1. Introduction
Dengue is the most common mosquito-borne viral disease of humans. In recent years, dengue has become a major international public health concern [
1]. The global incidence of dengue fever has grown dramatically in the past 50 years. The disease is transmitted to humans mainly by the mosquitos
Aedes aegypti and
Aedes albopictus. Clinical symptoms varying from dengue fever (DF) or classic dengue to dengue hemorrhagic fever (DHF), which may progress into a severe form known as dengue shock syndrome (DSS) [
2]. Normally, dengue virus circulating in the blood of viraemic humans gets ingested by female mosquitoes during feeding. The virus then infects the mosquito mid-gut and subsequently spreads systemically over a period of 8–12 days. After this extrinsic incubation period, the virus can be transmitted to other humans during subsequent probing or feeding. Dengue can be a severe, flu-like illness that affects infants, young children and adults, but seldom causes death. The clinical features of dengue fever vary according to the age of the patient. Infants and young children may have a non-specific febrile illness with a rash. Older children and adults may have either a mild febrile syndrome or the classical incapacitating disease with abrupt onset and high fever, severe headache, pain behind the eyes, muscle and joint pains, and rash [
3]. Till now, the disease has been reported in over 100 countries located in Africa, the Americas, the Eastern Mediterranean, Southeast Asia and the Western Pacific, which threaten the life of more than 2.5 billion people in urban, periurban, and rural areas of the tropics and subtropics. Dengue fever incidence is high in countries with tropical and warm climates [
1]. Causes of increasing dengue transmission may include rapid expansion of urbanization, inadequate water supplies, increased movement of mosquito and human populations within and between countries, and spread of insecticide resistance in mosquito vector populations [
4]. In 2010, dengue incidence in several Asian countries constituted a leading cause of pediatric hospitalization [
5].
In Thailand, an upward trend in the incidence of dengue has been observed, with acute and severe forms of dengue virus infection, since the first dengue outbreak in 1958 [
6]. In 2008, according to dengue surveillance data from the Thai Ministry of Public Health (MOPH), the total numbers of reported cases of dengue infections in Thailand were 43,911, with 46 deaths nationwide, including 18,797 DF cases, 24,455 DHF cases, and 659 DSS cases.
Epidemiological studies of dengue are not easy, for many reasons. The first reason is the strong silent transmission (about 80% of cases display no symptoms). Only severe cases were reported; patients with low or mild symptoms were not considered by the health system. In Thailand as seen in other Southeast Asian countries, the disease can also be confused with others, like influenza, if the diagnostic is only based on symptoms, as the two diseases pose almost the same temporal pattern. The second difficulty correlates with the first one, which concerns with unknown immune status of the population towards dengue infections. Dengue virus have four known serotypes, and each serotype induces a lifelong immunity in recovery cases. Immunity in the population is therefore very important to understand the epidemiology of the disease. The third difficulty lies in the fact that dengue is a vector borne disease, and many factors of outbreaks are related to the vector behavior and its relationship with the environment, like climate, breeding site density probability and vector control, urbanization, human population movement, etc. The presence and density of the vector (mainly
Aedes aegypti in the urban and peri-urban environment) is difficult to estimate. Climatic factors such as rainfall, temperature, humidity all influence dengue transmission. The high level of humidity during the rainy season makes the survival of the mosquito to be longer [
7,
8]. By implication Thailand’s rainy season from May to September provides optimal temperatures for
Aedes aegypti mosquitoes to thrive [
9]. Consequently, these conditions facilitate dengue epidemic outbreaks. Moreover, the Thai Meteorological Department (TMD) has reported higher dengue outbreaks in El Niño years. El Niño events in Thailand are actually related to high temperature and low rainfall [
10]. Hence a major objective of this study was to identify dengue diffusion patterns with respect to space and time.
GIS can be used to assess and identify potential risk factors involved in disease transmission such as socio-economic, climatic, demographic, and physical-environment variables. GIS technologies have been applied in epidemiological public health studies for many years [
11–
13]. GIS and spatial analysis are powerful tools in addressing epidemiological problems, allowing the identification of critical areas and variables intimately related to the modulation of the disease dynamics [
14,
15].
Spatial analyses and statistics, such as spatial autocorrelation analysis, cluster analysis, temporal analysis, are commonly used to highlight spatial patterns of diseases and to test whether there is a pattern of disease incidence in a particular area [
16–
19]. Recent advances in spatial statistics in GIS have led to a growing interest in the detection of disease clusters or “hotspots” for public health surveillance, in particular for improving the understanding of the growing incidence of dengue fever [
20]. Spatio-temporal patterns can provide clues in understanding the dynamics of disease spread. Detection of spatial, temporal and space-time clustering is useful in identifying higher risk areas and times, where disease surveillance and control need to be targeted [
21]. For instance, Rotela
et al. [
13] investigated the spreading dynamic of dengue fever outbreak in Tartagal city by Knox’s test method. Cummings
et al. [
22] examined the spatial-temporal dynamics of dengue occurrence in Thailand by applying empirical mode decomposition method to show the existence of a spatial-temporal traveling wave. Maidana and Yang [
23] measured the speed of dengue dissemination following the invasion and colonization by only the movement of adult mosquitoes. Tran and Raffy [
24] developed model for spatial and temporal dynamics of dengue. Hence, in this study spatial statistical analyses were used to investigate spatio-temporal diffusion patterns of dengue cases.
4. Discussion
Spatial epidemiological research has a long history, but epidemiology studies using GIS has emerged only recently. With the development of computer technology and spatial analysis methods, GIS is becoming more and more important [
33]. Monitoring and planning control measures for dengue epidemics have recently become vital to control disease outbreaks. This article aimed at providing useful information on dengue incidences and mapping their patterns and dynamics of diffusion. Spatial autocorrelation analysis proved to be a valuable tool to analyze the spatial patterns change over time.
The study revealed useful information on age group and gender vulnerability to dengue. Incidence of dengue observed to be greater than expected in the 0–24 years old age group and lower in population with less mobility like older than 25 years old. Additionally, several studies confirmed that dengue risk exposure is greater at home because of the endophilic habits of
Aedes aegypti [
13,
27,
44]. However, clinical symptoms may also be reported to a lesser degree by young people because of better self-recovery ability [
13].
Climate also plays important role and it was seen that dengue is generally prevalent in the province of Chachoengsao during the months of May to September. Temporal analysis of climatic factors (rainfall, temperature, and humidity) showed that dengue generally occurs when average temperatures increase, when the humidity is higher than average, and when the rainfall season has already started. As shown, rainfall and relative humidity data of one month before (t−1) showed very high correlation with dengue incidence. Globally, the vector-borne disease and associated vector activity are positively associated with temperature (<40 °C) [
45]. There are number of studies in the literature dealing with relationships between temperature and dengue occurrences and dengue vector abundance [
10,
12,
46]. Nakhapakorn and Tripathi [
10] reported that the dengue occurrences in Thailand were positively associated with rainfall and negatively associated with temperature and humidity, whereas during the rainy season they were positively associated with rainfall and temperature and negatively with humidity. Similarly, in Taiwan, Wu
et al. [
12] found a positive association between the number of dengue occurrences and the monthly maximum or minimum temperature and the cumulative rainfall with a lag of two months. These observations are coherent with the biology of vectors of viruses. It was shown in many regions that the minimum temperature is the most critical factor for the threshold of mosquito survival and developing rate in sustaining the population density. Likewise, Sriprom
et al. [
46] found dengue virus infection incidence to be positively associated with the monthly minimum temperature, consistent with the literature, and for the extrinsic period as the virus would not amplify in the vector when the temperature becomes less than 18 °C.
Using spatial analysis methods in GIS, the spatial patterns of dengue cases in Chachoengsao province from 1999–2007 were mapped and analyzed. The nature of spatial distribution was found to be clustered in high density population centers. Concerning the empirical Bayes smoothing (EBS) method, raw rates were used to estimate this underlying risk, which reduced differences in population size and in turn addressed variance instability and spurious outliers. In short, rate smoothing presented one way to address this variance instability [
47]. The study showed that spatial distribution patterns of dengue cases were significantly clustered, and identified the dengue hotspots in Chachoengsao province. Kernel density estimation illustrated variation in the grouping of dengue areas across the study area, and strongly confirmed the visible pattern on the point location map. Consequently, the village locations were chosen as the best way to analyze the spatio-temporal patterns of the outbreak dynamics over 153 days (May–September) in the year 2007 to study the temporal dynamics in space and time. During the epidemic, there were as many as 551 suspected dengue cases spread throughout the province, affecting 0.08% of the total population. Approximately 24.76% of the cases occurred in June (weeks 2–4). The outbreak dynamics showed a clear non-random pattern of spreading from the first village to other villages each day. The tracking analysis of the disease shows a cluster pattern in the south-west (Bang Pakong district) and in the center (Phanom Sarakham district) of the study area, and also showed how the dengue occurrence locations of disease changed in space and time by movement of days 1 to 153 (
Figure 8). Hotspot movement by week did not show clear spread pattern or trend. If related to the temporal distribution of the cases, it showed that concentration of hotspot occurs and then disappears, even when the incidence is high. This result suggested that the disease is spreading locally around foci (radially), with waves of concentration-diffusion process of hotspot. However, the limitation in the study was the dengue cases data. Due to administrative reorganization, some new villages were formed and dengue cases data for these villages was not available for earlier years.