Dynamic and classifier-based model SARS-CoV-2 Omicron spillover risk assessment in China

The coronavirus disease 2019 (COVID-19) continues to have a huge impact on health care and economic systems around the world. The first question to ponder is to understand the flow of COVID-19 in the spatial and temporal dimensions. We collected 7 Omicron clusters outbreaks in China since the outbreak of COVID-19 as of August 2022, selected outbreak cases from different Provinces and cities, and collected variable indicators that affect spillover outcomes, such as distance, migration index, PHSM index, daily reported cases number and so on. First, variables influencing spillover outcome events were assessed and analyzed retrospectively by constructing an infectious disease dynamics model and a classifier model, and secondly, the association between explanatory variables and spillover outcome events was constructed by fitting a logistics function. This study incorporates 7 influencing factors and classifies the spillover risk level into 3 levels. If different outbreak sites could be classified into different levels of spillover, it may reduce the pressure of epidemic prevention in some cities due to the lack of a uniform standard, which might be more conducive to achieving the goal of "dynamic zero".

Reproduction number of SARS-CoV-2 omicron variants, China, December 2022-January 2023.

China adjusted the zero-COVID strategy in late 2022, triggering an unprecedented Omicron wave. We estimated the time-varying reproduction numbers of 32 provincial-level administrative divisions from December 2022 to January 2023. We found that the pooled estimate of initial reproduction numbers is 4.74 (95% CI: 4.41, 5.07).

Impact of public health and social measures on contact dynamics during a SARS-CoV-2 Omicron variant outbreak in Quanzhou, China, March to April 2022.

OBJECTIVES: To evaluate the impact of early implementation of public health and social measures (PHSMs) on contact rates over time and explore contact behavior of asymptomatic versus symptomatic cases. METHODS: We used the largest contact tracing data in China thus far to estimate the mean contacts over time by age groups and contact settings. We used bootstrap with replacement to quantify the uncertainty of contact matrixes. The Pearson correlation was performed to demonstrate the number of contacts over time in relation to the evolution of restrictions. In addition, we analyzed the index cases with a high number of contacts and index cases that produced a high number of secondary cases. RESULTS: Rapidly adapted PHSMs can reduce the mean contact rates in public places while increasing the mean contact rates within households. The mean contact rates were 11.81 (95% confidence interval, 11.61-12.01) for asymptomatic (at the time of investigation) cases and 6.70 (95% confidence interval, 6.54-6.87) for symptomatic cases. The percentage of asymptomatic cases (at the time of investigation) meeting >50 close contacts make up more than 65% of the overall cases. The percentage of asymptomatic cases producing >10 secondary cases account for more than 80% of the overall cases. CONCLUSION: PHSMs may increase the contacts within the household, necessitating the need for pertinent prevention strategies at home. Asymptomatic cases can contribute significantly to Omicron transmission. By making asymptomatic people aware that they are already contagious, hence limiting their social contacts, it is possible to lower the transmission risk.

Simulating potential outbreaks of Delta and Omicron variants based on contact-tracing data: A modelling study in Fujian Province, China.

Although studies have compared the relative severity of Omicron and Delta variants by assessing the relative risks, there are still gaps in the knowledge of the potential COVID-19 burden these variations may cause. And the contact patterns in Fujian Province, China, have not been described. We identified 8969 transmission pairs in Fujian, China, by analyzing a contact-tracing database that recorded a SARS-CoV-2 outbreak in September 2021. We estimated the waning vaccine effectiveness against Delta variant infection, contact patterns, and epidemiology distributions, then simulated potential outbreaks of Delta and Omicron variants using a multi-group mathematical model. For instance, in the contact setting without stringent lockdowns, we estimated that in a potential Omicron wave, only 4.7% of infections would occur in Fujian Province among individuals aged >60 years. In comparison, 58.75% of the death toll would occur in unvaccinated individuals aged >60 years. Compared with no strict lockdowns, combining school or factory closure alone reduced cumulative deaths of Delta and Omicron by 28.5% and 6.1%, respectively. In conclusion, this study validates the need for continuous mass immunization, especially among elderly aged over 60 years old. And it confirms that the effect of lockdowns alone in reducing infections or deaths is minimal. However, these measurements will still contribute to lowering peak daily incidence and delaying the epidemic, easing the healthcare system's burden.

A comparative study on the three calculation methods for reproduction numbers of COVID-19.

Objective: This study uses four COVID-19 outbreaks as examples to calculate and compare merits and demerits, as well as applicational scenarios, of three methods for calculating reproduction numbers. Method: The epidemiological characteristics of the COVID-19 outbreaks are described. Through the definition method, the next-generation matrix-based method, and the epidemic curve and serial interval (SI)-based method, corresponding reproduction numbers were obtained and compared. Results: Reproduction numbers (R eff ), obtained by the definition method of the four regions, are 1.20, 1.14, 1.66, and 1.12. Through the next generation matrix method, in region H R eff = 4.30, 0.44; region P R eff = 6.5, 1.39, 0; region X R eff = 6.82, 1.39, 0; and region Z R eff = 2.99, 0.65. Time-varying reproduction numbers (R t ), which are attained by SI of onset dates, are decreasing with time. Region H reached its highest R t = 2.8 on July 29 and decreased to R t < 1 after August 4; region P reached its highest R t = 5.8 on September 9 and dropped to R t < 1 by September 14; region X had a fluctuation in the R t and R t < 1 after September 22; R t in region Z reached a maximum of 1.8 on September 15 and decreased continuously to R t < 1 on September 19. Conclusion: The reproduction number obtained by the definition method is optimal in the early stage of epidemics with a small number of cases that have clear transmission chains to predict the trend of epidemics accurately. The effective reproduction number R eff , calculated by the next generation matrix, could assess the scale of the epidemic and be used to evaluate the effectiveness of prevention and control measures used in epidemics with a large number of cases. Time-varying reproduction number R t , obtained via epidemic curve and SI, can give a clear picture of the change in transmissibility over time, but the conditions of use are more rigorous, requiring a greater sample size and clear transmission chains to perform the calculation. The rational use of the three methods for reproduction numbers plays a role in the further study of the transmissibility of COVID-19.

A comparative study on epidemiological characteristics, transmissibility, and pathogenicity of three COVID-19 outbreaks caused by different variants.

OBJECTIVES: The Omicron BA.2 variant is probably the main epidemic strain worldwide at present. Comparing the epidemiological characteristics, transmissibility, and influencing factors of SARS-CoV-2, the results obtained in this paper will help to provide theoretical support for disease control. METHODS: This study was a historical information analysis, using the R programming language and SPSS 24.0 for statistical analysis. The Geoda and Arc GIS were used for spatial autocorrelation analysis. RESULTS: Local spatial autocorrelations of the incidence rate were observed in Delta and Omicron BA.1 outbreaks, whereas Omicron BA.2 outbreaks showed a random distribution in incidence rate. The time-dependent reproduction number of Delta, Omicron BA.1, and Omicron BA.2 were 3.21, 4.29, and 2.96, respectively, and correspondingly, the mean serial interval were 4.29 days (95% confidence interval [CI]: 0.37-8.21), 3.84 days (95% CI: 0-8.37), and 2.77 days (95% CI: 0-5.83). The asymptomatic infection rate of cases in Delta, Omicron BA.1, and Omicron BA.2 outbreaks were 21.71%, 6.25%, and 4.35%, respectively. CONCLUSION: The Omicron BA.2 variant had the greatest serial interval, transmissibility, and transmission speed, followed by BA.1, and then Delta. Compared with Delta and Omicron BA.1 variants, the Omicron BA.2 variant may be less pathogenic and more difficult to control than Omicron BA.1 and Delta.

A comparative study on the three calculation methods for reproduction numbers of COVID-19

Objective This study uses four COVID-19 outbreaks as examples to calculate and compare merits and demerits, as well as applicational scenarios, of three methods for calculating reproduction numbers. Method The epidemiological characteristics of the COVID-19 outbreaks are described. Through the definition method, the next-generation matrix-based method, and the epidemic curve and serial interval (SI)-based method, corresponding reproduction numbers were obtained and compared. Results Reproduction numbers (Reff), obtained by the definition method of the four regions, are 1.20, 1.14, 1.66, and 1.12. Through the next generation matrix method, in region H Reff = 4.30, 0.44;region P Reff = 6.5, 1.39, 0;region X Reff = 6.82, 1.39, 0;and region Z Reff = 2.99, 0.65. Time-varying reproduction numbers (Rt), which are attained by SI of onset dates, are decreasing with time. Region H reached its highest Rt = 2.8 on July 29 and decreased to Rt < 1 after August 4;region P reached its highest Rt = 5.8 on September 9 and dropped to Rt < 1 by September 14;region X had a fluctuation in the Rt and Rt < 1 after September 22;Rt in region Z reached a maximum of 1.8 on September 15 and decreased continuously to Rt < 1 on September 19. Conclusion The reproduction number obtained by the definition method is optimal in the early stage of epidemics with a small number of cases that have clear transmission chains to predict the trend of epidemics accurately. The effective reproduction number Reff, calculated by the next generation matrix, could assess the scale of the epidemic and be used to evaluate the effectiveness of prevention and control measures used in epidemics with a large number of cases. Time-varying reproduction number Rt, obtained via epidemic curve and SI, can give a clear picture of the change in transmissibility over time, but the conditions of use are more rigorous, requiring a greater sample size and clear transmission chains to perform the calculation. The rational use of the three methods for reproduction numbers plays a role in the further study of the transmissibility of COVID-19.

A follow up study of cycle threshold values of SARS-CoV-2 in Hunan Province, China.

Since the epidemic of the severe acute respiratory syndrome coronavirus 2 (SARS-COV-2), many governments have used reverse transcription polymerase chain reaction (RT-PCR) to detect the virus. However, there are fewer measures of CT values information based on RT-PCR results, and the relationship between CT values and factors from consecutive tests is not clear enough. So in this study, we analyzed the connection between CT values and the factors based on cohort data from Delta variant of SARS-CoV-2 in Hunan Province. Previous studies have showed that the mean age of the cases was 33.34 years (±18.72 years), with a female predominance (55.03%, n = 71), and the greatest proportion of clinical symptoms were of the common type (60.47%, n = 78). There were statistical differences between the N and ORF1ab genes in the CT values for the cases. Based on the analysis of the association between CT values and the factors, the lowest CT values were obtained for the unvaccinated, older and clinically symptomatic group at 3-10 days, the maximum peak of viral load occurred. Therefore, it is recommended to use patient information to focus on older, clinically symptomatic, unvaccinated patients and to intervene promptly upon admission.

Contact pattern, current immune barrier, and pathogen virulence determines the optimal strategy of further vaccination.

Background: The current outbreak of novel coronavirus disease 2019 has caused a serious disease burden worldwide. Vaccines are an important factor to sustain the epidemic. Although with a relatively high-vaccination worldwide, the decay of vaccine efficacy and the arising of new variants lead us to the challenge of maintaining a sufficient immune barrier to protect the population. Method: A case-contact tracking data in Hunan, China, is used to estimate the contact pattern of cases for scenarios including school, workspace, etc, rather than ordinary susceptible population. Based on the estimated vaccine coverage and efficacy, a multi-group vaccinated-exposed-presymptomatic-symptomatic-asymptomatic-removed model (VEFIAR) with 8 age groups, with each partitioned into 4 vaccination status groups is developed. The optimal dose-wise vaccinating strategy is optimized based on the currently estimated immunity barrier of coverage and efficacy, using the greedy algorithm that minimizes the cumulative cases, population size of hospitalization and fatality respectively in a certain future interval. Parameters of Delta and Omicron variants are used respectively in the optimization. Results: The estimated contact matrices of cases showed a concentration on middle ages, and has compatible magnitudes compared to estimations from contact surveys in other studies. The VEFIAR model is numerically stable. The optimal controled vaccination strategy requires immediate vaccination on the un-vaccinated high-contact population of age 30-39 to reduce the cumulative cases, and is stable with different basic reproduction numbers ( R 0 ). As for minimizing hospitalization and fatality, the optimized strategy requires vaccination on the un-vaccinated of both aged 30-39 of high contact frequency and the vulnerable older. Conclusion: The objective of reducing transmission requires vaccination in age groups of the highest contact frequency, with more priority for un-vaccinated than un-fully or fully vaccinated. The objective of reducing total hospitalization and fatality requires not only to reduce transmission but also to protect the vulnerable older. The priority changes by vaccination progress. For any region, if the local contact pattern is available, then with the vaccination coverage, efficacy, and disease characteristics of relative risks in heterogeneous populations, the optimal dose-wise vaccinating process will be obtained and gives hints for decision-making.

Optimal control strategies of SARS-CoV-2 Omicron supported by invasive and dynamic models.

BACKGROUND: There is a raising concern of a higher infectious Omicron BA.2 variant and the latest BA.4, BA.5 variant, made it more difficult in the mitigation process against COVID-19 pandemic. Our study aimed to find optimal control strategies by transmission of dynamic model from novel invasion theory. METHODS: Based on the public data sources from January 31 to May 31, 2022, in four cities (Nanjing, Shanghai, Shenzhen and Suzhou) of China. We segmented the theoretical curves into five phases based on the concept of biological invasion. Then, a spatial autocorrelation analysis was carried out by detecting the clustering of the studied areas. After that, we choose a mathematical model of COVID-19 based on system dynamics methodology to simulate numerous intervention measures scenarios. Finally, we have used publicly available migration data to calculate spillover risk. RESULTS: Epidemics in Shanghai and Shenzhen has gone through the entire invasion phases, whereas Nanjing and Suzhou were all ended in the establishment phase. The results indicated that Rt value and public health and social measures (PHSM)-index of the epidemics were a negative correlation in all cities, except Shenzhen. The intervention has come into effect in different phases of invasion in all studied cities. Until the May 31, most of the spillover risk in Shanghai remained above the spillover risk threshold (18.81-303.84) and the actual number of the spillovers (0.94-74.98) was also increasing along with the time. Shenzhen reported Omicron cases that was only above the spillover risk threshold (17.92) at the phase of outbreak, consistent with an actual partial spillover. In Nanjing and Suzhou, the actual number of reported cases did not exceed the spillover alert value. CONCLUSIONS: Biological invasion is positioned to contribute substantively to understanding the drivers and mechanisms of the COVID-19 spread and outbreaks. After evaluating the spillover risk of cities at each invasion phase, we found the dynamic zero-COVID strategy implemented in four cities successfully curb the disease epidemic peak of the Omicron variant, which was highly correlated to the way to perform public health and social measures in the early phases right after the invasion of the virus.

Mathematical Models Supporting Control of COVID-19.

Mathematical models have played an important role in the management of the coronavirus disease 2019 (COVID-19) pandemic. The aim of this review is to describe the use of COVID-19 mathematical models, their classification, and the advantages and disadvantages of different types of models. We conducted subject heading searches of PubMed and China National Knowledge Infrastructure with the terms "COVID-19," "Mathematical Statistical Model," "Model," "Modeling," "Agent-based Model," and "Ordinary Differential Equation Model" and classified and analyzed the scientific literature retrieved in the search. We categorized the models as data-driven or mechanism-driven. Data-driven models are mainly used for predicting epidemics, and have the advantage of rapid assessment of disease instances. However, their ability to determine transmission mechanisms is limited. Mechanism-driven models include ordinary differential equation (ODE) and agent-based models. ODE models are used to estimate transmissibility and evaluate impact of interventions. Although ODE models are good at determining pathogen transmission characteristics, they are less suitable for simulation of early epidemic stages and rely heavily on availability of first-hand field data. Agent-based models consider influences of individual differences, but they require large amounts of data and can take a long time to develop fully. Many COVID-19 mathematical modeling studies have been conducted, and these have been used for predicting trends, evaluating interventions, and calculating pathogen transmissibility. Successful infectious disease modeling requires comprehensive considerations of data, applications, and purposes.

Assessing the mechanism of citywide test-trace-isolate Zero-COVID policy and exit strategy of COVID-19 pandemic.

BACKGROUND: Countries that aimed for eliminating the cases of COVID-19 with test-trace-isolate policy are found to have lower infections, deaths, and better economic performance, compared with those that opted for other mitigation strategies. However, the continuous evolution of new strains has raised the question of whether COVID-19 eradication is still possible given the limited public health response capacity and fatigue of the epidemic. We aim to investigate the mechanism of the Zero-COVID policy on outbreak containment, and to explore the possibility of eradication of Omicron transmission using the citywide test-trace-isolate (CTTI) strategy. METHODS: We develop a compartmental model incorporating the CTTI Zero-COVID policy to understand how it contributes to the SARS-CoV-2 elimination. We employ our model to mimic the Delta outbreak in Fujian Province, China, from September 10 to October 9, 2021, and the Omicron outbreak in Jilin Province, China for the period from March 1 to April 1, 2022. Projections and sensitivity analyses were conducted using dynamical system and Latin Hypercube Sampling/ Partial Rank Correlation Coefficient (PRCC). RESULTS: Calibration results of the model estimate the Fujian Delta outbreak can end in 30 (95% confidence interval CI: 28-33) days, after 10 (95% CI: 9-11) rounds of citywide testing. The emerging Jilin Omicron outbreak may achieve zero COVID cases in 50 (95% CI: 41-57) days if supported with sufficient public health resources and population compliance, which shows the effectiveness of the CTTI Zero-COVID policy. CONCLUSIONS: The CTTI policy shows the capacity for the eradication of the Delta outbreaks and also the Omicron outbreaks. Nonetheless, the implementation of radical CTTI is challenging, which requires routine monitoring for early detection, adequate testing capacity, efficient contact tracing, and high isolation compliance, which constrain its benefits in regions with limited resources. Moreover, these challenges become even more acute in the face of more contagious variants with a high proportion of asymptomatic cases. Hence, in regions where CTTI is not possible, personal protection, public health control measures, and vaccination are indispensable for mitigating and exiting the COVID-19 pandemic.

Comparison of epidemiological characteristics and transmissibility of different strains of COVID-19 based on the incidence data of all local outbreaks in China as of March 1, 2022

Background The epidemiological characteristics and transmissibility of Coronavirus Disease 2019 (COVID-19) may undergo changes due to the mutation of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) strains. The purpose of this study is to compare the differences in the outbreaks of the different strains with regards to aspects such as epidemiological characteristics, transmissibility, and difficulties in prevention and control. Methods COVID-19 data from outbreaks of pre-Delta strains, the Delta variant and Omicron variant, were obtained from the Chinese Center for Disease Control and Prevention (CDC). Case data were collected from China's direct-reporting system, and the data concerning outbreaks were collected by on-site epidemiological investigators and collated by the authors of this paper. Indicators such as the effective reproduction number (Reff), time-dependent reproduction number (Rt), rate of decrease in transmissibility (RDT), and duration from the illness onset date to the diagnosed date (DID)/reported date (DIR) were used to compare differences in transmissibility between pre-Delta strains, Delta variants and Omicron variants. Non-parametric tests (namely the Kruskal-Wallis H and Mean-Whitney U tests) were used to compare differences in epidemiological characteristics and transmissibility between outbreaks of different strains. P < 0.05 indicated that the difference was statistically significant. Results Mainland China has maintained a “dynamic zero-out strategy” since the first case was reported, and clusters of outbreaks have occurred intermittently. The strains causing outbreaks in mainland China have gone through three stages: the outbreak of pre-Delta strains, the outbreak of the Delta variant, and outbreaks involving the superposition of Delta and Omicron variant strains. Each outbreak of pre-Delta strains went through two stages: a rising stage and a falling stage, Each outbreak of the Delta variant and Omicron variant went through three stages: a rising stage, a platform stage and a falling stage. The maximum Reff value of Omicron variant outbreaks was highest (median: 6.7;ranged from 5.3 to 8.0) and the differences were statistically significant. The RDT value of outbreaks involving pre-Delta strains was smallest (median: 91.4%;[IQR]: 87.30–94.27%), and the differences were statistically significant. The DID and DIR for all strains was mostly in a range of 0–2 days, with more than 75%. The range of duration for outbreaks of pre-Delta strains was the largest (median: 20 days, ranging from 1 to 61 days), and the differences were statistically significant. Conclusion With the evolution of the virus, the transmissibility of the variants has increased. The transmissibility of the Omicron variant is higher than that of both the pre-Delta strains and the Delta variant, and is more difficult to suppress. These findings provide us with get a more clear and precise picture of the transmissibility of the different variants in the real world, in accordance with the findings of previous studies. Reff is more suitable than Rt for assessing the transmissibility of the disease during an epidemic outbreak.

Natural history and cycle threshold values analysis of COVID-19 in Xiamen City, China.

Objective: This study elaborated the natural history parameters of Delta variant, explored the differences in detection cycle thresholds (Ct) among cases. Methods: Natural history parameters were calculated based on the different onset time and exposure time of the cases. Intergenerational relationships between generations of cases were calculated. Differences in Ct values of cases by gender, age, and mode of detection were analyzed statistically to assess the detoxification capacity of cases. Results: The median incubation period was 4 days; the detection time for cases decreased from 25 to 7 h as the outbreak continued. The average generation time (GT), time interval between transmission generations (TG) and serial interval (SI) were 3.6 ± 2.6 days, 1.67 ± 2.11 days and 1.7 ± 3.0 days. Among the Ct values, we found little differences in testing across companies, but there were some differences in the gender of detected genes. The Ct values continuous to decreased with age, but increased when the age was greater than 60. Conclusion: This epidemic was started from aggregation of factories. It is more reasonable to use SI to calculate the effective reproduction number and the time-varying reproduction number. And the analysis of Ct values can improve the positive detection rate and improve prevention and control measures.

Modeling Cross-Regional Transmission and Assessing the Effectiveness of Restricting Inter-Regional Population Movements in Controlling COVID-19 - Xi'an City, Shaanxi Province, China, 2021.

Introduction: The aim of this study was to construct an assessment method for cross-regional transmission of coronavirus disease 2019 (COVID-19) and to provide recommendations for optimizing measures such as interregional population movements. Methods: Taking Xi'an City as the example subject of this study's analysis, a Cross-Regional-Gravitational-Dynamic model was constructed to simulate the epidemic in each district of Xi'an under three scenarios of controlled population movement (Scenario 1: no intensive intervention; Scenario 2: blocking Yanta District on December 18 and blocking the whole region on December 23; and Scenario 3: blocking the whole region on December 23). This study then evaluated the effects of such simulated population control measures. Results: The cumulative number of cases for the three scenarios was 8,901,425, 178, and 474, respectively, and the duration of the epidemic was 175, 18, and 22 days, respectively. The real world prevention and control measures in Xi'an reduced the cumulative number of cases for its outbreak by 99.98% in comparison to the simulated response in Scenario 1; in contrast, the simulated prevention and control strategies set in Scenarios 2 (91.26%) and 3 (76.73%) reduced cases even further than the real world measures used in Xi'an. Discussion: The constructed model can effectively simulate an outbreak across regions. Timely implementation of two-way containment and control measures in areas where spillover is likely to occur is key to stopping cross-regional transmission.

Model-Based Evaluation of Transmissibility and Intervention Measures for a COVID-19 Outbreak in Xiamen City, China

Background In September 2021, there was an outbreak of coronavirus disease 2019 (COVID-19) in Xiamen, China. Various non-pharmacological interventions (NPIs) and pharmacological interventions (PIs) have been implemented to prevent and control the spread of the disease. This study aimed to evaluate the effectiveness of various interventions and to identify priorities for the implementation of prevention and control measures. Methods The data of patients with COVID-19 were collected from 8 to 30 September 2021. A Susceptible-Exposed-Infectious-Recovered (SEIR) dynamics model was developed to fit the data and simulate the effectiveness of interventions (medical treatment, isolation, social distancing, masking, and vaccination) under different scenarios. The effective reproductive number (Reff) was used to assess the transmissibility and transmission risk. Results A total of 236 cases of COVID-19 were reported in Xiamen. The epidemic curve was divided into three phases (Reff = 6.8, 1.5, and 0). Notably, the cumulative number of cases was reduced by 99.67% due to the preventive and control measures implemented by the local government. In the effective containment stage, the number of cases could be reduced to 115 by intensifying the implementation of interventions. The total number of cases (TN) could be reduced by 29.66–95.34% when patients voluntarily visit fever clinics. When only two or three of these measures are implemented, the simulated TN may be greater than the actual number. As four measures were taken simultaneously, the TN may be <100, which is 57.63% less than the actual number. The simultaneous implementation of five interventions could rapidly control the transmission and reduce the number of cases to fewer than 25. Conclusion With the joint efforts of the government and the public, the outbreak was controlled quickly and effectively. Authorities could promptly cut the transmission chain and control the spread of the disease when patients with fever voluntarily went to the hospital. The ultimate effect of controlling the outbreak through only one intervention was not obvious. The combined community control and mask wearing, along with other interventions, could lead to rapid control of the outbreak and ultimately lower the total number of cases. More importantly, this would mitigate the impact of the outbreak on society and socioeconomics.

Assessing the Impacts of Meteorological Factors on COVID-19 Pandemic Using Generalized Estimating Equations

Background Meteorological factors have been proven to affect pathogens;both the transmission routes and other intermediate. Many studies have worked on assessing how those meteorological factors would influence the transmissibility of COVID-19. In this study, we used generalized estimating equations to evaluate the impact of meteorological factors on Coronavirus disease 2019 (COVID-19) by using three outcome variables, which are transmissibility, incidence rate, and the number of reported cases. Methods In this study, the data on the daily number of new cases and deaths of COVID-19 in 30 provinces and cities nationwide were obtained from the provincial and municipal health committees, while the data from 682 conventional weather stations in the selected provinces and cities were obtained from the website of the China Meteorological Administration. We built a Susceptible-Exposed-Symptomatic-Asymptomatic-Recovered/Removed (SEIAR) model to fit the data, then we calculated the transmissibility of COVID-19 using an indicator of the effective reproduction number (Reff). To quantify the different impacts of meteorological factors on several outcome variables including transmissibility, incidence rate, and the number of reported cases of COVID-19, we collected panel data and used generalized estimating equations. We also explored whether there is a lag effect and the different times of meteorological factors on the three outcome variables. Results Precipitation and wind speed had a negative effect on transmissibility, incidence rate, and the number of reported cases, while humidity had a positive effect on them. The higher the temperature, the lower the transmissibility. The temperature had a lag effect on the incidence rate, while the remaining five meteorological factors had immediate and lag effects on the incidence rate and the number of reported cases. Conclusion Meteorological factors had similar effects on incidence rate and number of reported cases, but different effects on transmissibility. Temperature, relative humidity, precipitation, sunshine hours, and wind speed had immediate and lag effects on transmissibility, but with different lag times. An increase in temperature may first cause a decrease in virus transmissibility and then lead to a decrease in incidence rate. Also, the mechanism of the role of meteorological factors in the process of transmissibility to incidence rate needs to be further explored.