Clinical characteristics and prognostic factors of COVID-19 patients progression to severe: a retrospective, observational study

The outbreak of coronavirus disease 2019 (COVID-19) has become a world-wide emergency. The severity of COVID-19 is highly correlated with its mortality rate. We aimed to disclose the clinical characteristics and prognostic factors of COVID-19 patients who developed severe COVID-19. The study enrolled cases (no=1848) with mild or moderate type of COVID-19 in Fangcang shelter hospital of Jianghan. A total of 56 patients progressed from mild or moderate to severe. We used least absolute shrinkage and selection operator regression model to select prognostic factors for this model. The case-severity rate was 3.6% in the shelter hospital. They were all symptomatic at admission. Fever, cough, and fatigue were the most common symptoms. Hypertension, diabetes and coronary heart diseases were common co-morbidities. Predictors contained in the prediction nomogram included fever, distribution of peak temperature (>38°C), myalgia or arthralgia and distribution of C-reactive protein (≥10 mg per L). The distribution of peak temperature (>38°C) on set, myalgia or arthralgia and C-reactive protein (≥10 mg per L) were the prognostic factors to identify the progression of COVID-19 patients with mild or moderate type. Early attention to these risk factors will help alleviate the progress of the COVID-19.

AGING been diagnosed every day, which requires enormous medical resources. The surge of infections placed huge pressure on the national medical system [2].
The Fangcang shelter hospitals in Wuhan were largescale, temporary hospitals, rapidly built by converting existing public venues, such as exhibition centers and stadiums, into health-care facilities. They were served to isolate patients with mild or moderate COVID-19 from their families and communities, while providing basic medical care, disease monitoring, food, shelter, and social activities [3]. Patients with mild or moderate COVID-19 who met additional admission criteria were isolated and treated in the Fangcang shelter hospitals, whereas patients with severe or critical COVID-19 received medical care in traditional hospitals [3][4][5][6]. Fangcang shelter hospitals provide basic medical care and monitored the progression of disease. As some patients remain experienced progression of COVID-19 or development of severe chronic diseases, they were transferred in a timely manner to the designated higherlevel hospitals. The clinical characteristics of patients transferred to the designated hospital were important for the revision of admission criteria of COVID patients in Fangcang shelter hospitals.
The severity of the COVID-19 determines the fatality rate and the medical resource usage. The case-severity (mild or moderate progressing to severe case) rate was an important benefit index for therapeutic efficacy assessment in shelter hospital. Dynamic observation the risk factors of mild to severe patients is contribute to great value for early prognosis and treatment. Therefore, a retrospective review of overall medical record was performed in Fangcang shelter hospital of Jianghan, which received the largest number of patients among Fangcang shelter hospitals in Wuhan. A total of 1848 cases with mild or moderate COVID-19 were included and 521 cases transferred to the designated hospital were analyzed. The clinical characteristics and prognostic factors of patients from mild or moderate to severe were detected as well.

Outcomes and case-severity rate of patients in Fangcang shelter hospital
Among 1848 enrolled patients, the age range was from 15 to 81 years, and 49.0% were men. From Feb 5th to Mar 9 th 2020, 521 (28.2%) patients were transferred to the designated hospitals for further treatment. Meanwhile, the other 1327 (71.8%) patients reached the criteria of isolation release or discharge, (Figure 1). Among 521 patients transferring to the designated hospitals, 10.7% patients with severe type from mild or moderate type (56 cases), 6.9% (36 cases) patients with body temperature more than 38.5°C for 3 days or more after treatment, 15.0% (78 cases) patients with cancer or severe liver/kidney/heart disease, 45.9% (239 cases) patients with the persistent positive nucleic acid testing after 2 weeks treatment, and 21.5% (112 cases) patients with other reasons (including new onset severe symptoms or mental illness or tremendous mental pressure or pregnant woman). The basic clinical characteristics of patients transferring to the designated hospitals were in Table 1.
The case-severity rate was identified as the proportion of mild or moderate type progressing to severe type in this study. A total of 56 patients have progressed from mild or moderate type to the severe type. The caseseverity rate of COVID-19 was 3.6% (67/1848) in the Fangcang shelter hospital of Jianghan.
According to the symptoms and signs, patients received different treatments. There were 16/42 (38.1%) of patients with antibiotics, antivirus and traditional Chinese medicine, 5/42 (11.9%) of patients with antivirus and antibiotics, 6/42 (14.3%) of patients with antivirus and traditional Chinese medicine. There were 14 patients with no medicine records due to the short time of staying in Fangcang shelter hospital.
On March 12, those 56 patients were followed up by telephone and 4 lost. There were 5.8% (9/52) of the patients whose nucleic acid test results remain positive.
As shown in the CT scan, 96.2% (50/52) of the patients were recovery or improved, only 3.8% (2/52) of the patients got deterioration or new lesions. The clinical manifestations and laboratory findings were shown in Table 3.

The prognostic factors of patients transforming from mild or moderate type to severe type
There were 45 variables in the LASSO regression and the outcomes displayed that median incubation (onset to shelter) period (≥14 days), fever, distribution of peak temperature (>38°C), sputum production, sore throat, shortness of breath, myalgia or arthralgia, poor appetite, headache, diabetes and CRP (≥10 per L) were predictive factors for patients transforming from mild or moderate type to severe type with nonzero coefficients ( Figure  2A, 2B). Ten variables mentioned above were then included in the multiple logistic regression analysis. The results demonstrated that fever, distribution of peak temperature (>38°C), myalgia or arthralgia and distribution of CRP (≥10 per L) were significantly predictive factors (Table 4). Moreover, the model that incorporated the above independent predictors was developed and presented as the nomogram ( Figure 2C). The calibration curves of this nomogram showed good agreement in this cohort ( Figure 2D). The C-index of the prediction nomogram was 0.888 (95% CI 0.839-0.937) for this cohort, which was confirmed to be 0.876  Data are median (IQR), n (%) or n/N (%).   AGING by bootstrapping validation. In this risk nomogram, effective performance showed a good prediction capability. Furthermore, the decision curve demonstrated that the prediction nomogram had superior standardized net benefit while the threshold probability of a patient and a doctor is 4% and 83%, respectively ( Figure 2E).

AGING
Furthermore, by utilizing receiver operating characteristic (ROC) analysis, we compared the predictive value of our prediction model for incidence of severe illness with that of the MuLBSTA, CURB-6 and NLR models. The analysis revealed that our prediction nomogram had the highest area under curve (AUC) (0.902) than the other three models (Supplementary Figure 1).

DISCUSSION
It is the first time to observed dynamically and comprehensively disclose the clinical characteristics and prognostic factors of COVID-19 patient progression to severe. The number of patients with severe type determines the final mortality rate of COVID-19. In this study, the case-severity rate was observed with a relatively large prospective cohort, which might be a valuable complement to the characteristics of COVID-19. In the Fangcang shelter hospital of Jianghan, about 3.0% of the patients transformed to severe, which was significantly lower than the 14% cases classified as severe or critical in the spectrum of COVID-19 disease [7,8]. The most common symptom at admission was fever and 76.8% (43/56 cases) mild to severe patients got fever on cabin admission. However, some patients with Covid-19 did not have fever abnormalities on initial presentation, which has complicated the diagnosis [9]. In this study, 55.4% patients were with peek temperature more than 38.0°C, while 12.5% of patients were once with fever over 39.0°C. High fever was associated with the development of severity and critical death [10][11][12]. Therefore, keep vigilance of those mild patients whose peak temperature over 38.0°C. They were quickly transferred to the designated higher-level hospitals once the blood oxygen saturation of those patients was less than 93% in Fangcang shelter hospital of Jianghan.
For more specialized monitoring, chest imaging and laboratory services were applied in the Fangcang shelter hospitals. Ground-glass opacity (53/56, 94.6%) was the most common morphological depiction in CT scan on admission. However, only 19.6% of patients showed multiple lesions, 42.9 % patients presented bilateral lesions. Compared with mild patients, most severe patients took CT scan in this study within seven days. The lesions that were present in asymptomatic individuals progressed to bilateral diffuse disease with consolidation around day 10 after the symptom onset [13][14][15]. The predominant CT pattern was unilateral and multifocal ground-glass opacities in early stage, then lesions quickly evolved to bilateral, diffuse groundglass opacity in later stage [13]. However, those characteristics were not consistent with what we had expected in this study. Therefore, the value of lung CT in determining the prognosis of mild COVID-19 patients still needs further research. Meanwhile, most of the patients had elevated levels of CRP at ill onset with 33.3% over 10 mg/L. Similarly, compared to mild or moderate cases, severe cases more frequently had higher levels of CRP [9,16,17]. Therefore, imaging and laboratory results could contribute to make the quick decision of transferring to the designated hospitals.
Previously, older age (over 65 years) was associated with higher odds of progression to severity of COVID-19, which also has been reported as an important independent predictor of mortality in SARS and MERS [18][19][20]. In order to better display the clinical characteristics of mild or moderate to severe patients below 65 years old, 92 age-and sex-matched mild or moderate patients stayed in the Fangcang shelter hospital at the same time. In this study, several factors in adults who were hospitalized in Fangcang shelter hospital were associated with mild progressed to severe COVID-19. In particular, patients aged 40-65 years constituted the highest proportion within the severe group in this study. It had reported that 75% of COVID-19 death cases previously suffered 1-2 underlying diseases, a majority of which were diabetes and cardiovascular diseases [21]. In line with above evidence, our study also found that 42.9% of the mild to severe patients had 1-2 basic diseases, such as cardiovascular diseases, cerebrovascular diseases and endocrine system diseases.
Notably, several clinical manifestations were identified as prognostic factors for progression from mild to severe in the univariate logistic regression analysis. We found that myalgia or arthralgia on admission was associated with increased odds of mild to severe (Table  4). Furthermore, 30.4% patients had the myalgia or arthralgia. Less common symptoms include poor appetite, sore throat and shortness of breath. However, respiratory system affection remained as the primary symptom [9,22,23]. Overall, onset of fever and myalgia or arthralgia symptoms should be closely monitored among cabin hospital, more attention should also be paid to patients on those isolation patients at home.
In this study, we developed a prediction nomogram included fever, distribution of peak temperature AGING (>38°C), myalgia or arthralgia and distribution of Creactive protein (≥10 per L) for patients with mild or moderate to severe COVID−19. Previous studies suggested that both MuLBSTA score and CURB-65 score were widely used to assess the mortality of pneumonia [24][25][26]. Moreover, a recent study had confirmed that neutrophil-to-lymphocyte ratio (NLR) could predicts severe illness patients with COVID−19 [27]. However, compared with the above three models, our prediction nomogram model seemed to have a higher C-index and AUC than them. Additionally, our model seems to be more clinical significance. Because we can predict the incidence of patients with COVID-19 from mild to severe. And the other three models mainly predict the mortality rate. Therefore, our model can play a warning role in the early stages of disease. According to the results, we suggested that patients with fever (peak temperature over 38.0°C), myalgia or arthralgia and CRP more than 10 per L should be vigilant by doctors and nurses in Fangcang shelter hospital.  [28,29]. We believe that the information in the article will provide valuable data for other countries facing the COVID-19 pandemic, when they are establishing the national public health emergency management for COVID-19.

Patient enrollment
The Jianghan Fangcang shelter hospital opened on the 5 th Feb and closed on the 9 th Mar 2020.

Follow-up
During the following days in Fangcang shelter hospital, the patients were re-examined for laboratory and imaging examination, and recorded symptoms, signs, treatments and outcome events. The throat swab specimens of RT-PCR test and chest CT scan were performed according to the symptoms and signs. After admission, all patients were given antiviral therapy (e.g., abidor hydrochloride) and other individualized treatments (such as antibiotics, antihypertensive and hypoglycemic therapy) according to the doctor's advice and NHC's interim guidelines [7]. The clinical outcomes of patients in the Fangcang shelter hospital were divided into three ways. They were the patients transferred to the designated hospitals for further treatments, the patients reaching the criteria of isolation release, and the patients kept treatment in the mobile cabin.

COVID-19 nucleic acid detection and chest CT scan
Throat swab samples were stored in virus transport medium and transported to Wuhan Union Hospital for laboratory diagnosis. Throat swab specimens of all patients were subject to real time PCR tests by amplifying ORF1ab gene and N gene of SARS-CoV-2 (BioGerm, Shanghai, China). The CT examinations were carried out with a 16-row multidetector CT scanner (μCT550, Shanghai LianYing medical technology co., LTD) using the following parameters: detector collimation widths 64 ×0.6 mm, 128×0.6 mm, 64×0.6 mm, and 64×0.6 mm; and tube voltage 120 kV.

Statistical analysis
Continuous variables were represented by mean (standard deviations, SDs) or median (interquartile range, IQR) as appropriate, categorical variables were described as number (%). Significant differences between the 2 groups (mild patients and mild to severe patients) were compared by Student t test, Mann-Whitney U test, chi-square or Fischer exact test where appropriate. In addition, we also used least absolute shrinkage and selection operator (LASSO) method to select the optimal predictive features in risk factors from the patients with COVID-19, and features with nonzero coefficients were selected in the LASSO model. Then, multivariable logistic regression model was used to build a predicting model by incorporating the features selected in the LASSO regression model. The variables with the P-value less than 0.05 were included in this model. The results were present as odds ratio (OR) [95% confidence interval]. Because all tests were two-sided, P value less than 0.05 was considered statistically significant. Analyses were performed using SPSS 20.0 statistical package and R version 3.6.0 was used to build nomogram, calibration and decision curve.

AUTHOR CONTRIBUTIONS
HS designed the study and had full access to all of the data in the study and was responsibility for the integrity of the data and the accuracy of the data analysis. HW, JH, YW, ZW, XC, CY and XF collected data. YL, YF and BW analyzed data. YL and YF wrote the article. All authors critically revised the manuscript for important content and gave final approval for the version to be published.

ACKNOWLEDGMENTS
The Jianghan Fangcang shelter hospital was managed daily by Union Hospital, Tongji Medical College, Huazhong University of Science and Technology and twenty-one medical teams from other provinces associated with six local hospitals participated in the work. We thank 1573 staff in the Jianghan Fangcang shelter hospital, including 308 doctors, 872 nurses, 120 AGING SUPPLEMENTARY MATERIALS

Supplementary Figure
Supplementary