Space-time scan analysis of HFMD data was conducted year by year in mainland China from 2011 to 2018. The results showed that the cases of HFMD were not randomly distributed in mainland China during the study period. Each year at least five significant spatiotemporal clusters were found, including the most likely cluster and secondary clusters. The red part on the map is the area of the most likely cluster (Fig. 5).
Table 3 shows the cluster time, cluster province, relative risk, and other information of the most likely cluster from 2011 to 2018. In most years (2011–2017), the most likely cluster was in the southern coastal provinces, including Hainan, Guangxi, and Guangdong. The range of cluster time for these years was between April and October. The relative risk ranged from 3.75 to 6.53, indicating that the incidence risk in the clusters was 3.75 to 6.53 times that of outside the gathering area in these years. The most likely cluster was different from the past in 2018, which appeared in the southeast coastal areas, including Shanghai, Jiangsu, and Zhejiang. The cluster time was from May to September, and the relative risk was 3.45. Summarizing the results of space-time scan analysis from 2011 to 2018, we observed that the risk areas for HFMD were mainly southern coastal provinces (Hainan, Guangxi, Guangdong), followed by eastern coastal provinces (Zhejiang, Shanghai, Jiangsu, Fujian), which with a developed economy, high population density and strong population mobility. The high-risk time frame was at the end of spring, throughout the summer, and early autumn, when the climate is warm and humid. Although the central and southwest regions were not included in the most likely cluster, their relative risk was at the forefront for many years, and these provinces need to be paid attention to.
Discussion
Since HFMD was listed as a notifiable infection in China, it has caused an enormous disease burden. It has a severe impact on the health of children under five years. The launch of the vaccine in 2016 might change the epidemiological character of HFMD in mainland China. Using GIS technology to analyze the spatiotemporal distribution patterns of HFMD in mainland China will help further to explore the high-risk areas and time of the disease. Combining it with interrupted time series analysis will provide a basis for developing more effective HFMD prevention and control policies.
This study found that the incidence of HFMD in mainland China showed an overall downward trend after 2014, and a substantial reduction in the incidence of HFMD was observed in the moments after the introduction of the vaccine, which may be related to HFMD vaccines launched against EV-A71 enterovirus in mainland China in 2016[26, 43]. However, the interrupted time series analysis results indicated that, after the introduction of the vaccine, the incidence of HFMD showed an upward trend, and the increase was more extensive than before in mainland China. The incidence of HFMD in Mainland China was 169.41 per 100,000 in 2018, which was only lower than the incidence in 2014 and 2016, higher than Thailand's incidence rate of 102.51/100,000[44] and Malaysia's incidence rate of 138.6/100,000[45]. So, the incidence of HFMD was still at a high level in mainland China in 2018. There may be three reasons for this. On the one hand, the monovalent inactivated EV-A71 vaccines cannot provide protection against non-EV-A71 virus infections of HFMD[46]. On the other hand, due to the stable birth rate and no cross-protection of alternate serotypes, the previous HFMD epidemics did not reduce the susceptible persons in the population probably[47]. In addition, the vaccines of HFMD have not been included in the National Immunization Program in China, which means that parents need to vaccinate their children at their own expense and voluntarily. Hence, the vaccination coverage rate was not high[48]. A study found that the annual EV-A71 vaccination coverage rate of children under 5 years old ranged from 5.53–15.01% in Yunnan from 2016 to 2019[49]. In Guangdong, the coverage rate of EV-A71 vaccination among the 6-month to 5-year children was 2.67% in 2016 and 10.07% in 2017[50]. A study in Chengdu showed that the EV-A71 vaccination coverage rate should reach 94.0% to end the EV-A71-related HFMD transmission[51]. Therefore, efforts should be made to develop multivalent vaccines against major enteroviruses and increase the vaccination rate in the future.
The study results suggested the cyclical and seasonal pattern of the HFMD epidemics. The HFMD epidemic took two years as a cycle. This cyclical pattern was observed in many countries, and there was evidence that this cyclicality is related to the time required to accumulate enough susceptible children in the population[7, 52, 53]. In this study, the seasonal pattern of HFMD was reflected in the peak incidence of twice a year, the first peak occurred in April to July, and the second peak occurred from September to November in mainland China. Seasonal patterns of HFMD have also been observed in other countries. HFMD usually breaks out from June to August in Thailand[54], cases of HFMD increased during the rainy season in Vietnam[55]. However, the seasonal patterns observed in temperate, tropical, and subtropical Asia are different[56]. In China, the seasonal characteristics of the south and the north were also inconsistent[57]. In mainland China, the first peak of each year may be related to climate factors such as temperature, humidity, and atmospheric pressure. Appropriate climatic conditions are conducive to the reproduction and survival of pathogens. Warm temperature promotes children’s activities and mutual contact, thereby promoting the transmission of pathogens. Although the temperature and humidity are still high in August, the school has summer vacation, which reduces the gathering and contact of students, so the incidence is reduced. September coincides with the end of summer and the beginning of autumn, so the temperature and humidity are still high. The second peak may be due to the opening of kindergarten in September to increase the contact between children, thereby increasing the possibility of cluster cases[24, 45, 57–60].
We also found that children under 5 years are the main morbid population, accounting for 94.17% of all reported cases. The result was consistent with other research’s findings in other countries and China[1, 7, 44, 48]. The low incidence of children over 5 years old maybe because they have had HFMD before, thereby protecting them and their peers from HFMD[61, 62]. In 2018, the incidence of children aged 5 and under was the lowest level in 10 years, which may be related to the increase in the cumulative coverage of vaccines[26]. Therefore, we call for attention to children in this age range, vaccinate them as soon as possible, and introduce the HFMD vaccine into the routine immunization program.
The global Moran’s I for 2011–2018 ranged from 0.419 to 0.547. The global spatial autocorrelation analysis results showed that the spatial distribution of HFMD cases was non-random in mainland China and tended to cluster at the provincial level, which was the same as the analysis result for 2008–2011 by Xiao[63]. Combined with the incidence distribution map (Fig. 3), we found that the incidence of HFMD was quite different between the southern and northern of China. Overall, the incidence of south China was much higher than that in north China. This phenomenon has also been found in other studies[1, 23, 24, 64]. Meteorological factors are a fundamental reason for the difference in incidence between south and north China. South China is mostly in high temperature, high humidity, and rainy weather conditions, which to a certain extent promotes the reproduction, survival, and spread of the pathogens of HFMD[23, 65]. Each year, there was only one peak incidence in late spring and early summer (May-July) in the northern region. At the same time, there were two peak incidences in April-June and September-November in the southern region[1, 25, 57].
The most likely cluster appeared in southern coastal provinces (Guangxi, Guangdong, Hainan) from 2011 to 2017. The results of local spatial autocorrelation analysis and spatiotemporal clusters analysis indicated that the southern coastal provinces (Guangxi, Guangdong, Hainan) were the areas with a high incidence of HFMD in China. Their incidence had been at the highest level in mainland China for many years, the same as previous research results[23, 66, 67]. The climate of these provinces is characterized by short and mild winter, while summer is long, hot, and very humid. In addition to climatic factors, the high incidence in southern coastal provinces may also be due to the high population density, strong population mobility, developed economy, and more frequent communication and close contact between individuals, which led to a higher risk of spreading HFMD[45, 59, 66–68]. The cluster time ranged between April and October, which was consistent with the peak time of HFMD in mainland China[1]. In 2017, the relative risk of the most likely cluster (Hainan, Guangxi, Guangdong) was 6.53, which was higher than before. Because the incidence of HFMD had decreased in other provinces, while the incidence of Hainan, Guangxi, and Guangdong were still at a high level, the low coverage rate of the vaccine may be one of the reasons. In 2017, the cumulative coverage rate of EV-A71 vaccine among 6-month to 5-year children was 9.26% in Guangxi[26] and 10.07% in Guangdong[50], while it was 18.94% in Yunnan[49] and 15% in Beijing[69]. In addition, some studies have found that non-EV-A71 enteroviruses were the dominant serotype in Guangxi in 2017, while the monovalent EV-A71 vaccine cannot generate cross-immunity against other enterovirus infections[26, 59, 70]. The interrupted time series analysis indicated that, after the introduction of the EV-A71 vaccine, the incidence of HFMD was lower than before in Guangdong, Guangxi, and Hainan. The incidence of HFMD showed a downward trend in Guangxi and Hainan. In 2018, the southern coastal provinces (Guangxi, Guangdong, Hainan) were no longer the most likely cluster. The decline in the incidence of HFMD was large in these three provinces, and the incidence had reduced to the lowest level during the study period, which was 258.10 per 100,000 in Hainan and 354.28 per 100,000 in Guangxi, 265.23 per 100,000 in Guangdong. This result may be related to the increase in vaccine coverage. Since Guangxi implemented comprehensive prevention and control measures focusing on promoting the EV-A71 vaccine among susceptible populations, the vaccine coverage rate of susceptible children had approached 30% in Guangxi in 2018, and the EV-A71-related incidence had declined[26, 70]. The same finding was also found in the study of Guangdong[59, 71]. For this reason, it is necessary to expand the HFMD vaccination program.
According to the results of spatiotemporal clusters analysis, we found that three southeast coastal provinces (Shanghai, Jiangsu, Zhejiang) were included in secondary cluster 1 from 2011 to 2016. The relative risk of this cluster was only second to the most likely cluster. This result indicated that these provinces were areas with a high incidence of HFMD, and they had a higher incidence than the northern and inland provinces. These provinces have a subtropical climate, so they are tended to be hotter and more humid. Furthermore, they have a higher population density and a larger migratory population. These factors are the main reason for the high incidence of HFMD[72]. In 2018, these provinces (Shanghai, Jiangsu, Zhejiang) constituted the new most likely cluster, and the cluster time ranged between May and September. We also discovered that the new High-High cluster area appeared in Jiangsu in 2018 through local spatial autocorrelation analysis. However, previous research by Chao Wang found that the most likely cluster was observed in the Shandong, Hebei, Henan, and Shanxi provinces in 2009, and were located in Guangdong, Guangxi, and Hainan provinces in 2010–2012[25]. Combined with our findings, it can be demonstrated that the spatiotemporal clustering pattern of HFMD had changed in 2018. The results suggested that the incidence of HFMD in Jiangsu and its neighboring Shanghai and Zhejiang was higher than that of other provinces in 2018, which was found in other studies[73, 74]. However, the incidence of the previous most likely cluster (Hainan, Guangxi, Guangdong) had decreased, so the southeast coastal provinces (Shanghai, Jiangsu, Zhejiang) became the new most likely cluster. According to the interruption time series analysis, after the introduction of the EV-A71 vaccine, the incidence of HFMD showed an upward trend in Zhejiang, Shanghai, and Jiangsu. In 2018, the incidence in Zhejiang and Jiangsu reached the highest value after the vaccine was launched, which was 436.02 per 100,000 in Zhejiang and 237.50 per 100,000 in Jiangsu, and Zhejiang had the highest incidence in the country. EV-A71 vaccine coverage rate had increased year by year to 24.05% in Zhejiang, higher than 19.4% in Guangdong[75]. The proportion of EV-A71 positive cases declined from 22.6% in 2017 to 3.3% in 2018 in Zhejiang[76]. Another study found that CV-A16 and CV-A6 became the primary pathogens of HFMD in Jiangsu, and the cases caused by these two pathogens accounted for more than 96% of all cases[77]. Therefore, the rise in the incidence of HFMD may be related to other serotypes becoming dominant after the introduction of the EV-A71 vaccine. Even large-scale vaccination of the EV-A71 vaccine would not greatly reduce the number of cases[26, 77]. This result suggested that the dominant pathogens in different regions and at different times are various. Therefore, it is necessary to monitor the enterovirus genotypes of HFMD cases in various places and promote HFMD multivalent vaccines.
Compared with studies in various provinces[66, 67, 78, 79], we found that clusters were not only distributed in one province but also cross-linked to multiple provinces, which showed the importance of cooperation between provinces to prevent and control the spread of HFMD[25]. This strong spatial and temporal correlation of HFMD was also found in other studies, which may be related to meteorological factors. Because meteorological factors promote the prevalence of HFMD, the meteorological conditions in adjacent areas and adjacent time points are similar[72].
The southern inland province Hunan was included in the High-High cluster areas in 2014 and 2016–2018. Although Hunan had not appeared in the most likely cluster, the results of spatiotemporal clusters analysis showed that Hunan was one of the high-risk areas for HFMD. It was observed to have a high incidence, which was also found in other studies[80]. Because Hunan has a subtropical monsoon climate with abundant heat, concentrated rainfall, and high humidity, in addition to high population density and high levels of migration, both suitable natural and social environments have promoted the development and spread of enteroviruses in Hunan[79, 81]. This result suggested that the HFMD epidemic in Hunan was worthy of attention.
For the prevention and control strategy of HFMD, we make the following suggestions: First, the prevention and control of HFMD should focus on southeastern and southern coastal provinces in mainland China, especially from April to September each year. Attention should also be paid to the southern inland provinces. Moreover, regarding the strong spatial and temporal association of HFMD in neighboring provinces, the provinces should strengthen cooperation and joint control to avoid a widespread HFMD across provinces. In addition, it is necessary to advance the EV-A71 vaccination plan and expand the vaccine coverage, especially in coastal areas of China. As the dominant pathogens have changed, future research should analyze the spatial-temporal distribution characteristics of different pathogens of HFMD and develop multivalent HFMD vaccines as soon as possible.
The main advantage of our research is using Chinese CDC's disease surveillance data, so the authenticity and completeness of the data are guaranteed. In addition, our study used spatiotemporal clusters analysis that combines time and space, revealing the temporal and spatial distribution patterns of HFMD at the province level in mainland China. We set 15% of the risk population as the maximum cluster size in the analysis, avoiding the most likely cluster containing low-risk regions and thus larger than the actual cluster. In this study, the data of HFMD cases were scanned year by year, and the trend of the most likely cluster could be found. As far as we know, it is the first study to analyze the spatiotemporal clusters of HFMD in the whole of mainland China after the HFMD vaccines were launched in China. Furthermore, we used interrupted time series analysis to assess the actual impact of the introduction of the EV-A71 vaccine on HFMD incidence.
The limitations of this study are worth mentioning. First, although Chinese CDC disease surveillance data could ensure the authenticity and completeness of the data to a certain extent, it may not avoid the differences in the quality of case reports in the surveillance systems of different regions and the underreporting of cases caused by mild cases that do not go to the hospital for treatment. In addition, the outbreak of COVID-19 in 2019 may change the spatial and temporal distribution pattern of HFMD in China. However, we have not obtained HFMD data after 2019. The incidence data of HFMD was at the provincial level. More accurate results may be attained if using smaller spatial scale (such as district, county) information. This study set 15% of the total population as the maximum cluster size based on previous research experience. Using MCS-P to select the optimal maximum cluster size can improve the performance of scan statistics[39]. But it does not apply to the data in this study. On the other hand, the spatial scanning window used to detect the cluster was circular, while the geographic shape was irregular, which may not represent the actual shapes of the clusters. What’s more, because we did not have access to pathogen data of HFMD cases, the spatiotemporal clusters analysis of HFMD pathogens was not possible. Substantial evidence indicated that other virus serotypes became dominant after the introduction of the EV-A71 vaccine. Future research should focus on the pathogen agents of HFMD.