Analysis of risk factors of severe COVID-19 patients

Objective: To explore relevant risk factors for severity of patients diagnosed with novel coronavirus pneumonia (COVID-19). Methods: The clinical data of 292 patients with COVID-19 admitted to Hubei Provincial Hospital of Integrated Chinese & Western Medicine from January 1, 2020 to February 29, 2020 were analyzed retrospectively. Patients were divided into mild or severe group according to the Guidance for Corona Virus Disease 2019 (7 th version) released by the Chinese National Health Committee. The clinical data were collected at the time of admission, including demographics, clinical characteristics, laboratory tests, imaging characteristics and outcomes of treatments. We applied univariable and multivariable logistic regression methods to explore the risk factors associated with severity of the disease. Results: The median age of patients in the severe group ((68.19±12.51) years) was signi�cantly older than mild group ((54.14 ± 13.62) years). The male sex was more predominant in severe group (63.45%) than that of mild group (38.1%). There were more smokers (8.97% vs 1.36%) and drinkers (4.14% vs 0%) in severe group than that of mild group. Patients in the severe group had more underlying diseases. Hypertension(cid:0)48.97% vs 23.81%(cid:0)(cid:0)coronary heart disease (22.07% vs 1.36%, P<0.0001) , chronic obstructive pulmonary disease (6.21% vs 1.36%), malignant tumor (7.59% vs 2.04%) and chronic kidney disease (3.45% vs 0%) were more frequent in severe group than in mild group. The dyspnea, chest tightness and dry cough were more common in severe group (43.45%, 66.9% and 66.21%) than in mild group (23.13%, 44.22% and 53.74%). Abnormality of chest radiography were more frequent in the severe group, there were more ground glass opacities, consolidation of lung and white lung in the severe cases (88.97%, 44.07% and 46.21%) than in mild cases (78.91%, 19.05% and 2.04%). Patients in the severe group were more likely to receive methylprednisolone, oxygen therapy and mechanical ventilation. Lasso algorithm showed that age, C-reactive protein (CRP), creatine kinase (CK) and α-hydroxybutyrate dehydrogenase (α-HBDB) were independent risk factors for severe COVID-19, but the count of CD 4+ T lymphocyte was the protective factor. Conclusion: This retrospective study of 292 COVID-19 patients revealed that age, CRP, CK, α-HBDB and the count of CD 4+ T lymphocyte were independent risk factors for severity of COVID-19. Identifying patients with risk factors at an early stage of the disease are helpful for outcome prediction and clinical management.

The world health organization declared that the outbreak of SARS-Cov-2 constituted a public health emergency of international concern on 30th January 2020.The global situation was very grave.One research had suggested that the SARS-Cov-2 infection caused severe pneumonia, which clinical manifestations are similar to sars-cov infection, associating with admissions to intensive care units and high mortality rate [7] .The other study published in the lancet shows that older age, D-dimer greater than 1 μg/mL and higher Sequential Organ Failure Assessment (SOFA) score could predict severe COVID-19 in the early stage [8] .
The purpose of this study is to compare the clinical and laboratory characteristics of patients with different clinical types of COVID-19 who admitted to the isolation ward of Hubei Provincial Hospital of Integrated Chinese & Western Medicine, which is one of the designated hospitals assigned by Chinese government.We aimed to explore the risk factors of severe patients, and to provide scienti c basis for reducing the incidence and mortality of severe COVID-19.

Methods
Study Design and Participants 292 patients diagnosed with COVID-19 in our hospital from January 1, 2020 to February 29, 2020 were selected as the study objects.Diagnostic and clinical typing criteria were according to the Guidance for Corona Virus Disease 2019 (7 th version) released by the Chinese National Health Committee [9] .Exclusion criteria as follows: (1) patients with psychiatric diseases who did not cooperate with the treatment.(2)   patients with incomplete data or transferred to other hospitals.(3) patients with pneumonia caused by other pathogens.This study had been approved by the ethics committee of our hospital.This case series were approved by the institutional ethics board of Hubei Provincial Hospital of Integrated Chinese & Western Medicine (No. 2020011).Written informed consents were waived because of the rapid emergence of this serious epidemic.

Data collection
The following data were collected and recorded: (1) The basic data included epidemiological history, demographic characteristics, signs and symptoms.(2) The results of laboratory examination included blood routine, C-reactive protein, SAA (Serum amyloid A protein), biochemistry, immunity, in ammation, coagulation function and other indicators.(3) The chest radiography characteristics.(4) Treatments.(5)  Outcomes.The data collection tables were independently reviewed by two researchers.

Statistical analyses
Continuous variables were presented as ± standard error of mean (SEM) or medians (interquartile range, IQR) depending on whether they tted the normal distribution, and the comparisons between two groups were performed by using Wilcoxon rank sum test.For qualitative variables, statistical description was expressed as frequency (percentage), and Fisher exact probability method was chosen for comparisons between two groups.To explore risk factors with the association of the severity of COVID-19 in our subjects, univariable and multivariable logistic regression models were used to estimate odds ratios (ORs) and 95% CIs, adjusting for confounders including age, sex, comorbidities, COVID-19 treatments.Cox proportional-hazards models were applied to determine hazard ratios (HRs) and 95% CIs for disease outcome, adjusted for the aforementioned confounders.Sensitivity analyses were done for all adjustment variables, comparing the results between univariable analysis without adjusting confounders and multivariable analysis with adjusting confounders for disease severity in patients with COVID-19.
Statistical software was R version 3.6.3(The R Foundation), and all hypothesis tests were two-sided tests with a signi cance level of 0.05.
Regarding clinical symptoms, the dyspnea, chest tightness and dry cough were more common in severe group (43.45%, 66.9% and 66.21%) than in mild group (23.13%, 44.22% and 53.74%).Details of other relevant information were shown in Table 1.

Laboratory tests
After admission, tests of patients' rst blood routine, biochemical, immune and coagulation were performed in the two groups.Substantial differences in laboratory ndings from the two groups were displayed in Table 2.
The patients in severe group had persistent and more severe lymphopenia (0.72 (IQR, 0.5-1.05))than the mild ones (1.11

Radiological characteristics
The lung lesions in the two groups were mostly distributed under the pleura, accompanied by multiple patchy or lumpy ground glass opacities, with or without pulmonary consolidation or white lung.In the severe group, the lung lesions were mainly distributed in two lungs, with multiple ground glass opacities (88.97% vs 78.91%, P=0.0252), pulmonary consolidation (42.07%vs 19.05%, P<0.0001) and white lung (46.21% vs 2.04%, P<0.0001,) more common than in the mild group which were summarized in Table 3.

Treatment and outcome
All patients in the two groups were treated with antiviral, anti-bacteria, nutrition, traditional Chinese medicine and symptomatic support.Compared with the mild group, the patients in the severe group received more methylprednisolone, oxygen therapy, noninvasive mechanical ventilation and invasive mechanical ventilation, which were signi cantly higher than that in the mild group.In 145 severe cases, 96 patients were cured and discharged, accounting for 65.31%, 51 patients died, accounting for 34.69%.
For the mild group, no dead patients were found and all of them were cured and discharged.There was statistical signi cance between this two groups, P<0.0001, displaying in Table 3.

Analysis of related risk factors
Multivariate analysis revealed that age, CRP, CD

Discussion
A recent epidemiological study made by China CDC showed that the mortality rate of critical COVID-19 patients could be as high as 49% [10] , aroused especial awareness in clinical management.
In this study, all 292 patients with COVID-19 were from Wuhan.Lasso algorithm concluded that age was the risk factor of severe patients.The risk of severe patients increased by 15.15% when the age increased 5 years.In the severe group, most of the patients with COVID-19 were elderly patients with basic diseases.Hypertension, coronary heart disease, chronic obstructive pulmonary disease, malignant tumor and chronic kidney disease were more frequent among severe group than that in the mild group.In 145 severe cases, 51 patients died, accounting for 34.69% and 90.2% of the dead patients were over 60 years.Additionally, 40 patients of the 51 deaths had underlying disease.The death rate of the patients with basic diseases was higher, accounting for 78.43%.The recent reports show that advanced age (>60) and comorbidities (particularly hypertension) were believed to be risk factors for severe disease and death from SARS-Cov-2 infection [4,5,7] , which was consistent with the results of previous studies [5,7,11] and our ndings.
In this study, C-reactive protein (CRP) in the severe group was higher than that in the mild group.CRP was selected as a risk factor of severe patients by lasso algorithm.The risk of severe patients increased 17.55% when CRP increased 20mg/L.Furthermore, we found that in ammatory markers such as SAA, IL-1, IL-6, IL-10, peripheral blood leukocytes, neutrophils and procalcitonin were signi cantly higher in severe patients, which indicated that severe patients might have had secondary bacterial infection.This might be closely related to death.Severe patients might also suffer from in ammatory cytokine storms, which caused fatal organ dysfunction and closely related to mortality [12] .
Through the lasso algorithm, we found that creatine kinase (CK) and α -hydroxybutyrate dehydrogenase (α -HBDB) were independent risk factors for patients with severe illness.The risk of severe patients increased 7.86% and 15.31% when CK increased 100u/L and α -HBDB increased 100u/L respectively.The CK mainly existed in skeletal muscle, cardiac muscle and smooth muscle.α -HBDH mainly existed in cardiac muscle and liver.We also found that the liver enzyme, LDH, Blood urea nitrogen and creatinine in the severe group were signi cantly higher than those in the mild group.These might be the results of multiple organ dysfunction caused by SARS-CoV-2 infection [13] .Recent research reported that severe and critical patients more often developed multi-organ dysfunction than that in mild patients [14] .A research reported that α -HBDH was an independent risk factor of SLE related to the liver injury [15] .
We found that CD 4 + T lymphocyte count was the risk factor of severe patients.The risk of severe patients increased 25.58% when CD 4 + T lymphocyte count reduced 50/UL.In this study, CD 4 + T lymphocyte count for 95.24% of these140 patients in the severe group was lower than the normal lower limiting value and the phenomenon indicated that the decrease of lymphocyte count might be an important factor to cause the aggravation of the patient's condition, which is consistent with the report of Central South Hospital [5] .Because of the continuous in ammatory reaction, lymphocyte apoptosis increased, the number rapidly decreased and the body entered the state of "immunosuppression" or "immune paralysis" [16] .In this situation, the proper use of immunosuppressive glucocorticoids might yield better results [17] .
The level of D-dimer, brinogen and PT in severe group were higher than that in mild group.The rise of D-dimer was in uenced by many factors.Acute in ammatory response caused by severe infection could affect coagulation and brinolysis via many ways.Studies have shown that high levels of D-dimer were related to the 4-week mortality [18] and its mechanism might be related to hemodynamic changes caused by systemic pro-in ammatory cytokine response [19] .Recent study showed that elevated level of D-dimer might be associated with the fatal outcome of COVID-19 infection [8] .In our study, D-dimer elevation occurred in 47 of the 51 deaths in the severe group, similar to the outcomes as previous.
The patients in the severe group showed multiple ground glass opacities of both lungs on the image, lung consolidation and white lung were more common than those in the mild group, similar to the research of Guan WJ [20] , about 30% of patients with COVID-19 would rapidly progress to ARDS [20] .These patients were more likely to develop to respiratory failure, severe pneumonia and ARDS.
This study found that α-HBDB was one of the independent risk factors for the onset of severe COVID-19 patients for the rst time which was not reported by any other relevant literatures before.But there are some limitations in this study, it was a single center retrospective study and no external validation cohort.It is hoped that the risk factors of severe COVID-19 can be veri ed through multi-center clinical research in the future.
Several limitations should be noted in our study.First, the samples is not too large, however, we did a power analysis for severity of COVID-19 in patients with or without cancer using the Pearson χ² test for two proportions method, and the power was more than 0•95.Second, less than 3% patients were exclude from our study sample because the are asymptomatic carrier so that they will not be more likely to die or more sick than others and will not introduce signi cant bias although they are missing data in our study.
In a word, as found by the cox regression, age, CRP, CK ,α-HBDB and CD 4 + T lymphocyte count were the independent risk factors for the onset of severe COVID-19 patients.Therefore, early attention to the above factors might be very helpful to improve the prognosis of COVID-19 in the future.

Declarations
Availability of data materials Concentrations of procalcitonin, high sensitivity C-reactive protein were obviously higher in severe group than in mild group (0.08 (IQR, 0.04-0.23)vs0.02(IQR,0.02-0.05),P<0.0001),(300(IQR, 153.03-300) vs 114.65(IQR, 19.41-300), P<0.0001).Concentrations of alanine aminotransferase, aspartate aminotransferase, total bilirubin, Direct bilirubin, alkaline phosphatase, and Gamma glutamyl transferase were especially higher in severe group than in mild group.Albumin concentrations were signi cantly lower in severe group than in mild group (33.8 (IQR,30.7-36.6)vs36.5 (IQR,34.25-40.2),P<0.0001).Concentrations of blood urea nitrogen, creatinine, creatine kinase, α-hydroxybutyrate dehydrogenase and lactate dehydrogenase were markedly higher in severe group than in mild group.Patients in the severe group had signi cantly lower concentrations of CD 4 + T lymphocyte, CD 8 + T lymphocyte and total T lymphocyte count more often than those of patients in the mild group.Concentrations of interleukin 1, interleukin 6 and interleukin 10 were signi cantly higher in severe group than in mild group.Median prothrombin time was signi cantly longer in severe group (13.2 (IQR, 12.6-14.3))than in mild group (12.5 (IQR, 12-13.35)),D-dimer and brinogen concentrations were markedly greater in severe group than in mild group.Details of other relevant information were shown in Table 2.