Clinical Characteristics, Risk Factors and Predictive Value of COVID-19 Pneumonia: A Retrospective Study of 173 Patients in Wuhan, China

Background: COVID-19 is a globally emerging infectious disease. As the global epidemic continues to spread, the risk of COVID-19 transmission and diffusion in the world will also remain. Currently, several studies describing its clinical characteristics have focused on the initial outbreak, but rarely to the later stage. Here we described clinical characteristics, risk factors for disease severity and in-hospital outcome in patients with COVID-19 pneumonia from Wuhan. Methods: Patients with COVID-19 pneumonia admitted to Cancer Center, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology from February 13 to March 8, 2020, were retrospectively enrolled. Multivariable logistic regression analysis was used to identify risk factors for disease severity and in-hospital outcome and establish predictive models. Receiver operating characteristic (ROC) curve was used to assess the predictive value of above models. Results: 106 (61.3%) were female. age of populations 62.0 underlying including

characteristic and risk factors of con rmed cases who were admitted to the Cancer Center of Wuhan Union Hospital from February 13 to March 8, 2020.

Study design
In this retrospective single-center study, we described clinical characteristics of patients with COVID-19 pneumonia admitted to Cancer Center, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology from February 13 to March 8, 2020, and analyzed risk factors for disease severity and inhospital outcome.

Diagnosis and clinical types
According to WHO interim guidance and the diagnostic and treatment guidelines issued by the Chinese National Health Committee (version 5), COVID-19 pneumonia were mainly con rmed by chest computed tomography (CT) and SARS-CoV-2 nucleic acids test in throat swab [10][11].
As suggested, COVID-19 pneumonia was divided into four types. Mild type had minimal clinical symptoms and no imaging ndings of pneumonia. Common type had fever, respiratory and other symptoms, and imaging ndings of pneumonia. Severe type met any of the following: (1) respiratory distress with respiratory frequency ≥ 30 times/minutes; (2) pulse oximeter oxygen saturation at rest ≤ 93%; (3) arterial oxygen partial pressure (PaO 2 ) /inspired oxygen fraction (FiO 2 ) ratio ≤ 300 mm Hg (1 mmHg = 0.133 kPa). Critically severe type met any of the following: (1) respiratory failure requiring mechanical ventilation; (2) shock; (3) combination with other organ failure requiring monitoring and treatment of intensive care unit [10][11]. Here we combined into severe illness (mild/common type) and non-severe illness (severe/critically severe type).

Discharge criteria
According to China guidelines, the discharge standards for patients with COVID-19 pneumonia met all of the following: (1) body temperature returned to normal for more than 3 days; (2) respiratory symptoms improved signi cantly; (3) lung imaging showed obvious absorption of in ammation; (4) two consecutive nucleic acid test for respiratory pathogens was negative (the sampling interval is at least one day) [11].

Data extraction
We collected relevant data from electronic medical records, including epidemiologic and demographic information, comorbidities, symptoms, laboratory and CT ndings, treatment, as well as in-hospital outcome (discharge, hospitalization/death) on March 8, 2020. Serum cytokine levels such as interleukins (IL)-2, IL-4, IL-6, IL-10, tumor necrosis factor-a, and interferon-γ, were measured on admission. The data of each patient was carefully examined by at least two researchers and then entered into a computer database.

Statistical analysis
Continuous variables were described by median and interquartile range (IQR), while categorical variables were expressed as the counts and percentages. Univariate analysis (Chi-square tests or Fisher's exact tests) and multivariable logistic regression analysis were used to identify risk factors for disease severity and in-hospital outcome, and establish predictive models. Further, receiver operating characteristic (ROC) curve was drawn and the area under the ROC curve (AUC) was calculated to assess the predictive value of above models. All tests were 2-tailed, and the level of signi cance was P = 0.05. Statistical tests were performed using SPSS version 22.0 software.

Clinical characteristics
Clinical characteristics are shown in Table 1. From February 13 to March 8, 2020, a total of 173 patients with COVID-19 pneumonia were admitted to the Cancer Center of Wuhan Union Hospital and enrolled in the study. Among them, 67 (38.7%) were male while 106 (61.3%) were female. The patients ranged in age from 18 to 96 years, with the median age of 62 years (IQR 49-69 years), of whom 64 (37.0%) were over 65 years of age.

Risk factors for in-hospital outcome and their predictive value
As of March 8, 2020, 90 patients (52.0%) were discharged, and 83 patients (48.0%) were not discharged (2 deaths). For discharged and undischarged patients, the median time from onset to admission were 14 days (IQR 7-20 days) and 15 days (IQR 7-24 days), and the median time for hospitalization were 11 days (IQR 9-17 days) and 21 days (IQR 12-22 days). Two patients died on the 7th and 9th days after admission, respectively.

Discussion
In this study, 67 (38.7%) of the patients were males, and the severe illness rate was 44.8% in males but 34.9% in females. Sex differences have been reported regarding the susceptibility, severity and outcome of COVID-19 pneumonia. In addition to estrogen protection and different lifestyle, angiotensin-converting enzyme (ACE) 2 may explain sex differences. ACE2 is a receptor for SARS-CoV-2 entry into cells, and its higher expression in male lungs may correlate with high prevalence, severe illness rate and mortality in males [12][13][14]. However, our data did not con rm sex differences in COVID-19 pneumonia. This may result from different study periods and trial designs.
The median age of our cases [62.0 years (IQR 49-69)] was much higher than that reported by Chen [15] and Zhang [16], potentially leading to more comorbidities. Similar to other recent reports [17][18], the most common comorbidities were hypertension (24.9%), diabetes mellitus (9.8%) and heart disease (8.7%). Cumulative studies con rmed that older age and comorbidities were associated with severe COVID-19 pneumonia and poor outcomes [6,8,19]. Also, our data supported older age as a risk factor. In the process of developing severe illness, in ammatory storm usually plays an important role. Comprehensive genomic analyses exhibited that older animals had stronger host innate immune response to SARS-CoV infection, and tended to develop more severe pneumonia [20].
Although the main symptoms were fever and respiratory symptoms, digestive symptoms were also very common. In our report, almost a quarter of patients (22.0%) had digestive symptoms, consistent with that reported by Dan [21], but much higher than that reported by Guan [6]. Previous univariate analysis showed that the patients with digestive symptoms were at greater risk of severe illness of COVID-19 pneumonia [22][23]. On multivariate logistic regression analysis, we further found that diarrhea was associated with patients with severe illness. After cells in lungs are infected by SARS-CoV-2, effector CD4 + T cells migrate to the small intestine through the gut-lung axis, thereby causing intestinal ora imbalance and diarrhea. In turn, abnormal intestinal ora fails to regulate the host's immune response, leading to in ammation storm closely associated with lungs and vital organ systems [24][25].
The lymphocytes counts serves as a reference indicator in the diagnosis of COVID-19 pneumonia [11]. In this study, peripheral blood lymphocytes counts were even lower in severe illness than in non-severe illness, possibly due to the recruitment of lymphocytes from the peripheral blood to lungs by the SARS-CoV-2 [26][27].
More important, we supported two important indicators of in ammation response, white blood cell and IL-6 as parameters for assessing in-hospital outcome in patients with COVID-19 pneumonia. There is increasing evidence that IL-6 has become the main driver of in ammation associated with the disease. Of all tested cytokines, IL-6 is increased most signi cantly, with an upward trend of more than 10 times [28][29].
Among 173 patients, 19.7% had an increased LDH levels and only 10.4% had an increased creatine kinase levels. The results suggested that some patients with increased LDH levels may not myocardial injury. LDH levels may re ect tissue necrosis associated with immune hyperactivity [30]. Thus, LDH may be a useful and easy to test parameter in order to predict poor clinical outcome. Notably, antiviral treatment does not seem to improve in-hospital outcomes in this study. Even a cohort study from Italy showed that the patients receiving antiviral treatment had lower odds of discharge [31]. The results suggested that we need to use them with caution before con rming the effects of these antiviral drugs on COVID-19 pneumonia.
This study had also some limitations. First, this was a retrospective single-center study, so it might lack of highlevel evidence. Next, all patients in this study were enrolled after Chinese Spring Festival travel rush, and the results might only re ect clinical characteristics of the epidemic in Wuhan during a certain period. Finally, since there were no recognized treatment guidelines of COVID-19 pneumonia, the impact of treatment on in-hospital outcome was not focused on analysis.

Conclusion
Older patients with diarrhea and lymphopenia need early identi cation and timely intervention to prevent the progression to severe COVID-19 pneumonia. However, older patients with leucopenia, increased lactic dehydrogenase and interleukins-6 levels are at a high risk for poor in-hospital outcome.

Declarations
Ethics approval and consent to participate The study was approved by the Ethics Committee of Hospital of Chengdu University of Traditional Chinese Medicine.

Consent for publication
Not applicable.

Availability of data and materials
The data used to support the ndings of this study are included within the article. Figure 1 ROC Curve of predictive logistic regression model for disease severity of COVID-19 pneumonia. A prediction model for disease severity on admission is constructed by three indicators of age, diarrhea and lymphopenia.