Clinical and paraclinical predictive factors for in-hospital mortality in adult patients with COVID-19

This retrospective cross-sectional study (February 2019-May 2020) was conducted on patients with con�rmed diagnosis COVID-19, who were admitted in Yazd Shahid Sadoughi Hospital, in middle of Iran. The patients with uncompleted or missed medical �les were excluded from the study. Data were extracted from the patients' medical �les and then analyzed. The patients were categorized as survivors and non-survivors groups, and they were compared


Background
Since December 2019, a type of coronavirus has emerged in Wuhan, China, which has become the focus of global attention due to an epidemic of pneumonia of unknown cause, called COVID-19.According to statistics of the World Health Organization, in this pandemic, more than 110 million de nitive cases of patients with COVID-19 were identi ed until February 22, 2021.Also, in Iran, until the same date, more than 1.5 million cases and about 60,000 deaths have been reported due to the virus (1).
Early diagnosis of this disease is very important because it affects the prognosis of patients.On the other hand, controlling risk factors and identifying high-risk individuals are considered essential (9).Given that different studies on different communities have reported scattered results on common clinical symptoms and paraclinical ndings as well as factors affecting the severity and mortality, this study aimed to investigate the factors affecting in-hospital mortality of patients with COVID-19 hospitalized in one of the main hospital in central Iran.

Methods
This retrospective cross-sectional study (February 2019-May 2020) was conducted on patients' medical les with diagnosis COVID-19, who were admitted and hospitalized in Shahid Sadoughi Hospital, Yazd, Iran, one of the biggest teaching and referral hospital in middle of Iran.The inclusion criteria were all adult patients (> 18 years), with con rmed diagnosis of COVID-19 using polymerase chain reaction (PCR) test.The patients with uncompleted or missed medical les were excluded from the study.
After relevant coordination, the patients' medical les were extracted from the hospital's archives unit and assessed.Data were recorded in a data gathering form, which was designed by the researchers according to previous researches.It was consist of below parts: 1) patients' demographic information (age, gender, marital status, type of residence, education levels); 2) medical history and clinical ndings at the time of admission; 3) laboratory ndings; 4) computed tomography (CT) scan ndings; 5) treatments; 6) complications; and 7) outcomes (discharge or in-hospital mortality).
All analyses were performed by SPSS version 16.0 for Windows.The Shapiro-Wilk t-test was used to test normal distribution of numerical variables.Independent sample t or Mann-Whitney tests was used for twogroup comparisons of continuous variables.Chi-square and Fisher's exact tests were used for proportions.In the univariate logistic regression analysis, each variable was separately entered.Variables with a P < 0.2 from the univariate analysis were entered into the multivariate logistic regression analysis, using the Forward Stepwise methods to determine predictive factors for in-hospital mortality, and odds ratio (OR) were reported.It is noteworthy that despite the large number of variables studied in this survey, the variables were entered the regression model in cluster form (medical ndings, laboratory ndings, treatments, and complications); and nally, signi cant variables in each cluster were entered the nal multivariate logistic regression model.Results were presented as mean ± standard deviation (SD) for continuous variables and were summarized in number (percentage) for categorical ones.Two-sided P-value less than 0.05 and con dence interval (CI) of 95% were considered to be statistically signi cant.
The current study was conducted in accordance with the Declaration of Helsinki, and it was approved by the vice-chancellor of research and technology, as well as the local ethics committee of Shahid Sadoughi University of Medical Sciences (IR.SSU.REC.1399.028).To consider ethical issue, the collected data were not revealed to anyone, except for the researchers; hence, patients' names were kept con dential.
In the present study, 93 patients (16.23%) died in the hospital.The mean ± SD of age of these patients was 14.36 ± 69.71 years.However, the results showed that age, gender, marital status, place of residence, cigarette, hookah, and drug use, recent travel history, and the history of contact with the suspect person were not different between survivors and non-survivors.In a study conducted in China, Wei et al. showed that the mean age of individuals who died from COVID-19 was 51 years, and most of them were elderly men (12).In a study by Zhou et al., the mean age of individuals who died from COVID-19 was reported as 69 years, which was signi cantly higher than the survived group, and most of the deceased patients were male (2).In a study by Tang et al., this rate was obtained 64 years which was signi cantly higher than the discharged group (13).
The most common underlying diseases in patients with COVID-19 in the present study were hypertension (36.4%),DM (26.6%), and CHD (12.8%), respectively, all of which were observed in non-survivors.Chen et al.
also reported that 51% of patients with a de nitive diagnosis of COVID-19 had an underlying disease (10).
Similarly, Liang et al. showed that 40% of patients with COVID-19 had an underlying disease, including cardiovascular, pulmonary, and cerebrovascular diseases, as well as DM and cancer, respectively (14).In another study performed by Zou et al., 51.59% of individuals had an underlying disease, including hypertension, cardiovascular diseases, DM, chronic respiratory disease, or cancer (15).In another study, the most common clinical manifestations of COVID-19 were reported as fever, cough, and fatigue (16).In Zhang et al.'s study, gastrointestinal symptoms, hypertension, and DM were reported as the main underlying diseases in these patients (17).
The present study showed that the most common clinical symptoms in patients were cough, fever, shortness of breath, and myalgia.Shortness of breath and loss of consciousness were signi cantly more common in non-survivors.Previous similar studies have shown that the most common symptoms observed in patients with COVID-19 were fever and chills, shortness of breath, cough, myalgia, weakness, lethargy, and gastrointestinal symptoms such as nausea, vomiting, and diarrhea (6, 10, 15).
The mean ± SD of PR and RR per minute in this study were signi cantly higher in non-survivors.On the other hand, the percentage of O 2 saturation in non-survivors was lower.GCS of 15 was more common in survived patients.In a study by Liu et al., the means of heart rate and RR per minute were 24 and 94, respectively (19).A RR > 24 per minute was reported 29% in Chen et al.'s study, which was signi cantly higher in deceased patients (63% versus 16%).Also, a heart rate > 125 beats per minute was observed in only 1% of patients.Fever was recorded in 94% of patients and it was similar in survivors and non-survivors (10).
In the present study, the most ndings of lung CT scan were bilateral in ltration (90.5%), peripheral lobes involvement (65.3%),GGO (45%), and air bronchogram (43%), generally.In non-survivors, mixed GGO/consolidation, air bronchogram, bilateral in ltration, mixed central-peripheral lobe involvement, LAP, crazy paving, and septal thickening were observed, but consolidation and peripheral lobe involvement were signi cantly higher in survivors.Chen et al. found that pulmonary involvement was mostly as bilateral pneumonia followed by GGO lesions (10).Francone et al. showed that the most common view observed on CT scan (less than 7 days from the onset of symptoms) was the GGO; and after 7 days, crazy paving, consolidation, and brosis were the most common views, respectively (20).In a study by Huang et  pneumothorax are very rare but may be seen as the disease progresses.Imaging patterns related to clinical improvement usually occur after 2 weeks of illness and include the gradual removal of opacities and the decrease in the number of lesions and involved lobes (21).
In this study, the means ± SD of WBC, blood sugar, urea, LDH, aspartate transaminase (AST), serum potassium, phosphorus, and ESR were higher in non-survivors, but the mean ± SD of lymphocytes, serum potassium, calcium, and albumin were higher in survivors.In Huang et al.'s study, laboratory features in patients with COVID-19 included leukopenia (25%), lymphopenia (25%), and increased AST (37%) (6).Zhang et al. also found that prothrombin and D-dimer levels were higher in patients with ICU than (17).In a meta-analysis, Lippi et al. reported that in almost all patients with COVID-19 (99%), the troponin level increased to the maximum normal range.In addition, troponin levels highly increased in patients with severe infection than those with milder disease; and it could predict the likelihood of heart damage and disease progression toward worse clinical signs, and protective cardiac treatments may be helpful in these patients (23).In another systematic review and meta-analysis, it was reported that out of 4,663 patients, the most common laboratory nding related to COVID-19 was C-reactive protein (CRP), followed by decreased albumin, increased ESR, decreased eosinophil, increased interleukin 6, decreased lymphocyte count, and nally increased LDH, respectively.Their meta-analytic ndings on 1905 patients also showed that the increased CRP and LDH levels, as well as decreased lymphocyte in the patients' blood samples, would be signi cantly associated with increased disease severity and mortality (25).
Our ndings demonstrated that the most treatments performed were Kaletra, hydroxychloroquine, and oseltamivir, respectively.In non-survivors, vitamin D3, antibiotics, tavanex, corticosteroids, non-invasive mechanical ventilation, invasive mechanical ventilation, and RRT were further used.However, hydroxychloroquine was more administered to survivors.Zhou et al. stated antibiotics and corticosteroids as the most commonly used treatments (2).Moreover, Chen et al. mentioned the most commonly used treatments for patients as antiviral drugs, oxygen therapy, and antibiotics, but found no evidence of their effectiveness (10).Various treatments have been suggested for patients over time, which the reason for the observed differences may be due to the increased knowledge and experience of physicians regarding drugs effectiveness in the treatment of COVID-19 and its complications.
The most common complications observed in our studied patients included respiratory failure, sepsis, and acidosis, respectively.All complications were more common in non-survivors.Zhou et al. mentioned sepsis, respiratory failure, ARDS, and heart failure as the most common complications, all of which were signi cantly higher in non-survivors.The results regarding the difference between the onset of symptoms and the onset of complications in our study were similar to the obtained results by Zhou et al.'s(2).In Chen et al.'s study, ARDS was observed in 17% of patients, which was the most common complication (10).
Hospital readmission was observed in 5.2% of patients in the present study, which was higher in nonsurvivors.ICU admission was observed in 20.5% of patients, which was higher in non-survivors.In Zhou et al.'s study, 26% of patients were admitted to the ICU, which was signi cantly higher in non-survivors (2).Also, in our study, the mean of hospital stay were higher in patients with in-hospital mortality, but it was lower in deceased patients in Zhou et al.'s study.However, similar to their results (2), the duration of ICU admission in patients of both groups was not signi cantly different in the present research.Also, in line with their results, the means of the time between the onset of clinical symptoms and outcome were more common in non-survivors (2).
The results of the present study showed that 76.7% of patients with ICU admission died, which was signi cantly higher than survivors.In the study by Auld et al., the mortality rate of patients with ICU admission was reported as 33.9%, which was lower than the result obtained in the present study.This rate was reported as 52-62% in other similar studies (28).Another study in the United States found that 50-67% of patients with ICU admission died (29).
The results of this study showed that plural effusion in lung CT scan, WBC, albumin, non-invasive mechanical ventilation, and ARDS were the predictive factors for in-hospital mortality in patients with COVID-19.Wang et al. found that CRP could be a valuable marker for predicting the likelihood of exacerbation of the disease in adult patients with non-severe COVID-19 (24).By examining the clinical ndings of 82,719 patients with coronavirus that resulted in the treatment of 4632 patients who died, Deng et al. considered old age and male gender as risk factors for mortality.It was also observed that the time from the onset of symptoms to the treatment center, the time from the onset of symptoms to laboratory con rmation of COVID-19, and the duration of onset of symptoms to the patients' hospitalization of were directly related to higher mortality (30).
The retrospective nature of the study, lack of recording all data accurately, and lack of follow-up of discharged patients were among the limitations of this study.On the other hand, the high sample size of patients and the study of various factors were among the strengths of this study.Using the results of the current study can be effective in physicians' clinical decisions and also policy makers.However, performing multicenter and prospective studies with larger sample sizes and assessing other factors, especially the effect of vaccination, as well as the drug doses and their complications, can be valuable.

Conclusions
This study showed that most inpatients were male.In-hospital mortality was obtained at about 16% and ICU admission was observed in about 20% of patients with COVID-19.Plural effusion in lung CT scan, WBC, albumin, non-invasive mechanical ventilation, and ARDS were obtained as predictive factors for in-hospital mortality in these patients.
AbbreviationsARDS acute respiratory distress syndrome AKI Acute kidney injury ALT alanine transaminase

Table 4
Complications and outcomes in patients with de nitive diagnosis of COVID-19 * Statistically signi cant; AKI: Acute kidney injury; ARDS: acute respiratory distress syndrome; SD: standard deviation * Statistically signi cant; AKI: Acute kidney injury; ARDS: acute respiratory distress syndrome; SD: standard deviation

Table 5
Results of multivariate logistic regression test in each category (medical ndings, paraclinical ndings, treatments, and complications) * Statistically signi cant; ARDS: acute respiratory distress syndrome; AKI: Acute kidney injury; CI: con dence interval; GCS: Glasgow coma scale; ICU: intensive care unit; OR: odds ratio; WBC: white blood cell

Table 6
Results of multivariate logistic regression test to determine the predictive factors for in-hospital mortality The results showed that the mean age of all patients with COVID-19 was 46.7 years, of which 51.8% were male (3))atistically signi cant; ARDS: acute respiratory distress syndrome; CI: con dence interval; OR: odds ratio; WBC: white blood cell The comparison of con rmed COVID-19 patients in regards of ICU and regular wards' admissions was shown in supplementary Table1.DiscussionThe current study investigated the factors affecting in-hospital mortality of patients with COVID-19 hospitalized in one of the main teaching hospital in central Iran.The results showed that the mean ± SD of age was 17.53 ± 56.29 years, and about 60% of inpatients were male.In line with our ndings, in a study by Chen et al., the mean age of inpatients was reported as 55 years, of which 67% were male(10).In another study by Wu et al., the highest mortality rate was reported in older men with underlying disease(3).During a systematic review and a meta-analysis, Li et al. epidemiologically investigated clinical features, risk factors, and treatment outcomes in patients with COVID-19.
Common symptoms in patients with COVID-19 were reported by Wei et al. as fever and cough (12).Zou et al. stated that the most common symptom present in patients included fever, cough, shortness of breath, hemoptysis, and diarrhea, respectively (15).In a meta-analysis, Cao et al. showed that the most prevalent clinical manifestations in patients with COVID-19 were fever, cough, shortness of breath, myalgia or fatigue, and respiratory distress al., 98% had bilateral lung involvement, and in general GGO was more common (6).Cao et al. mentioned the main ndings of imaging as bilateral pneumonia and GGO (18).Salehi et al. stated that one of the known features of COVID-19 in patients' early lung CT scans is GGO with peripheral or posterior distribution, mainly in the lower lobes and less in the middle lobe.Septal thickening, bronchiectasis, pleural thickening, and subpleural involvement are some of the less prevalent ndings that are mainly seen in the later stages of the disease.Pleural effusions, pericardial effusions, lymphadenopathy, cavitation, halo symptoms, and