared the pathophysiological response to SARS-CoV-2 infection in hospitalized patients with a severe disease, divided into two groups according to their clinical evolution (survival and death groups).For this purpose, we analyzed the sociodemographic and clinical characteristics, the biochemical and immunological ATR-FTIR spectra from saliva samples, and their correlation with chemometric fndings.

Study Population.
Te study was conducted at the Central Military Hospital (HCM)-SEDENA, Mexico.Te study period was from May 19, 2020, to March 1, 2021.In this research, we divided the population into survival and death groups.Te survival group was integrated with COVID-19 patients who were hospitalized and those who 2 Journal of Spectroscopy recovered from the illness, and the death group was composed of patients who died during the hospitalization.Te inclusion criteria were volunteers that accepted to participate in the study by writing informed consent, were aged over 18 years, diagnosed positive for COVID-19 by qRT-PCR test, category 3 risk factors for severe disease according to COVID-19 Treatment Guidance of Massachusetts General Hospital, and a fasting period of 8 hours.Contrarily, intubated patients and samples from patients who brushed or rinsed the oral cavity with mouthwash before sampling were excluded.

Medical Data Collection, Laboratories, and Evolution.
When the sample was collected, patients with symptoms (cough, dyspnea, headache, sputum, and fever, among others) and comorbidities such as diabetes, obesity, hypertension, and asthma were interrogated, and the obtained data were recorded as present or absent.To determine overweight and obesity, we followed the World Health Organization guidelines, which establish a body mass index (BMI) ≥ 25 for overweight and a BMI ≥ 30 for obesity [16].
At the same time, a blood sample for laboratory blood tests (hematic biometry, blood chemistry test, hepatic function test, and infammatory biomarkers) was collected.
For the hematic biometry analysis, 5 ml of complete blood was collected in EDTA (ethylenediaminetetraacetic acid) tubes.Immediately, the blood sample was gently mixed and subjected to continued agitation for 30 minutes.After that, analysis was developed using ADVIA® 360 (SIEMS) equipment.
Five millilitres of blood were collected in a serumseparating tube for the blood chemistry test.After that, the blood sample was centrifuged at 1096g for ten minutes, and the obtained serum was collected and analyzed using the automatic equipment VITROS® 5600 (Ortho Clinical Diagnosis).
To evaluate clinical evolution, days of hospital stay were used as a time scale to calculate the median time for recovery or death.

Saliva Sample Collection.
Both study groups were asked to donate 1 ml of saliva, collected in a sterile 1.5 ml microcentrifuge tube.Te samples were kept cold (4 °C) during their transfer and until their analysis in the following three hours.Personal protective equipment was used by personnel in charge of sampling and analyzing the samples.All experiments were evaluated and approved by the appropriate Ethics Committee, and the ethical standards laid down in the 1964 Declaration of Helsinki were followed.Moreover, the Institutional Human Research Ethical Committee approved the protocol and the written informed consent forms.

Saliva Sample Analysis through ATR-FTIR Spectroscopy.
Te FTIR spectra analyzed in the attenuated total refectance mode (ATR-FTIR spectra) were examined in the spectral range between 4000 and 400 cm −1 , employing an FTIR spectrometer (6600, Jasco).Te apparatus has a fxed spectral resolution of 4 cm −1 .For the sample analysis, three μl from the saliva were placed onto the ATR diamond crystal (PKS-D1F, Jasco).Subsequently, the sample was dried at room temperature for about 15 minutes to eliminate excess water until the band attributed to water was not detected [17].Te sample spectra were captured 120 times and averaged to obtain one spectrum.Each sample was analyzed in triplicate; then, the average of the spectra was obtained for the corresponding normalization and analysis.

ATR-FTIR Spectrum Analysis.
Standard normal variate (SNV) normalization was performed employing the Unscrambler X software (version 10.3, Camo).Ten, the second derivative spectra were determined by the Savitzky-Golay algorithm with fve-point windows for smoothing and the second polynomial order using the Unscrambler X software.Next, the obtained spectra were inverted by factoring by −1 [18].Ten, the mean of each population was obtained to identify relevant diferences (displacements or area).Te Origin software (version 8.5, Origin Lab Corporation) was used for getting the graphs.

Statistical Analysis.
Descriptive nonparametric survival analysis, such as the Kaplan-Meier survival curve, was used to estimate survival probability.Days of hospital stays were used as a time scale to calculate the median time to recovery.Categorical variables were expressed as rate and percentage; for quantitative variables, normality tests were applied, and these were expressed as median ± standard deviation (SD) media or interquartile range (IQR).X 2 , Mann-Whitney U test, or T-student test, was used to determine signifcant diferences between groups according to the variable type and normality of behavior.Te analysis was performed using IBM SPSS Statistics 25.0 and Graph Pad 8.0.Te median age was 55 ± 12.5 years, and most patients were male.However, it is important to mention that patients who died (16.5%) were older (63 ± 12 years), highlighting that the majority of patients that integrated into this population were over 50 years old; nevertheless, no statistical signifcance was evidenced related to sex (Table 1).

Baseline Clinical Characteristics of Patients.
Te main comorbidities were being overweight, followed by diabetes and obesity, and had an O+ blood type.Patients who died showed major smoking; however, no statistical signifcance was observed between the groups.Regarding the blood type, the most common blood type in both groups was O+, followed by A+, without evidencing statistical signifcance.Te main symptoms were dyspnea and cough; death patients showed more sputum compared with recovery patients.Te ratio of arterial oxygen partial pressure to fractional inspired oxygen (PaO 2 /FiO 2 ) was low, and dead patients had a high respiratory rate (Table 2).

Evolution of Patients.
Regarding the days at sampling and the hospital stay, no signifcant diference was evident; the Kaplan-Meier cuve showed a median of 46.3 days.Nevertheless, it is essential to mention that 75% of dead patients required intubation and showed a median of 23.38 days (Table 3).

Biochemical Characteristics of Patients by Serum and
ATR-FTIR Analyses.Te death group showed higher creatinine, urea, CRP, and LDH, and lower lymphocyte and ALT levels than the survival group (Table 4).
Once lipid biogenesis pathways are related to the virus propagation, we analyzed this process through the 2923/ 2958 (CH2/CH3) ratio, the peak at 1075 cm −1 , and the total area of lipids/total area of proteins ratio (Figure 2).Considering those above, the death group evidenced less lipid biogenesis (Figure 2(a)).Contrary, the peak at 1075 cm −1 , related to phospholipids (Figure 2(b)), showed a higher content in the death group.Finally, regarding the total area of lipids/total area of proteins ratio, we did not fnd changes (Figure 2(c)).
Te ATR-FTIR spectra analysis on the second derivative showed, in the death group, a decrease in the alpha helix of proteins and an increase in urea (Figures 3(a)-3(c)) according to the serum parameters.
Patients who died showed higher levels of leucocytes, neutrophils, and N/L ratio; and low levels of lymphocytes (Table 5).
Finally, Figures 5(a) and 5(b) show the averaged and normalized spectra, raw and in the second derivative of the cytokine's regions, respectively, where there was an increase in IFN-c, IL-10, and IL-6 in the death group compared with the survival group, which was also seen in the integrated areas of the peaks, as shown in Figure 5(c).

Discussion
Te COVID-19 pandemic remains unpredictable due to the spectrum of presentation varying from asymptomatic infection to a fulminant systemic infammatory syndrome unleashed by the cytokine storm.Risk factors for mortality in COVID-19 reported in various studies include advanced age, male gender, and some comorbidities such as diabetes and hypertension [24].
On the other hand, Dessie and Zewotir showed a mortality rate between 3.14-61.51%,with a mortality prevalence among hospitalized patients of 17.62%, similar to our population (16.5%) [25].
Moreover, this research found signifcant aged patients in the dead group; 88.2% of the deceased patients were >50 years old.Dessie and Zewotir declared that age is one of the main risk factors for death.Aging was associated with other viral infections such as MERS-CoV, SARS, and infuenza, which might be related to lower immunity levels; it is known that aging reduces T cells' and B cells' clonal diversity and the dysregulated production of type 2 cytokines that could prolong the proinfammatory immune response, contributing to the poor outcomes [25,26].
Furthermore, looking for a relationship between sex and death risk, we did not fnd any diferences; however, the most predominant sex was male, similar to Alharbi et al., who also reported a majority of male hospitalized COVID-19 patients.Terefore, the women's advantages could be related to the immune regulatory genes on the X chromosome and the higher expression of TLR7 [5,25].
Regarding the comorbidities, we found that the main comorbidities in both groups were overweight, but smoking was the most prevalent in the death group, which agrees with the meta-analysis of Brodin, who showed an association between COVID-19 progression and smoking.In addition, smoking depresses pulmonary immune function and is a risk factor for contracting other infectious diseases and 4 Journal of Spectroscopy more severe outcomes among people who become infected; also, it is reported that cigarette smoke induces expression of ACE2, which allows SARS-CoV-2 to enter cells [2,25,27].About the symptoms, it has been reported that the natural evolution of COVID-19 involved a 4-5 days incubation period, showing fever and dry cough at admission time, followed by dyspnea, arthralgias, myalgias, and headache.Herein we found that the main symptom on sampling day (8 days) was cough.It is essential to mention that eight days is when severe COVID-19 cases progress to acute respiratory distress syndrome (ARDS) [7].ARDS may lead directly to respiratory failure, which is the cause of death in 70% of fatal COVID-19 cases and 52.9% of intubated patients death (9/17 patients).
Regarding the days of hospitalization, in our study, we found a median of 18 days of hospital stay, similar to that reported by Garibaldi et al., who showed a median of 19 days (IQR, 11.9 to 30.0 days).Regarding our Kaplan-Meier analysis, including all patients, it showed a median of 46.3 days, while it was 23.38 days in intubated patients.In contrast, the same study of Garibaldi showed that the median time for intubation was 66.5 days [15].
Concerning ATR-FTIR results, shifts in the spectra are reported on serum between COVID-19 patients and standard controls, indicating protein and lipid conformation changes according to biochemical and immunological changes described below [28].Journal of Spectroscopy  6 Journal of Spectroscopy Moreover, shifts on bands at 2923 cm −1 , 2853 cm −1 , 1240 cm −1 , and 1075 cm −1 suggest the potential conformational changes in lipid components; lipid biogenesis pathways afect receptor-mediated virus entry at the endosomal cell surface and modulate virus propagation [29].
Lipid changes are corroborated by the decreased ratio of 2923/2958 (CH2/CH3) in dead patients; this ratio shows the proportion of unbranched and branched molecules of lipids and fatty acids that also are associated with cell damage like apoptosis [30].
In the same way, changes in the ratio related to CH2/ CH3 might refect an increase in the intermolecular chain disorder; the higher concentration of a long chain of CH2 suggests a weakening of the cell membrane-skeleton   Journal of Spectroscopy structure.Tat is according to the fndings of Bel'skaya, who reported a signifcant decrease in polysaturated fatty acids with the main content of CH2, such as palmitic acid.Te decrease in palmitic acid could be related to the requirement of lipid chains by the SARS-CoV-2; the process of cysteine palmitoylation regulates lipid addition to conserved cysteine residues that are located adjacent to the transmembrane sections of the virus spike and envelope proteins, which has also been demonstrated on infuenza A virus; in addition, the decrease in the activity of the 19-desaturase [30][31][32][33][34].
Likewise, in comparison with healthy patients, an increase in oleic, linoleic, and arachidonic acids in COVID-19 patients has also been reported [30,35].Oleic acid modulates the infammatory process, apoptosis, necrosis, and oxidative stress, and it stimulates iNOS and oleic acid increase by decreasing the activity of D9-desaturases on the phospholipids.Although the linoleic acid increase could be related to damage to mitochondrial dysfunction, modulated by the SARS-CoV-2 infection, higher levels of arachidonic acid are a part of a defense mechanism that can cause infammation and modulate the development of the cytokine storm [32].
About phospholipids, the band at 1075 cm −1 attributed to this component, specifcally sphingomyelin and lysolecithin (symmetric stretches from PO 2 − ), increased in the death group; other studies have also shown that in the mild, severe, and fatal COVID-19 patients, they are higher [28].
Related to the protein changes, the amide I band shifts 10 cm −1 , suggesting transitions in protein secondary structures.Tis has been reported by Zhang et al.In this research, we found that the proportion in the 1650/1626 area is related to changes in the conformational structure of proteins α-helix/β-pleated sheet; this ratio decreased in the death group.In the same way, a decrease in α-helix and an increase in β-pleated sheet structure were evidenced; those are related to a reduction in albumin that predominates the α-helix structure, and an increase in immunoglobulins correlate with the increase of the β-pleated sheet [28].
Regarding the ratio of total lipids area/total proteins area, we did not fnd changes; in this way, the changes in the acyl chain length of lipids can serve as an indicator of more subtle diferences in the structure of lipids that are not related to the total content of proteins [19].
Te changes in lipids and proteins are related to the cytopathogenesis efect of the SARS-Co-V2 infection, which could be related to increased lactate dehydrogenase (LDH) levels in dead patients.In the same way, LDH could be related to the necrosis process.While CRP is an infammatory marker, LDH is usually employed as a marker of cell damage, and it has been associated with the severity of COVID-19.LDH is an enzyme involved in carbohydrate metabolism by the conversion of lactate and pyruvate.Abnormal LDH levels can result from decreased oxygenation, leading to an upregulation of the glycolytic pathway and multiple organ injuries.In cell damage, LDH is released, and high levels also have an unfavorable prognosis in other pathologies, e.g., infections, malignancies, sepsis, or cardiopulmonary compromise [6,24,[36][37][38][39].
Here, we found a signifcant decrease in serum levels of albumin and an increase in CRP in the dead group; the reduction in the albumin level is related to the acute phase response of patients to viral infection with an increase in CRP.Te decrease in hepatic synthesis of proteins is associated with dysregulation of hepatic function; even though the signifcant ALT levels are in survival patients, we also fnd increased levels of ALT 43 U/L (29, 72 U/L) when compared with the previous report of 26.20 U/L (17.00-53.00U/L), suggesting hepatic dysfunction [28,40].
Biochemical changes are related to mortality, such as an increase in AST, ALT, urea, creatinine, and CRP, that are similar to our fndings in serum levels and ATR-FTIR analysis, such as urea area peaks elevated at 1452 cm −1 [1,3,24,41].
In severe respiratory failure and death, multiple studies suggest that the dysregulated immune response is one of the main factors [10].
Lasso et al. showed that the higher peak of IgG correlates with the length of hospitalization and disease severity in a kinetic assay during the hospitalization stay, but they did not fnd any signifcant diference at admission.Likewise, herein, at the time of the sampling test, we did not fnd any change in IgG levels by ATR-FTIR spectroscopy; according to the fndings of Lasso et al., they showed similar IgG responses in severe and mild patients [10,42].
Te increase in IgM area (1160-1028 cm −1 ) was significantly lower in the survival group than in the death group.About this, Martinez-Cuazitl et al. demonstrated that IgM is lower in mildly COVID-19 patients in comparison with severely ill patients [13,42].One explanation is an antibodydependent enhancement (ADE), a phenomenon described for infections with viruses such as dengue.ADE occurs when the antibody does not block the viral entry; the antibody might facilitate Fc-receptor-mediated endocytosis of the virus, enhanced viral replication, and massive infammatory responses [2].Lasso et al. and many other studies showed higher levels of leukocytes, neutrophils, and low levels of lymphocytes in dead patients, similar to our fndings.Te lymphopenia could be explained by apoptosis, according to the lipids and LDH changes described before [10,24,37,43,44].
Also, the lymphopenia and increased neutrophillymphocyte ratio seen in patients with SARS-CoV-2 infection may be explained due the secretion of such cytokines and chemokines (IL-6, IFN-c, MCP1, and IP-10), inducing the recruitment of immune cells from the blood and the infltration of lymphocytes into the airways [7].
Severe symptoms of COVID-19 and death patients showed an increased serum concentration of many proinfammatory mediators such as IL-10, IL-6, IL1-β, TNF-α, and IFN-c, which is correlated with lymphopenia, and infltration of hyperactive mononuclear cells in the lungs, that is according to our fndings.Furthermore, although we did not fnd signifcant diferences in IL-6 and IL-1β, high levels of these interleukins in the death group were evident [45].
Te immune response against severe SARS-CoV-2 infection is characterized by a combination of delayed-type I/III interferon response and production of proinfammatory cytokines that recruit efector cells; type III IFNs are structurally related to IL-10 family cytokines, that is, according to our fndings by ATR-FTIR that increase IL-10 on death patients [10,46].
In addition, the increase of type II interferon (IFN-c) is secreted by natural killer and Tcells but not directly by virusinfected cells; IFN-c activates macrophages that induce more proinfammatory cytokines, dysregulating the immune system.Also, higher levels of IFN-c are related to the CRP levels described by Paolini et al. [44,47].In another way, the cell death pathways (apoptosis, necroptosis, and proptosis) induced by SARS-CoV-2 infection could be modulated by the higher levels of tumor necrosis factor (TNF) and IFN-c.Te cell death is directly correlated with lung injury, multiple organ failure, and critical prognosis [44].
Te increase in cell death also could be regulated by the NLR family pyrin domain containing 3 (NLRP3) infammasome that induces pyroptotic cell death mediated by the caspase-1 dependent cleavage and release the IL1-β, increasing LDH levels, as our fndings [2].Journal of Spectroscopy

Conclusions
Here, we showed that patients in the death group were older, overweight, and smokers.Moreover, it is essential to mention that it showed a higher respiratory rate than the survival group.ATR-FTIR spectroscopy allowed to establish an association between clinical course (survival or death) and biochemical and immunological status once a decrease in lipid biogenesis pathways was evidenced as well as an increment of phospholipids, which are related to virus propagation and cell damage in the death group.
Shifts in the amide I band, a decrease in the α-helix/ β-pleated sheet related to the reduction of albumin and increase of urea, and immunoglobulin production, especially IgM, demonstrate the host cell damage induced by the SARS-CoV-2 infection.
Cell damage also induces the cellular immune response that modulates the cytokines IL-10, IL-6, IL1-β, TNF-α, IFNc, and the lymphopenia shown in the death group, also refected in higher levels of infammatory markers (CRP and LDH).

Figure 1 :
Figure 1: ATR-FTIR spectra of saliva samples of hospitalized patients with severe COVID-19 who survived or died.(a) Mean of the ATR-FTIR spectra of both groups, including the biological fngerprint region at 1800-800 cm −1 .(b) Biochemical components assignation to the peaks.p represents the statistical signifcance after the Mann-Whitney U analysis.
Patients.From May 19, 2020, to March 1, 2021, 103 COVID-19 patients with category 3 risk factors for severe disease, according to the COVID-19 Treatment Guidance of Massachusetts General Hospital, and admitted to Central Military Hospital were recruited.

Table 1 :
Demographics of patients.

Table 3 :
Evolution of patients.
Te bold values refer to statistical signifcance in each group.