Transmissibility of Hand, Foot, and Mouth Disease in 97 Counties of Jiangsu Province, China, 2015-2020

Background: Hand, foot, and mouth disease (HFMD) has been a serious disease burden in the Asia Pacic region represented by China, and the transmission characteristics of HFMD in regions haven’t been clear. This study calculated the transmissibility of HFMD at county levels in Jiangsu Province, China, analyzed the differences of transmissibility and explored the reasons. Methods: We built susceptible-exposed-infectious-asymptomatic-removed (SEIAR) model for seasonal characteristics of HFMD, estimated effective reproduction number (R eff ) by tting the incidence of HFMD in 97 counties of Jiangsu Province from 2015 to 2020, compared incidence rate and transmissibility in different counties by non -parametric test, rapid cluster analysis and rank-sum ratio. Results: The average daily incidence rate was between 0 and 4 per 100,000 in Jiangsu province from 2015-2020. The 97 counties could be divided into three levels: low incidence, medium incidence and high incidence, and occurred that the average daily incidence rate dropped sharply in 2016-2017, and increased sharply in 2017-2018 years. The Quartile of Reff in Jiangsu Province from 2015 to 2020 was 1.54 (0.49, 2.50), Rugao district in Central Jiangsu and Jianhu district in Northern Jiangsu had the highest transmissibility by rank-sum ratio. R eff generally decreased in 2017 and increased in 2018 in most counties, and the median level of R eff was lowest in 2017 (P<0.05). Conclusion: Transmissibility was different in 97 counties of Jiangsu Province, and the reasons for the differences may be related to the climate, demographic characteristics, virus subtypes, vaccination and other infectious diseases. (SEIAR) model for seasonal characteristics of HFMD, estimated effective reproduction number (R eff ) by tting the incidence of HFMD in 97 counties of Jiangsu Province from 2015 to 2020. The prevalence and transmissibility of HFMD in Jiangsu have regional and seasonal characteristics. The reasons for the differences may be related to the climate, demographic characteristics, virus subtypes, vaccination and other infectious diseases. This study provides a reference for the control of HFMD in different districts, and provides a new idea for the study of HFMD transmissibility in the future.


Introduction
Hand foot and mouth disease (HFMD) is an infectious disease caused by enteroviru, enterovirus 71(EV71) and coxsackievirus A16 (CA16) are the main pathogens causing HFMD worldwide 1 . The virus is mainly transmitted by fecal-oral route 2 . The disease mainly occurs in children under the age of 10, especially in school-age children aged 5-6 3 . The main manifestations of patients are low-grade fever, macular papules or papular rashes on the hands and soles of feet, and oral pain ulcers, among them which infected Cox A16 may be accompanied by some atypical lesions, such as large herpes and nodules on the trunk, limbs and face, neurological complications may occur when it's serious 4 .
HFMD is a serious public health problem. Since the rst report of HFMD in Canada in 1950s, HFMD has been popular all over the world, and Asia is a high incidence area of HFMD 5 . In recent years, the disease burden of HFMD in China, Singapore, Vietnam, Malaysia and Japan has been increasing, China had become the largest epidemic area of HFMD in Asia Paci c region in 2008, and it was classi ed as class C legal infectious disease which the incidence rate was the highest among all the diseases that be reported in China, and caused 2 million children to be infected each year 6 . However, up to now, there is no speci c effective treatment for HFMD, and the HFMD vaccine with better e cacy (94.8% -97.4%) is mainly for EV71 7 . Therefore, it is particularly important to study the incidence, transmission characteristics and in uencing factors of HFMD and nd appropriate prevention and control measures. The transmission dynamics model can be used to study the transmissibility and in uencing factors of HFMD, and susceptible-infectious-removed (SIR), time-susceptible-infectious-removed, susceptibleexposed-infectious-removed (SEIR) or SEIAR models have been mostly used to t and estimate the transmissibility of HFMD and its different serotypes [8][9][10][11][12] . Based on these models, some studies introduce isolation measures or environmental factors to build susceptible-exposed-infectious-hospitalizedremoved (SEIQR) and susceptible-exposed-infectious-asymptomatic-recovered-environment (SEIARW) models to evaluate the impact of isolation or environmental factor control on transmissibility [13][14][15] , to put forward targeted prevention and control measures. Most of these studies focused on a large region such as a country or a province, and the results of these studies didn't further explore the transmissibility and in uencing factors in different counties in country or province. However, studies have shown that weather can affect the spread of HFMD 16 , a study found that the relationship between climate and HFMD varies from region to region 17 , and climate couldn't fully explain the spread of HFMD, another research considered that isolation measures affect the epidemic peak of HFMD 18 , and semester, Spring Festival holiday and highway passenger volume were the main factors affecting the peak 19 . Therefore, it is necessary to further study the regional transmissibility of HFMD and explore its related in uencing factors, to provide more effective information for the actual prevention and control of HFMD in different counties.
In order to further explore the transmissibility and its in uencing factors of HFMD in different counties, we used the incidence data of 2015-2020 HFMD in Jiangsu Province. We built a seasonal SEIAR model to t the incidence rate of HFMD, then calculated the change of HFMD transmissibility of 97 counties in Jiangsu from 2015-2020. Finally, we compared the transmissibility of HFMD in three regions (Central Jiangsu, Northern Jiangsu and Southern Jiangsu) and 97 counties which were contained in three regions in Jiangsu Province, and analyzed the in uencing factors of the transmissibility, to provide a reference for controlling the outbreak of HFMD.

Data sources
Jiangsu province locates in the eastern coastal area of Chinese mainland, between the east longitude 116º18'-121º57', and the north latitude 30º45'-35º20'. The whole province is in the transition from subtropical zone to warm temperate zone, with mild climate and moderate rainfall. According to the statistical yearbook of Jiangsu Province, Jiangsu Province is divided into three regions.
The three regions are bounded by the Huaihe River and irrigation canal, it has subtropical humid monsoon climate in the South and a warm temperate humid and semi-humid monsoon climate in the north. The Yangtze River is the natural boundary of Jiangsu Province. Jiangsu province divides the south of the Yangtze River into Southern Jiangsu, and the north of the Huaihe River into Northern Jiangsu, while the area between the Yangtze River and the Huaihe River belongs to Central Jiangsu. The area of the three regions is similar, but there are obvious differences in social and economic development, the regional economy is best in Southern Jiangsu, then in Central Jiangsu and the last is in Northern Jiangsu 20 .

Case de nition
The diagnosis of HFMD was carried out according to the guide issued by the National Health and Family Planning Commission of the people's Republic of China 21 . The mild form of HFMD with or without fever was the most common form of HFMD, accompanied by neurological complications (aseptic meningitis, encephalitis, encephalomyelitis, acute accid paralysis or autonomic nervous system dysfunction) HFMD was con rmed by RT-PCR, real-time PCR or virus isolation using throat swabs or stool samples.
The transmission models of HFMD According to the epidemiological feature of HFMD and our previous studies 8,9,11 the SEIAR model could be used for the simulation in the model, the population was divided into susceptible individuals (S), exposed individuals (E), infectious individuals (I), asymptomatic individuals (A) and recovery individuals (R). The model diagram is shown in Figure S1.
The differential equations of the model were used to describe the dynamic changes of each state. The corresponding model equations were as follows: 1 The model assumed that HFMD cannot propagate vertically, so the new individuals born in all kinds of people are susceptible. Then we set birth rate (br), the natural mortality rate (dr), and the mortality rate of the infectious individuals (f). The mortality rate of all kinds of people in the disease spectrum is low, and the mortality rate of population attributable to HFMD is even lower, we set the mortality rate of the whole population as the sum of the mortality of the whole population and the mortality of HFMD.
2 Transmission of HFMD occurs via person-person, and the transmissibility between infectious individual and asymptomatic one may be different. So, the k was de ned as the relative transmissibility rate of asymptomatic to symptomatic individuals. At the same time, we assumed the S will be potentially infectious as long as they are in contact with infectious individuals or asymptomatic individuals, and the coe cient of the infection rate was set as β.
3 Infectious individuals (I) and asymptomatic individuals (A) came from the susceptible individuals, so we considered that there was a certain proportion of exposed individuals pE (0⩽p⩽1) transformed into I after incubation, another part of exposed individuals (1-p) E were transformed into A after incubation as well. At a certain time (t), the development speed from the E to I pathway is the same as the E to A pathway and we set the speed as ω (0≤ω≤1). So the proportional coe cient of E to I was set as pω, and E to A was set as (1 − p) ω.

Parameter estimation
The parameters β, ω, ω ', γ, γ', k, p and f represented the infection rate coe cient, incubation period coe cient, latent period coe cient, removal rate coe cient of dominant infection, removal rate coe cient of recessive infection, infectivity coe cient of recessive infection compared with dominant infection, the proportion of dominant infection and fatality rate of dominant infection respectively.
1) The birth rate (br) and death rate (dr) were collected from 97 counties' statistical yearbooks in Jiangsu Province.
4) The duration of symptomatic infection was 2 weeks 10,25 , therefore, the rate of disease removal γ = 0.0714. The duration of asymptomatic infection ranged from 2 to 4 weeks 22,23 , Median of 3 weeks was chosen as the disease removal rate of asymptomatic patients, therefore, γ'= 0.0476.

5)
The mortality of symptomatic infection ranged from 0.0001 to 0.0005 26,27 , selecting the median value 0.0003. Parameter β is estimated by curve tting.
6) There is no clear data or references to support the parameter κ, which is still uncertain. Therefore, in this study, we assumed κ = 1 for calculation, and sensitivity analysis was carried out to calculate its impact on the model.
The signi cance of each variable and parameter of the model is shown in Table 1.

Transmissibility evaluation index
In this study, the population was not completely susceptible and arti cially adopted some prevention and control measures, so we chose the effective reproduction number (R eff ) to calculate transmissibility. The calculation formula was as follows: Simulation methods and statistical analysis The coe cient of determination (R 2 ) was used to assess the goodness-of-t. SPSS 13.0 software (IBM Corp. Armonk, NY, USA) was used to calculate the R 2 . Non-parametric tests, fast clustering analysis and rank-sum ratio and linear regression analysis were used to further analyze the differences in different regions. Rank sum ratio process: the R eff values from 2015 to 2020 were divided into the mean value in the rst half of the year and the mean value in the second half of the year. Rank principle, the smaller the R eff was, the larger the rank was. The rank-sum ratio RSR was calculated by ranking. Probit was calculated through RSR distribution, and the regression equation of RSR and probit was constructed. The comprehensive comparison results of R eff in various regions are determined through the regression equation.

Result
County-level incidence map of HFMD in Jiangsu Province from 2015 to 2020 The average daily incidence of HFMD in various counties of Jiangsu Province was in the range from 0 per 100,000 to 4 per 100,000. In 2018, the median average daily incidence rate M (0.5 per 100,000) was the highest. In 2020, the median average daily incidence rate M (0.003 per 100,000) was the lowest. Comparing the average daily incidence rate in Jiangsu Province from 2015 to 2020 with that in 2009-2013 28 , except the average daily incidence rate in 2020 was smaller than in previous years, the average daily incidence rate in other years had a larger range and the highest daily average incidence rate was 6.67 times the highest in 2009-2013.
According to the incidence map ( Figure 1), we found that in 2020, the average daily incidence rate of three regions (Southern Jiangsu, Northern Jiangsu and Central Jiangsu) all was in the range from 0 per 100,000 to 0.5 per 100,000, however, from 2015 to 2019, the average daily incidence rate in Southern Jiangsu was generally more serious than that in Northern Jiangsu and Central Jiangsu. From 2015 to 2020, the average daily incidence rate of all counties in Central Jiangsu was in the range from 0 per 100,000 to 0.5 per 100,000, except for 2018, it has been well controlled in other years. In northern Jiangsu, with the exception of Huai'an, where the incidence rate was lower, the incidence rate in other counties showed an alternating trend of increase and decrease. According to Figure 2-4, we found that the HFMD outbreaks in Jiangsu Province showed obvious seasonality. The outbreaks in Southern Jiangsu were basically two seasons a year, and the peak height and duration of the two outbreaks were relatively consistent. The counties of Central Jiangsu were also basically two seasons a year. The peak height and duration of the outbreak in two seasons a year were relatively consistent, but the peak height of the outbreak in 2018 was signi cantly higher than that in other years. The outbreaks in Northern Jiangsu were more complex. the counties in three major cities (Huai'an, Lianyungang, and Suqian) showed a trend of seasonal outbreaks. While the counties in Yancheng city showed 2-3 outbreaks a year and the counties in the Xuzhou city showed a steady two-season outbreak.
According to the change of the average daily incidence rate in the region from 2015 to 2020, we could divide 97 counties into three typical situations by fast cluster analysis. The rst kind was that the average daily incidence was at a high level, basically maintained at 1 per 100,000. Gangzha District in Nantong City, Gongyeyuan District in Suzhou City and Sucheng District in Suqian City were represented which the highest average daily incidence rate in 2015-2020 was 3.7 per 100,000, 1.8 per 100,000, 1.5 per 100,000, respectively. Except for the gradual decline of the Kaifa Disticts since 2017, other counties showed the average daily incidence rate was one year down, one year up which descending signi cantly in 2017, and increasing signi cantly in 2018, and incidence rate in 2018 is almost higher than that in 2016. Medium epidemic counties of HFMD was the second kind. The average incidence rate of HFMD in the middle epidemic counties is basically in the range from 0.5 per 100,000 to 1 per 100,000 in 2015-2019. Huishan District in Wuxi City, Suyu District in Suzhou City and Tinghu District in Zhenjiang City were represented which the average daily incidence rate was 0.57 per 100,000 to 1 per 100,000, 0.54 per 100,000 to 1 per 100,000. 0.50 per 100,000 to 0.89 per 100,000, respectively. Except for Yancheng (Tinghu and Yandu District), the incidence rate increased slowly in 2017 and declined in 2017-2020 years, other counties showed the average daily incidence rate was one year down, one year up and in 2016 and 2018 was the most prominent, about 2 times that of the year before and after. Low incidence counties of HFMD was the third kind. The average incidence rate in 2015-2019 of HFMD in the low epidemic counties was basically in range of 0.01 per 100,000 to 0.5 per 100,000. Among them, the three lowest incidence rate counties were the counties of Binhai value showed that the mean of coe cient of correlation R 2 was 0.50 ± 0.15, so the model was tted well (Table S1). The tting result of Southern Jiangsu ( Figure 2) was that except for Tianning (R 2 =0.240,

Comparison of the transmissibility of 97 counties
We compared the transmissibility of 97 counties by the Rank -sum ratio (RSR). According to the RSR distribution table (Table 2), we constructed RSR and Probit regression equation which could be obtained as: (F=1813.37, P=0.000), through this equation, the of each district could be calculated and using it to classify the transmissibility into 6 levels which showed in Table 3. From 1 to 6, the transmissibility was getting weaker and weaker. The result showed that counties with the strongest transmissibility were Rugao in central Jiangsu and Jianhu in northern Jiangsu, while the weakest were Liyang and Jintan in southern Jiangsu, and Sihong in northern Jiangsu. Most counties were in 3-4 level indicating that those counties' transmissibility was relatively similar, especially in the same region or city.

Comparison of R eff in different years in Jiangsu Province from 2015 to 2020
We compared the R eff in different years by Kruskal-Wallis H test, the result showed that the median of R eff for each year from 2015 to 2019 was different (χ 2 =21.283, P=0.000), and the median of R eff in 2017 was smaller than that in other years (P<0.05).

Discussion
In this study, the seasonally adjusted SRIAR model was used to study the transmissibility of HFMD among 97 counties in Jiangsu Province, to provide suggestions for local CDC, community in Jiangsu Province and other areas with similar transmissibility of HFMD.

Analysis of the different incidence rate in various counties
The incidence rate of Southern Jiangsu was higher than Northern Jiangsu and with a peak of two seasons in a year which was consistent with earlier studies of HFMD in Jiangsu Province 28 , but in this study, we found that some counties of Northern Jiangsu had one seasonal peak. We considered the reasons for the different incidence rate and seasonal in various counties may be as following: 1) The climate zone of the regions is inconsistent. Liu W et al. found that the incidence rate of HFMD in Jiangsu was proportional to the average temperature and rainfall, but negatively correlated with the days of rainfall(≥0.1mm), low temperature, high temperature and sunshine duration 28 . A systematic review showed that the incidence of HFMD increased signi cantly when the temperature and relative humidity increased by 1℃ and 1%, respectively 28 , moderately warm environment promotes the spread of the HFMD virus. We thought that Southern Jiangsu is warmer than the Northern Jiangsu what may cause the incidence rate is higher, and the winter in the northern region was too cold to prevent the spread of hand foot mouth disease.
2) The demographic characteristics of regions are different. The south of Jiangsu Province is a densely populated area. Studies have found that most of the cases in this area are infants and children under 5 years old, so the incidence rate was different.
3) The epidemic virus serotypes are different. Zhuang ZC et al. thought CA-V16 may lead to the peak of HFMD in autumn or winter and the high incidence of adults 29 , CA6 often causes herpangina (HA), which is characterized by salivary blister pain. However, in many countries, HFMD does not contain HA data in hand-to-mouth disease reported to NNDSS, which often leads to a loss of reporting and a reduction in its incidence rate 5  Analysis of the different incidence rate in various years at the same regions We analyzed the different incidence rates in various years as following aspects: 1) From the perspective of climate change. Although the incidence and spread of HFMD are related to climate factors 28 , according to some meteorological researches, the temperature and rainfall in 2017 and 2018 are not abnormal compared with other years 31,32 . Therefore, the average incidence rate of 2017 and 2018 decreased signi cantly, and the increase may not be related to climatic factors. 2) Protective effect of the vaccine. Since 2016, the HFMD vaccine for EV71 has been put into use 33 and Changchun 40 in China increased to more than 60% in 2013. In this study, we also found that there are two peaks a year in most years, while many prevalent peaks occurred in 2018 in incidence rate tting results. Therefore, the repeated outbreaks after 2017 may be caused by CA16 infection or new virus subtypes after vaccination. 4) The impact of other infectious diseases. In this study, we found that the average daily incidence rate of HFMD was lower than that of the previous 10 times in the rst half of 2020. This indicated that the protective measures against COVID-19, such as school closures, business discontinued, frequent hand washing and wearing masks, and maintaining social distance, have affected the prevalence of HFMD to some extent. Other research also showed that the incidence rate of HFMD was affected by road passenger volume and population mobility during the term and Spring Festival. The combined effect was more signi cant than that of meteorological factors on the epidemic of hand foot mouth disease 19 .

Analysis of the difference of transmissibility of three regions in Jiangsu Province
The average R eff of HFMD in Jiangsu Province from 2015 to 2020 was 1.54, which was similar to the research results of foreign and most domestic provinces and regions, but the R eff was lower than that of Shenzhen, Guangdong Province 25 . We found that the R eff in Southern Jiangsu was less than that in Northern Jiangsu which we contrary to the incidence rate of the regions. We considered the reason may be as the following: 1) Because of the area and population of Southern Jiangsu are bigger than that of in Northern and Central Jiangsu, causing the number of susceptible persons was larger in Southern Jiangsu, so that the transmissibility of Southern Jiangsu was lower, while the incidence rate was higher.
2) The large population base of Southern Jiangsu will also increase the risk of HFMD indirectly caused by other infectious diseases. Studies have shown that the incidence rate of onychomycosis is related to HFMD with Cox A16 serotype 41,42 , especially adult population. Based on the above analysis, we suggested that South Jiangsu should pay more attention to a wide range of publicity in the season of HFMD onset, and for the central and Northern Jiangsu areas with strong transmissibility of HFMD, strengthening the implementation of protective measures is more helpful to reduce its prevalence.

Analysis of the comprehensive comparison results of transmissibility in 97 counties
Jianhu District in Northern Jiangsu and Rugao District in Central Jiangsu had the strongest comprehensive evaluation of transmissibility, but the trend of transmissibility was different. Jianhu District maintained high transmissibility from 2015 to 2016, and has a downward trend from 2017 to 2019. From the previous research we could nd that in Yancheng City, where Jianhu is located, HFMD is highly prevalent among infants., while the higher the level of maternal antibody to EV71, the stronger the protection for infants 43 . Therefore, the implementation of vaccine immunization in Jianhu District has a certain protective effect, but it is also necessary to further strengthen the propaganda and education and detect whether there is a new virus subtype epidemic. The transmissibility in Rugao District had uptrend from 2017-2020, Research showed that the HFMD in Rugao District had been more serious in recent years, and the incidence rate and incidence ratio were the rst places in class C infectious diseases 44 where also had critically ill patients in 2015-2020. There are inappropriate nursing and poor health environment in the rural areas of this district, and the vaccination situation is also low. We need to focus on improving the health environment, strengthen the publicity and health education, improve the awareness of epidemic prevention, and improve the epidemic situation monitoring, especially the analysis and monitoring of virus subtypes of severe patients.

Analysis of the different transmissibility of various years
The trend of HFMD transmissibility over time showed that R eff was the lowest in 2017 which may be related to the implementation of EV71 vaccine in 2016. Because we found the incidence rate in 2017 was also lowest, so the immunity of the vaccine to EV71 and the publicity of vaccination reduced the number of susceptible people and infected people, which reduced the actual transmissibility of HFMD. But what's interesting is that in many counties, most of the transmissibility suddenly increased in 2018, and the peak height could be higher than that in 2015 and 2016. We think this may be due to the different subtypes of was also found in Suzhou 30 . In a study on HFMD in Changsha, China, EV71 interacts with Cox A16, and the interactions between EV71 and other enteroviruses and between Cox A16 and other enteroviruses are all directional 49 . Therefore, we suggested that based on the classi cation of different transmissibility described by results to select counties to monitor the subtypes of HFMD, and the HFMD vaccine for different subtypes should be developed to cope with the change of epidemic pathogens.

Limitations
Due to the limitation of data, this study has some limitations. In this model, factors that may affect the transmissibility, such as age and gender, are not included, which may have some impact on the results, and the actual data of possible in uencing factors such as climate characteristics, virus types, population data were not collected for correlation analysis with transmissibility of HMFD in various counties. 2. The prevalence and transmissibility of HFMD in Jiangsu have regional and seasonal characteristics.
The higher the incidence rate in the three regions, the lower the transmissibility. The peak period of the epidemic will be changed from one season to two seasons.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. supplementTableS1.docx