| Abstract | Basic elements of outbreak examinations and e pidemiology are person, place and time. The epidemiology research has focused on “person” and “time” over a hundred years. The element of “place” has not been much addressed. However, only the disease mapping has been carried out in some areas. The develop ment of Geographic Information Systems (GIS) in recent years has provided a more powerful and rapid ability to investigate spatial patterns of diseases and processes. This is referred to and more useful in epidemiologic investigations and also disease surv eillance including policy relevant issues such as health services and planning. Spatial epidemiology approaches (i.e. probability mapping, spatial interpolation, spatial autocorrelation analysis and space - time analysis), using GIS, were used to visualize and analyze the geographic distribution of diseases through time (patterns, trends and relationships) that would be more benefic ial to understand the spatial spread or diffusion of disease outbreaks (i.e. the occurrence of a large number of disease cases i n a restricted geographical area over a short period of time). Recen tly, the attempt to apply these approaches and GIS for investigating disease outbreaks has become the important tool in epidemiological studies in Thailand. Chiang Mai province, the stud y area, is the largest province in northern Thailand. In this province some of the most dominant epidemic are diarrhea, food poisoning, pneumonia and dengue fever. Diarrhea cases are surprisingly very high. A better understanding of the spatial spread or d iffusion of these disease outbreaks is central to the design of prevention and control strategies for public health officers who work in this area. The major objective of this study is to apply the spatial epidemiology approaches for studying diseases patt erns, hotspots and diffusion in Chiang Mai province. The specific objectives are to i) adjust the incidence rate of diseases by applying the empirical Beyes method and illustrate values by diseases mapping, ii) analyze the disease patterns in the terms of seasonal, population and geographic distribution patterns, iii) detect the disease hotspots under different years by using the local spatial autocorrelation analysis and iv) utilize the space - time permutation model and GIS for investigating and visualizing the spatial diffusion of disease outbreaks. The study covers 2,070 villages of Chiang Mai for the period 2001 - 2006. Data about patients with diseases and population at village level were obtained from the Chiang Mai. Provincial Public Health Office (CMPHO ), Thailand. These records included disease cases (diarrhea, food poisoning, malaria, dengue fever (DF), dengue hemorrhagic fever (DHF), influenza, pneumonia and fever of unknown origin) referred from other hospitals and the population figures from the Min istry of the Interior, Thailand. The spatial data in this study included the village location points in the year 2006, which were collected from the Geo - Informatics and Space Technology Centre (Northern Region) (GISTC), Thailand. All these data were incorp orated in GIS . The methodology included the spatial epidemiology approaches such as the empirical Bayes smoothing, kernel density interpolation, global and local spatial autocorrelation analyses, space - time permutation scan statistic and inverse distance weighed (IDW) were selected to support the objectives of this study. All incidence rate (IR) of diseases for each village were adjusted by using the empirical Bayes smoothing function in the GeoDa software and converted to the morbidity rate (MBR) by multi plying by 1,000. These values were represented with the choropleth maps of diseases by using the kernel density interpolation. The Moran’s I indices of global spatial autocorrelation analysis were applied to detect spatial patterns (clustered/random/disper sed) of diseases at the global level. The local indicator of spatial association (LISA) was used to measure and test spatial patterns at the local level and could be used to determine locations of clusters or hotspots of diseases. And also, space - time perm utation scan statistic and the spatial prediction method of inverse distance weighed (IDW) were used to investigate and visualize the spatial spread or diffusion of disease outbreaks in space and time. For disease mappings, the raw estimates of IR were sp atially smoothed using empirical Bayes smoothing technique. The spatial prediction method of kernel was then used to produce the choropleth maps of MBR, which highlight the risk of eight diseases at certain places in Chiang Mai to dye in a particular year. Disease patterns analysis revealed that the epidemic patterns of infected diseases have fluctuated every year under different seasons, from 2001 to 2006. The incidences of food and water bone diseases (diarrhea and food poisoning) and the respiratory dis eases (pneumonia and influenza) were increased in hot (March - May) and rainy (June - September) seasons. The vector bone diseases (dengue fever, DHF and malaria) and the fever of unknown origin were occurred with high infected case during rainy season. While v most of all diseases were distributed notable among children less than 5 years of age, the vector bone diseases were distributed among age group of 10 to 25 years. The global spatial autocorrelation analysis for annualized MBR of villages in Chiang Mai fro m 2001 to 2006 showed that the Moran's I (0.02 - 0.49) values were significant (0.01 significance level) for each year, implying that distribution of the affected villages with diseases was somewhat spatially autocorrelated (low clustered) though the overall tendencies were not so strong. For the hotspots detection of diseases, the LISA was used to examine the local level of spatial autocorrelation in order to identify villages where values of the MBR were both extreme and geographically homogeneous. The hot spot trends found over 2001 to 2006 periods were indicated increasing trend in four diseases. These were the fever of unknown origin (R 2 = 0.83), dengue fever (R 2 = 0.59), diarrhea (R 2 = 0.53) and food poisoning (R 2 = 0.47). Note that the hotspot villages of f our diseases, which were represented by the highly concentrated locations of risk, have indicated strong increasing in recent years. In effect, the disease hotspots were obtained and found to be significantly clustered in the high MBR zones of diseases. F or example, the most clusters of malaria hotspots were in the north of Chiang Mai which occurred at Wang Haeng, Chiang Dao, Fang, and Mae Ai districts. Especially, the villages, which were located nearby the Thai - Myanmar border, have produced higher incide nce or risk values. This confirms that the malaria disease would be distributed around the border of Thailand and Myanmar. And also, the most of the clusters of dengue fever and DHF hotspots occurred in the downtown or urban area of Chiang Mai. This confir ms that the dengue fever/DHF incidents were concentrated in the province’s Muang district of Chiang Mai and distributed in the urban areas of the tropical countries. The spatial diffusion of disease outbreaks from diarrhea is selected as the case study. A space - time retrospective analysis was conducted to detect the outbreak signals of diarrhea from 1 November 2003 to 31 October 2006, in order to include all confirmed diarrhea cases of patient less than five years of age (37,536 cases). In the study period , the strongest signals of outbreak (13 villages) was on 17 January 2006 (cool season) and occurred to the conurbations of Mae Chaem district (Chang Khoeng and Tha Pha sub - districts), which is located about 100 km to the south - west of Chiang Mai City. This signal had 44 cases observed over 8 days when 2.98 were expected (relative risk= 14.79). The weaker signal was on 29 June 2005 (begin of rainy season) that were located in southern Chiang Mai (Chom Thong district), there had 16 cases observed over 11 days when 1.21 were expected theoretically (relative risk= 13.24). The daily spatial spread or diffusion of diarrhea outbreak from the first day to the end day of incidence (17 - 24 January 2006) in the study site of Mae Chaem district was follow ing a pattern of contagious diffusion. They spread outward from a village of origin (Ban Rai) with high incidence to nearby villages, which related to the area of high population densities, within the urban and agricultural areas along the main course of Mae Chaem River. This study exhibits that proposed methods and tools can be useful for disease surveillance for public health officials. The outcome from the study demonstrated that integrating existing health data, spatial epidemiology concept and GIS can provide an vital information on disease clusters within infected areas, and also form a basis to pursue further investigation for related factors responsible for disease risk. To implement specific and geographically appropriate risk-reduction programs for public health officers, the use of such spatial analysis and tools may be adopted as an integral component in the epidemiologic description and risk assessment. The methodology is developed for the diarrhea but it is general and may be applied for any other epidemic. |