Exploration on the inﬂuencing factors and establishment of a risk model for the time of virus turning negative of asymptomatic or mild Covid-19

Background. Patients are allowed to be released from quarantine after virus turns negative for asymptomatic or mild covid-19. Factors aﬀecting the negative conversion of nucleic acid are unknown. Methods. This study included asymptomatic or mild covid-19 patients quarantined in Shanghai shelter hospital from April to May 2022. They were tested daily for SARS-CoV-2 nucleic acid. The clinical characteristics were collected. Univariate and multivariate Cox regression were used to explore the relationship between clinical characteristics and the time of virus turning negative (TTN), ﬁnally a risk model was established. Results. A total of 7836 patients enrolled were divided into training or validation set randomly. In training set, age ([?]40 Y), symptoms (yes) , diabetes (yes) and vaccination status(yes) were correlated with TTN. We used HR values of the above factors in multivariate Cox regression as risk coeﬃcients to build a risk model for predicting TTN. TTN of patients with high risk score was longer than with low risk score. This risk model has been well validated in the validation set and the complete set. Conclusions. Our study found that age, symptoms, diabetes and vaccination status were correlated with the TTN,and we ﬁrst build a risk model to predict TTN in asymptomatic or mild Covid-19.


Introduction
It has been more than two years since Covid-19 appeared.According to the data of the World Health Organization, the SARS-CoV-2 has infected about 5.29 billion people and has caused more than 6 million deaths.More importantly, the Covid-19 has had a serious impact on the economy and life of the world, although many countries have announced that the pandemic of Covid-19 has passed, life has slowly returned to its pre infection state [1].However, there are still many disputes on this issue [2,3].In any case, the Covid-19 is still one of the main problems that the world needs to deal with.
After more than two years of evolution, the variants of concern (VOCs) of SARS-CoV-2 has changed from the original strain (Alpha) to the present Omicron(BA.2.x), and the characteristics of virus infection have changed from the initial severe symptoms and high mortality to the present mild symptoms, mainly mild and asymptomatic infection [4][5][6].The symptoms of upper respiratory tract infection are main, and the mortality is low.This is also the reason why many countries adopt an open strategy.However, some data show that although the Omicron strain causes mild symptoms and low mortality, it is more infectious and spreads rapidly among people [7].On this occasion, most of the patients are quarantined at home, and they will be required to put into normal work and life after the virus turns negative and the transmissibility is reduced [8,9].So, what are the factors related to the virus negative conversion of patients with asymptomatic or mild Omicron?And how to predict?At present, there are few studies have been reported.We selected asymptomatic and mild Covid-19 infected patients in Shanghai from April to May 2022, and these patients were admitted to Shanghai shelter hospital.These patients were tested for SARS-CoV-2 every day until they turned negative and were discharged.The aim of this study to explore the factors that affect the virus turning negative and establish the risk model.

1.Patients and groups
All the patients were mild or asymptomatic patients diagnosed as Covid-19 by nucleic acid test from April 13, 2022 to May 22, 2022.The mild and asymptomatic diagnostic criteria were based on the 9th edition of the guidelines for diagnosis and treatment of Covid-19 issued by the National Health Commission of the People's Republic of China.According to the report of the Shanghai Health Commission in early April 2022, the new SARS-CoV-2 strain in Shanghai is Omicron (BA.2.2).After removing some patients with incomplete information, a total of 7836 patients were enrolled in the study.Each patient was given a continuous admission number at the time of admission, and the age, sex, symptoms, basic diseases, Covid-19 vaccination and other data were recorded as options.The above information is collected at the time of admission.After admission, all patients were tested for nucleic acid of SARS-CoV-2 by throat swabs once a day until they were negative for two consecutive times, and the test interval was >24 hours.It was confirmed that the patients' SARS-CoV-2 turned negative and were discharged from the hospital.The time of virus turning negative (TTN) was defined as the time from the first positive virus test of the patient to the two consecutive negative virus tests after entering the shelter hospital.We mainly studied the influence of patients' age, sex, symptoms, basic diseases and vaccination status of Covid-19 on the TTN, and built a risk model that affected the TTN.In order to better verify the risk model,7836 patients were randomly divided into two groups, the training group and the validation group.There were 3951 patients in the training group and 3885 patients in the validation group.

2.Nucleic acid detection of SARS-CoV-2
After the patients were admitted to the shelter hospital, the nucleic acid of SARS-CoV-2 was detected once a day.Pharyngeal swabs were used for sample collection.The collected samples shall be submitted for examination within 2 hours.The testing institution was Shanghai Lanwei medical laboratory, and the SARS-CoV-2 nucleic acid detection kit (product No. 2019-ncov-100-02) of Wuhan Mingde biology was used.The detection method was RT-PCR, and ORFlab gene and N gene were detected.The threshold CT value [?] 35 is weak positive or negative, which is clinically defined as negative.

Statistical Analysis
All patients were randomly divided into two groups, training group and validation group.Kaplan-Meier method and univariate Cox regression were used to analyze the effects of age, gender, symptoms, hypertension, diabetes, other basic diseases and vaccination status of Covid-19 on TTN.Then the effects of age, sex, symptoms, hypertension, diabetes, other basic diseases and vaccination status of Covid-19 on TTN were analyzed by multivariate Cox regression analysis.According to the results of multivariate Cox regression analysis, the statistically significant indicators were found.According to the sum of HR values of the factors with significant differences in multivariate Cox regression analysis, a risk model for the TTN of Covid-19 was constructed, and the risk score of each patient was calculated.The high-risk group with a risk score higher than the average indicates a longer TTN, whereas the low-risk group with a risk score lower than the average indicates a shorter TTN.Then the constructed risk model is further verified in the validation set and the complete set.The study data were analyzed by SPSS 26 software, and GraphPad Prism 9 was used for forest plot and Kaplan-Meier curve.

Ethics
This study was approved by the ethics committee of Affiliated Jinling Hospital ,Medical school of Nanjing University.

Characteristics of patients
From April 13, 2022 to May 22, 2022, 7836 patients with mild or asymptomatic Covid-19 were enrolled in the study.Among them, 4610 (58.83%) were male and 3226 (41.17%) were female.3111 (39.70%) were under 40 years old, and 4725 (60.30%) were over 40 years old.Most of them were asymptomatic patients 7019 (89.57%).Most patients have no basic diseases.Hypertension and diabetes were the main basic diseases, 599 (7.64%) and 258 (3.29%) respectively.In the history of vaccination, most patients received the Covid-19 vaccine, of which 174 (2.22%) received only one vaccination, 1920 (24.50%) received two vaccinations, and 2615 (33.37%) received three vaccinations.There are still 3127 (39.91%) patients who have not been vaccinated against Covid-19.Among all patients, the average TTN of Covid-19 was 8.65 days, including 821 (10.48%) patients who had not turned negative for a long time ([?] 14 days), and most patients turned negative within two weeks.According to the simple random method, all 7836 patients were divided into two sets, 3951 in the training set and 3885 in the validation set.The proportions of gender, age, symptoms, basic diseases, hypertension, diabetes, vaccination status and TTN of the two groups are similar.See Table 1 for details.
Table3 describes the basic diseases of patients in detail.Among all the patients, 899 had basic diseases, of which 559 (62.18%) had hypertension, followed by 258 (28.70%) with diabetes.The basic diseases of other systems, including digestive system diseases, heart diseases, lung diseases and nervous system diseases, accounted for 47 (5.23%), 14 (1.56%), 23 (2.56%) and 10 (1.11%) respectively.See Table 3 for details of basic diseases in training set and validation set.
Table4 describes in detail the types and brands of Covid-19 vaccines.Covid-19 vaccines mainly include inactivated vaccine, adenovirus vaccine and recombinant protein vaccine.Among them, inactivated vaccines accounted for the vast majority of 4577 (97.2%), while adenovirus vaccines and recombinant protein vaccines accounted for only 83 (1.76%) and 53 (1.13%).Among the inactivated vaccines, the inactivated vaccines of Kexing biology and Beijing biology accounted for the vast majority, which were 3209 (68.15%) and 1180 (25.06%) respectively.See Table4 for the type and brand of vaccination in the training set and validation set.

Factors affecting the TTN
In order to explore the factors influencing the TTN of asymptomatic and mild Covid-19,we used univariate Cox regression to analyze the correlation between age, gender, symptoms, hypertension, diabetes, other basic diseases, vaccination and TTN.The study found that in the training set, the patient's age ([?]40 years old), symptoms (yes), hypertension (yes), diabetes (yes), other basic diseases (yes) and vaccination status(yes) (HR=1.263,P =0.000; HR=1.174,P =0.002; HR=1.158,P =0.015; HR=1.346,P =0.001; HR=1.159,P =0.023; HR=0.937,P =0.047, respectively) were significantly correlated with the TTN.There was no significant correlation between gender and the TTN.From the univariate analysis results, it can be seen that age ([?]40 years old), symptoms (yes), hypertension (yes), diabetes (yes) and other basic diseases (yes) can lead to the extension of the time to negative.However, vaccination status (yes) can shorten the TTN (Table5 and Figure 1A).We further used multivariate Cox regression to analyze the correlation between age, gender, symptoms, hypertension, diabetes, other basic diseases, vaccination status and the TTN.The results showed that age ([?]40 years old), symptoms (yes) were significantly correlated with the TTN (HR =1.273, P =0.000; HR=1.247,P =0.000).While diabetes (yes) and vaccination status (yes) were approximately significantly correlated with the TTN (HR=1.200,P =0.058; HR=0.938,P =0.054).However, gender, hypertension and other basic diseases were not significantly correlated with the TTN (Table5 and figure 1B).

Construction of risk model affecting TTN
In the training set, we found that age ([?]40 years old), symptoms (yes) were significantly correlated with TTN (HR=1.273,P =0.000; HR=1.247,P =0.000) .However, diabetes (yes) and vaccination status (yes) were approximately significantly correlated with TTN(HR=1.200,P =0.058; HR=0.938,P =0.054).In order to better predict TTN in asymptomatic and mild Covid-19 infected patients, we used age, symptoms, diabetes and vaccination status to build a risk model.The risk coefficients of age, symptoms, diabetes and vaccination were defined according to the HR value of multivariate Cox regression analysis (Table 6).And then, the risk coefficients of the four clinical characteristics of each patient are added to obtain the risk score of the patient.In the training set, the average risk score of 3951 patients was 4.160 (3.938-4.720),and the risk score lower than 4.160 was defined as the low-risk group, indicating that the TTN was short.However, those with a risk score greater than 4.160 were defined as high-risk group, indicating a longer TTN.
In the training set, we divided the patients into low-risk group and high-risk group according to the risk score, including 1326 in the low-risk group and 2625 in the high-risk group.Kaplan-Meier method was used to analyze the difference of TTN between low-risk group and high-risk group.The results showed that there was a significant difference in TTN between the low-risk group and the high-risk group in the training set (Log rank test P =0.000)(Figure 2).The median TTN in the low-risk group was 8 days (95%CI: 7.683-8.317),while the median TTN in the high-risk group was 9 days (95%CI: 8.814-9.186).

Validation of risk model
In order to further validate the TTN risk assessment model for patients infected with Covid-19, we use Kaplan-Meier method to validate the model in the validation set and the complete set.In the validation set, the mean risk score of 3885 patients was 4.162 (3.938-4.720).The low-risk group was defined as the risk score lower than 4.162, and the high-risk group was defined as the risk score higher than 4.162.Among them, 1290 were in the low-risk group and 2595 in the high-risk group.Kaplan-Meier method was used to analyze the difference of TTN between low-risk group and high-risk group.The result shows that there is a significant difference in TTN between the low-risk group and the high-risk group in the validation set (Log rank test P =0.000) (Figure 3A).The median TTN in the low-risk group is 8 days (95%CI: 7.657-8.343),while the median TTN in the high-risk group is 9 days (95%CI: 8.820-9.180).
In the complete set, the average risk score of 7836 patients was 4.161 (3.938-4.720).The low-risk group was defined as the risk score lower than 4.161, and the high-risk group was defined as the risk score higher than 4.161.Among them, 2616 were in the low-risk group and 5220 in the high-risk group.Kaplan-Meier method was used to analyze the difference of TTN between low-risk group and high-risk group.The result shows that there is a significant difference in TTN between the low-risk group and the high-risk group in the complete set(Log rank test P =0.000) (Fig 3B).The median TTN in the low-risk group is 8 days (95%CI: 7.767-8.233),while the median TTN in the high-risk group is 9 days (95%CI: 8.871-9.129).

Discussion
This project mainly studies asymptomatic or mild Covid-19 patients who were admitted to the shelter in Shanghai, China,from April to May 2022.In Shanghai, the virus strain of this wave of Covid-19 epidemic is mainly Omicron (BA.2.2) [10,11].We studied the factors influencing the time of virus turning negative in these asymptomatic or mild Covid-19 infected patients.The study found that age, symptoms, diabetes and vaccination can all affect TTN.It is suggested that age [?]40 years old, symptomatic and diabetes predict the extension of the TTN, while vaccination status predicts a short TTN.On this basis, we build a risk model to predict TTN according to the HR value of the multivariate Cox regression of the four factors.TTN of low-risk group was significantly shorter than that of high-risk patients.And we have well verified this risk model in the validation set and the whole population.For the first time, we proposed a risk prediction model for the time of turning negative in patients with asymptomatic or mild Covid-19, which is of great clinical significance.
Our results show that old age, symptomatic and diabetes are high risk factors for prolonged TTN.However, TTN was shortened when the vaccine was received.This further suggests the importance of Covid-19 vaccination.Some studies have shown that although the Covid-19 vaccine can not reduce the infection rate of Omicron, it can significantly reduce the incidence and mortality of severe diseases [12][13][14][15].Our study further supplemented that the vaccination of Covid-19 vaccine can also shorten the time of nucleic acid negative conversion in mild or asymptomatic patients.
There are many factors that affect TTN of patients with mild or asymptomatic Covid-19 infection.At present, there are few related studies.Our research shows that besides vaccination status, there are also age, symptoms, basic diseases of diabetes, etc.How to combine so many factors?We use the HR of each factor obtained by multivariate Cox regression analysis to build a risk model of TTN.In this way, many influencing factors can be well combined.Each patient can get a risk score according to the risk model, so as to predict the length of TTN.This has important clinical value.
After more than two years of pandemic in the world, the SARS-CoV-2 has mutated from the original strain to the Omicron strain, and its clinical characteristics are different from the initial severe symptoms and high mortality.Now the Omicron strain is mainly asymptomatic or mild in clinical manifestations, in which the symptoms have also been dominated by upper respiratory tract infection symptoms and the mortality is lower.Our study also confirmed that the most common symptom is cough.Upper respiratory tract symptoms such as expectoration, sore throat, stuffy nose and runny nose are the main symptoms.Therefore, many countries have declared the end of the Covid-19 pandemic.However, studies have shown that although the pathogenicity of Omicron has decreased, its transmissibility has increased, which will still affect people's work and life.Government departments in many countries suggest that these patients with mild or asymptomatic infection should be quarantined at home to isolate the transmission of the virus [16][17][18][19][20].However, it is very important to determine the quarantine time based on the time virus turns negative.In this wave of epidemic in Shanghai, because there are so many infected people, the government adopts the shelter isolation method for those with mild or asymptomatic infections.And the daily SARS-CoV-2 detection provides a very good opportunity for us to focus on the negative time of patients with mild or asymptomatic Covid-19.Our results can provide an important basis for the negative conversion of mild or asymptomatic Covid-19 persons isolated at home in other countries.
This study is a retrospective study, and it still needs prospective studies to further verify the risk model of the turning negative time.And with the variation of virus, the coefficient of risk model will change to some extent.However, the method and process of our study also have important reference value for possible mutants in the future.
Our study found that age, symptoms, diabetes and vaccination of asymptomatic or mild Covid-19 were significantly correlated with the TTN.It is the first time to build a risk model for TTN, which provides an important prediction tool for the time of nucleic acid turning negative in asymptomatic or mild Covid-19.

Financial support.
There was no fund for this study.

Figure 1
Figure 1 Forest plots present analysis of univariate Cox regression(A)and multivariate Cox regression (B) between clinical characteristics and TTN.Figure A indicated age ([?]40 years old), symptoms (yes), hypertension (yes), diabetes (yes), other basic diseases (yes) and vaccination(yes) were correlated with the TTN.There was no significant correlation between gender and the TTN.Figure B indicated that only age ([?]40 years old), symptoms (yes) were significantly correlated with the TTN .
Figure 1 Forest plots present analysis of univariate Cox regression(A)and multivariate Cox regression (B) between clinical characteristics and TTN.Figure A indicated age ([?]40 years old), symptoms (yes), hypertension (yes), diabetes (yes), other basic diseases (yes) and vaccination(yes) were correlated with the TTN.There was no significant correlation between gender and the TTN.Figure B indicated that only age ([?]40 years old), symptoms (yes) were significantly correlated with the TTN .
Figure 1 Forest plots present analysis of univariate Cox regression(A)and multivariate Cox regression (B) between clinical characteristics and TTN.Figure A indicated age ([?]40 years old), symptoms (yes), hypertension (yes), diabetes (yes), other basic diseases (yes) and vaccination(yes) were correlated with the TTN.There was no significant correlation between gender and the TTN.Figure B indicated that only age ([?]40 years old), symptoms (yes) were significantly correlated with the TTN .

Figure. 2
Figure.2 Kaplan-Meier analysis of TTN according to the risk score in the training set(low risk: the risk score lower than 4.160 ;high risk: a risk score greater than 4.160).The results showed that there was longer TTN in high-risk group than the low-risk group in the training set (Log rank test P =0.000).

Figure 3
Figure3Kaplan-Meier analysis of TTN according to the risk score in the validation set(A) and the complete set(B) (low risk: the risk score lower than 4.160 ;high risk: a risk score greater than 4.160).In both sets ,TTN in high-risk group was longer than the low-risk group (both sets:Log rank test P =0.000).Hosted filetables.docxavailable at https://authorea.com/users/515282/articles/590703-exploration-onthe-influencing-factors-and-establishment-of-a-risk-model-for-the-time-of-virus-turningnegative-of-asymptomatic-or-mild-covid-19

Table 1
Baseline characteristics of patients

Table 5
Univariate and multivariate Cox regression for TTN

Table 6
Risk coefficient in risk model of TTN Characteristics Age(year) Age(year) Symptom Symptom Diabetes Diabetes Vaccination status Vaccination TTN: the time of Covid-19 turned negative