Spatial-Temporal Epidemiology Of HFMD On A Smaller Scale

Hand, foot and mouth disease (HFMD) is a common infectious disease caused by various intestinal viruses, and it mainly occurs in the children aged less than5years. CoxA16and EV71are the most common strain of HFMD. HFMD has extremely strong infectivity and is usually prevalent from April to August. The transmission routine is complexed. It can cause large scale epidemic or outbreak in a short time. HFMD was very prevalent in Shandong Province. Zhaoyuan in2000, Xintai in2003and Heze in2009had HFMD outbreak and prevalence since HFMD outbreak was reported firstly in Zaozhuang in1984. In recently years, especially in2009, HFMD was very severe in Liaocheng, which had high focus by the society.The study showed that HFMD had spatial heterogeneity and spatial structure except seasonality. However, HFMD spatial-temporal distribution revealed different characteristics in different spatial scale. It is very important and significant to detect the characteristics of HFMD spatial-temporal epidemiology in different scale to tract HFMD spatial-temporal outbreak point and make regional prevention measures. The study about HFMD spatial analysis mainly was in county scale as the geographical unit. However, making specific interventions for prevention and control of HFMD mainly was replied on city or county-level centers for disease control and prevention in actual condition. Therefore, there was not enough to analyze spatial-temporal epidemic dynamics in county scale. In order to help city or county-level centers for disease control and prevention to provide HFMD spatial-temporal epidemic dynamics, locate high incidence areas and make preventive measures, it is necessary to illustrate HFMD spatial-temporal dynamic epidemic features and risk factors on smaller scale. The study selected Liaocheng of high HFMD incidence in Shandong Province as study site, and used spatial autocorrelation analysis, space-time scan statistic and geographical weighted regression (GWR) to systematically analyze spatial-temporal epidemiological characteristics and ecological risk factors of HFMD cases aged less than5years old from2007to2011on town scale.Results:1. HFMD had some epidemiological characteristics in Liaocheng. Firstly, HFMD had obviously seasonality with unimodal distribution, generally increased in March and decreased in May. The number of cases in September was usually as the same as the number of cases in May. Secondly, the sex ratio was shown in the Table1each year from2007to2011. Although the sex ratio was different in different year, the ratio of males than females remained unchanged. Thirdly, although different occupation between HFMD constituent was statistical significance, the scattered children were the dominant. Fourthly, the proportion of cases aged from1to2was the largest each year, which indicated the younger children were more vulnerable to HFMD. Fifthly, EV71and CoxA16were the major epidemic strain to cause HFMD in Liaocheng.2. The high HFMD average incidence mainly located in the county boundary such as Guan, the center of Dongchangfu, the south of Yanggu and the north of Gaotang. In2007, the high incidence mainly located in the east of Dongchangfu, such as Donge and Chiping. In2008, the high incidence mainly located in Gaotang, Guan and Dongchangfu. In2009, the high incidence mainly located in Guan, Yanggu and Dongchangfu. In2010, the high incidence mainly located in Guan, Dongchangfu, Yanggu and Guhe Town in Gaotang. In2011, the high incidence mainly located in Dongchangfu, Guan, Gaotang and the part of Guan. It indicated that HFMD was presented in spatial-temporal heterogeneity. The high incidence areas were wandering in different year. 3. The high incidence areas mainly located in Guan, Gaotang and Dongchangfu after adjusting HFMD incidence using spatial empirical Bayes smoothing from2007to2011. The high incidence areas mainly located in Guang, Dongchangfu and Gaotang, which indicated that demographic composition had an great effect on spatial distribution of HFMD. High HFMD incidence areas were stable after eliminating the effect.4. The high excess risk ratio areas mainly located in Guan, the center of Dongchengfu and the south of Yanggu from2007to2011. In2007, the high excess risk ratio areas mainly located in the east of Liaocheng. In2008, the high excess risk ratio areas were in discrete state. In2009, the high excess risk ratio areas mainly located in Guan, Yanggu and Dongchangfu. In2010, the high excess risk ratio areas mainly located in Guan, Dongchangfu, Yanggu and Gaotang. In2011, the high risk ratio areas mainly located in the east of Guan, the center of Dongchangfu and Gaotang.5. Spatial distribution of HFMD was not random, with Moran’s I=0.16and P<0.01from2007to2011. It indicated that HFMD was positive spatial autocorrelation and spatial cluster. Moran’s I index about HFMD incidence was statistically significance each year. Local spatial autocorrelation analysis indicated that spatial cluster areas mainly located in Guan, Dongchangfu and Yanggu. Spatial cluster areas mainly located in the county boundary.6. The study obtained one most likely cluster area in the whole area from2007to2011using temporal scan analysis. Time frame was from2009/3/22to2011/7/18with RR as10.65and P as0.001. Although the study area had one most likely cluster with statistical significance each year, time frame was different and relative risk (RR) decreased each year.7. The study obtained9high incidence areas using spatial scan analysis from2007to2011in Liaocheng, including one most likely cluster and8secondary likely clusters. One most likely cluster had7towns with RR as3.38, mainly located in Gulou, Xinqu, Liuyuan, Dongcheng, Huxi, Beicheng and Jiangguantun in Dongchangfu. In2007, the study obtained2high incidence areas with statistical significance in Liaocheng. One most likely cluster had17towns with RR as6.49and P as less than0.01, located in Chiping and Donge. In2008, the study obtained9high incidence areas with statistical significance, including1most likely cluster and8secondary likely clusters. One most likely cluster had3towns with RR as12.83and P as less than0.01, located in Gulou, Liuyuan and Xinqu in Dongchangfu. In2009, the study obtained7high incidence areas with statistical significance, including1most likely cluster and6secondary likely clusters. One most likely cluster had20towns with RR as3.33and P as less than0.01, located in Dongchangfu, Chiping, Guan, Yanggu and Shen. In2010, the study obtained4high incidence areas with statistical significance, including1most likely cluster and3secondary likely clusters. One most likely cluster had7towns with RR as4.10and P as less than0.01, located in Dongchangfu. In2011, the study obtained3high incidence areas with statistical significance, including1most likely cluster and2secondary likely clusters. One most likely cluster had31towns with RR as4.08and P as less than0.01, located in Dongchangfu, Yanggu, Chiping and Guan.8. The study obtained5spatial-temporal clusters with statistical significance using space-time scan analysis in Liaocheng from2007to2011. One most likely cluster had29towns with RR as1.65and P as less than0.01, located in Dongchangfu, Guan and Shen. Its time frame was from2009/4/17to2009/9/29.9. CPD and NMW were selected into the regression model using stepwise ordinary linear regression (OLR) analysis. The adjusted R-square was0.592with F statistic as78.306and P as less than0.001. The adjusted R-square of GWR was0.613, greater than OLS. CPD and HFMD incidence had positive relationship in most towns of Liaocheng. In the regional of regression coefficient with statistical significance, the more CPD was, the higher HFMD incidence was. NMW and HFMD incidence had positive relationship in most towns of Liaocheng. In the regional of regression coefficient with statistical significance, the more NMW was, the lower HFMD incidence was. CPD and NMW effect on HFMD incidence was obvious variability of geographic gradientsConclusions:1. HFMD had obvious seasonality with unimodal distribution in Liaocheng. The number of males was more than females. Scattered children were the dominant. EV71and CoxA16were the major strains to cause HFMD.2. Spatial distribution of HFMD was spatial autocorrelation, not random in Liaocheng.3. HFMD had obviously spatial-temporal features. In the purely temporal dimension, HFMD not only had long trends with single period, but also had its own special unimodal disease with seasonality. In the purely spatial dimension, although it had high incidence areas in the whole region from2007to2011, the high incidence areas were not always the same each year, and in the wandering condition. In the spatial-temporal dimension, it had obviously spatial-temporal cluster. The high incidence areas were always located in the special space and time condition, which indicated that the high incidence and cluster areas had the risk factors to promote HFMD prevalence in the specific time.4. The risk factors of HFMD were child population density (CPD) and health resources (NMW) in Liaocheng. The greater CPD was, the higher HFMD incidence was. The greater NMW was, the lower HFMD incidence was. Spatial dependency was presented in a specific point with space-centered around obvious geographic gradients variability.5. Based on HFMD specific spatial-temporal epidemiological characteristics, its prevention and control not only pays attention to the spatial-temporal cluster areas to make regional measures, but also observe the wandering spatial trends to make emergency prevention and control measures immediately. Innovations:1. The diseases’ spatial-temporal epidemiological characteristics depend on specific spatial-temporal scale. The study systematically detected HFMD spatial-temporal epidemiology and ecological risk factors in smaller spatial and temporal scale. It is the first time to clarify HFMD spatial-temporal epidemiological characteristics on smaller scale.2. The study clarified HFMD had not only fixed high incidence areas but also wandering spatial trends to provide new epidemiological basis for its prevention and control.3. The study clarified CPD and NMW were risk factors of HFMD with geographic gradient variability in Liaocheng to provide new epidemiological basis for regional prevention and control.4. The study used spatial autocorrelation analysis, spatial-temporal scan analysis and GWR of spatial statistical methods to systematically clarify HFMD spatial-temporal epidemiological characteristics and provide spatial statistical methods on spatial-temporal epidemiology of infectious diseases.Deficiencies:1. Because of small study area, climate, vegetation, ecological and other important ecological factors failed to be included with small spatial variability in the study. The study will need to expand the study area in the future.2. Although we got the number of village cases, the study did not analyze spatial characteristics in village scale because of limited time. The study will replenish this part to clarify spatial-temporal epidemiology on a micro scale in the future.