Dengue risk map using GIS based on socio-environmental and climatic variables in residential area

Dengue, a major international health problem, is transmitted by aedes mosquitoes. Due to the lack of a vaccine, vector control by tackling the contribution factors may reduce dengue incidence. By understanding the factors contributed to the vector densities such as sociological, environmental and climatic factors, dengue prevention and control will succeed. This study is aimed at determining the major sociological, environmental and climatic factors contributing to dengue cases and to produce the risk map based on the contributing factors Questionnaire survey of 379 respondent with dengue history were utilized. The climatic data were obtained from Meteorological Department. Geographical Information System (GIS) technology has been used to integrate the socio-environmental and climatic factors with dengue cases

lakes, swamps, springs and rivers (13). Vegetation is also important for Aedes mosquitoes, especially Aedes albopictus; thus, landscaping can influence the abundance of mosquitoes. Aedes prefers thick and dense vegetation (14).
Vegetation also serves as a resting area during the day around residential houses (host availability) (11).
The uncontrolled expansion of urban environments with rapid population growth has accelerated the prevalence of dengue fever (15). Many studies attribute the increasing exposure to the dengue vector to overcrowded households (16,17).
Aedes mosquitoes feed on multiple humans per day and crowded conditions make it easy to transmit the virus efficiently (16). Several studies also suggested that dengue is associated with several sociological factors such as poor housing areas, household density, type of housing, multilevel housing, and human activities (13,(18)(19)(20).
Daily routines such as working and studying during the peak biting time are contributing factors to the incidence of dengue (21). Daily population movement is an inevitable activity in society. Thus, the risk of human exposure to the disease could vary physically (22). Previous epidemiological studies had incorporated population mobility and the location of daily activities for assessing the exposure and risk of disease transmission (22)(23)(24). Several studies assessed the crowdgathering places (market, school, parks, and bus/train station), and commuting behavior played an important role for dengue transmission (25)(26).
Because there is no available vaccine or specific drugs for dengue fever, control measures are based on the biological and chemical control towards aedes mosquitoes. In order to improve control, better understanding of factors that lead to the abundance and distribution of aedes mosquitoes is crucially needed (13).The aim of this study is to determine the sociological, environmental and climatic factors contributing to dengue cases at high risk areas in Kuala Lumpur, Malaysia and to produce the risk map based on the contributing factors.

Description of study sites
Kuala Lumpur is the capital of Malaysia with 243 km2 area and an average elevation  This study designed and estimated the sample of the respondents from the high risk areas considering the sociological and environmental determinants that contribute to dengue incidence within 5% of true prevalence with 95% confidence (31). The calculation formula of sample size was applied using OpenEpi Software (16) and also by Lemeshow et al., 1990. Based on the calculation, the statistically required sample is 381 respondents derived from dengue case data from January 2016 to May 2016. The respondents for this study were interviewed via phone call to obtain the answers to the questionnaire. The date interval was selected to avoid recall bias if a longer period was chosen (32)(33). Of the 730 respondents who were identified as dengue positive by the Kuala Lumpur City Health Department, 379 agreed to participate in this study and gave us full cooperation ( Figure 2). All the information gathered through the phone interview and questionnaire was documented The address of 379 respondent then were geocoded in order to produce the spatial point layer by using ArcGIS version 10 (34).

The predictive variables
The structured questionnaire used in this study consists of questions on sociological and environmental factors. The sociological questions were based on the following rationale: (i) information related to either work or study and the routine of daily activity during peak biting (early morning and late evening) (ii) history of dengue infection in the respondent (32), and in household members (16,(35)(36)(37), (iii) and number of people in the household. The environmental section consists of questions about environmental factors surrounding the home, such as (iv) the type of housing (bungalow, terrace, and apartment), (v) nearby water bodies such as lakes and river which could be breeding sites, (vi) nearby playgrounds as potential breeding sites, (vii) nearby cemetery which could provide potential artificial breeding sites; (viii) vegetation such as bushes, house plants, or shrubs near the houses as habitat and resting areas and (ix) construction areas which can serve as transmission areas and also offer breeding sites if there is any standing water, especially during the rainy season (4,(38)(39). The climatic data such as mean daily rainfall, temperature, relative humidity were obtained from Malaysia Meteorological Department.

Data Analysis
Data were entered and analyzed using Statistical Packages for Social Sciences (SPSS) version 22.0. The analysis undertaken by using this software at different levels. Chi-square tests followed by binary regression were used for sociological and environmental factors for the association of variables, however Pearson's Correlation test was used and followed by multivariate analysis for investigation of association between climate and dengue cases. The level of significance in this study was set at P < 0.05.

Estimation
The Kernel estimation was widely used in several vector borne disease (26,39). The distribution of dengue cases were transformed into density surface by kernel density estimation and it will be illustrated contiguous pattern instead of point patterns. Through the pattern, the risk concentration in the study area was clearly observed for further analysis and the same time can protect the privacy of dengue cases.

Weighted of socio-environmental and climatic
The risk categories were determined according to the variables in the questionnaires. The spatial resolution of socio-environmental factors was set according to the flight range below 800m (according to the risk level) and also based on the risk impact to the dengue infection by the literature review (34,40).
The climatic factors value were set according to the optimum range that impact to the survival and activities of dengue virus carrying the mosquitoes (41)(42). The data of climatic factors were recorded based on the day of respondent start feeling unwell ( Table 1).
The "weightage" of the parameters were given based on the literature review (36,43). The risk levels was given according from high risk to low risk respectively. Each parameter was converted into a raster layer with its weightage by using inverse distance weighing (IDW). IDW interpolation (nearest neighbour technique) was employed to produce the desired results. In the process, a neighbourhood around the interpolated point is identified, the weighted average is taken of the observation values within this neighbourhood (4,42,44).The final risk map was produced by using Weighted Overlay. The risk map is the combination of the risk factor layers with presented of low to high risk areas and it visualizes the risks in relation to each other (45).

Socio-environmental factors for respondent dengue history
A total of 379 respondents with a history of dengue answered the questionnaires provided. The study showed that the majority of the respondents who had only been infected by dengue once were living in a house with more than four people (87.7%), not working (88.8%) and routinely stayed outside during peak biting hours (87.2%) ( Table 2). This study found that there was no significant association between a respondent's dengue history and these sociological factors.
A majority of the respondents who had one instance of dengue infection lived in bungalow houses (90.2%). The study also found that majority of the people who had only experienced one dengue infection lived near a cemetery (89.4%) and also near water bodies such as lakes and rivers (88.3%). Houses of respondents who had only one experience of dengue were not near construction areas (88.6%) ( Table 3).
This study showed that there is a significant association between respondents who had one experience with dengue and lived in houses shaded with vegetation (95% CI = 0.263 and 0.836, p = 0.012) with playground areas near the house (95%CI = 0.304 and 0.859, p = 0.011) as shown in Table 2.  Table 3. However, monthly mean of climate variables revealed inconsistent pattern for the five years analysis (Figure 4,5, 6, 7).

Assessing the impact of socio-environmental and climatic variable
The study indicated spatial distribution of dengue cases at high risk (  (Table 5). Based on the kernel density map, the cemeteries and water bodies mostly not in the high risk and low risk areas. Furthermore, most of them lived in the attached houses with more than four people for both areas.  (49,57). This study collected the significant variables about sociological data such as number people in the house, routine activity during peak biting time and working status. The chi-square indicated majority of the respondent routinely stayed outside during the peak biting and mostly not working people. Students also among not working category. Although the majority among non-working category, however the percentage of working category also high (85.2%). This result was in line with outcome from kernel density estimation, which was high dengue cases areas showed the majority come from working category. Even though, dengue incidence occurred in all age groups, older people have more behavioural risk factors such as travel, work, study, immigration, and leisure activities (58). Daily population movement is an inevitable activity in society. Successful dengue transmission is dependent upon the time period when these people remain in certain areas, as well as the number of humans that visited the viremic areas (33,59). This particular study revealed that the most common areas for dengue-afflicted patients to be are residential areas, work areas, and school. Thus, several preventive control measures can be implemented by the public during peak biting hours, such as the application of insecticides, wearing long-sleeved shirts, and using mosquito repellent (59)(60)(61).
The finding of this study showed that the households with more than four members were majority among the respondent. In Malaysian, the average of household especially Aedes albopictus mosquitoes, will further expose the dangers of dengue and its spread during an outbreak due to the small distance between the source of food and the breeding site, respectively (37,62). Moreover, the present study also proved that flower vases and the lids of paint buckets commonly seen at apartment houses are a notable breeding site. It may worsen the overall condition should residents fail to manage their planting areas properly (63). Generally, the community that lives in apartments and terrace houses plant their trees and flowers in these vases due to the lack of ground areas, as most of the floors in their houses are covered with cement (64)(65). Several studies also revealed that a lesser concentration of vegetation could also serve as suitable resting areas for female mosquitoes, as high rise building apartments revealed high dengue incidences, too (66)(67).
Vegetation also serves as the resting areas for Aedes mosquitoes. This study indicated the presence of vegetation showed the significant result among the respondent's dengue history (p = 0.012). With regard to this, previous studies had reported that vegetation might remarkably impact dengue infection in combination with a myriad of factors causing dengue cases in study areas (23,68). Furthermore, plants and vegetation sites serve as a natural habitat and food source for Aedes mosquitoes as they feed on the nectar of the plants (69)(70). In consideration of their optimum flight range of approximately 200 meters, the growth of Aedes mosquito population will be uncontrollable if any breeding site is present near the human community, especially in the urban areas (43). These urban area having a highly dense population will subsequently result in an exploding mosquito population as the food source (human blood) is easily accessible (71). Therefore, a larger mosquito population will lead to an increased incidence of dengue cases attributable to the growing human-mosquito contact (26).
This study indicated a significant result for the presence of playground areas compared to the frequency of dengue history among the respondents (p=0.011).
The frequency of dengue infection for the respondents is important due to its potential association with the residential areas and their surroundings. The Malaysian Ministry of Health has identified parks (including playground areas), cemeteries, vacant lands, public infrastructure areas and construction sites as the favored breeding sites for these troubling insects (23,66). These areas are typically linked to overcrowding, whereby artificial containers are easily found (39,72). This study revealed that about 89.8% patients lived near playgrounds or parks, which were also concomitantly used as the garbage dumps by the public. Water that is trapped after rainfalls is the perfect point for mosquito oviposition areas (73). These playground areas are typically surrounded by trees, ornamental plants, and shrubs.
Thus, without the elimination of the rubbish dump, man-made containers, and other breeding sources, the female Aedes mosquitoes will oviposit and complete the life cycle. Following this, the number of adult mosquitoes will increase as the areas have already served as a natural habitat for them (3,74). Similarly, crowdgathering places like recreation parks, schools, and markets also played an important role in dengue transmission, particularly in case of an abundance of infected Aedes mosquitoes in such locations (24,74).
The study setting was in the crowded and urbanization areas. The ecological setting for both areas is the combination of residential, commercial and densely populated places with development in both infrastructure and residential housing. According to the study areas, most of the respondent in Parliament Batu lived in terrace houses and bungalow, however, those at Parliament Wangsa Maju mostly lived in an apartment. Terrace houses are usually associated with vegetation such as ornamental and house plants (56). Most of the residential areas are nearby to the playground. The socio-environmental and climatic factors considered in this study were likely responsible for creating the suitable condition for the increased breeding and transmission of aedes mosquitoes and thus increasing its density and spreads.
The transformation of the climate condition for the long period of time including temperature, precipitation, relative humidity, winds, and rainfall may lead to the changes in survival, replication, development and distribution of dengue virus and mosquitoes (56,74). Furthermore, the finding of this study such as vegetation and playground areas are strongly associated with breeding habitat. The prolonged rainfall will increase dengue transmission. Besides, under shrubbery area and lower temperature, the water may take longer period to dry and it will turn into breeding site (75).
The sociology factors in this study commonly gave general information for the Interrupting the larval and pupae stages alike will lead to a decreased number of adult mosquitoes, which would consequently result in a decline in the density of the vector population, and ultimately a reduction of dengue incidence.
The production of risk maps in the study used the spatial distribution of socioenvironmental and climatic factors from IDW maps. The distribution of the factors were calculated based on the optimum flight range distance from the residential areas (4,42). However, when the validation has been made using the kernel density of dengue case distribution, it showed that several factors only distributed nearby residential areas in the optimum range (Table 5). Dengue transmission can be influenced by a combination of several factors in different areas. Even though, not all the risk factors presented in particular areas, the density of Aedes mosquitoes can be increased (75)(76). The risk map and the validation results from this study can be used as a model to predict the factors contributed during the outbreak in urban areas especially.

Conclusion
The impact of socio-environmental and climatic factors for dengue transmission risk in this study gives meaningful additional information for the improvement of current control measures. This principal finding also will beneficial other researchers in academic institutions. The environmental factors that contributed towards dengue transmission uncovered in this study consisted of the playground areas, and vegetation at shaded houses. The climatic analysis revealed significant determinants such as relative humidity, temperature and rainfall. Although only a few factors were found to be significantly associated with dengue incidence, the combination of multi-factorial circumstances was also crucial as they may collectively and indirectly contribute to the abundance of Aedes mosquitoes. As the factors affecting dengue incidence and transmission are diverse, the implementation of dengue control is challenging. Significant sociological and environmental variables should be taken into consideration to better formulate local dengue control and preventive measures for the community and health authorities.

Ethics approval and consent to participate
The ethical approval number for this research: UKM PPI/111/8/JEP-2016-393

Consent for publication
Not applicable

Availability of data and material
No applicable

Competing interests
The authors declare that there is no conflict of interests regarding the publication of this paper.

No applicable
Author's contributions 1) Ruhil Amal Adnan carried out the experiment, performed the data analysis, and wrote the manuscript.  Map of study areas in Kuala Lumpur, Malaysia.

Figure 2
Map of study areas in Kuala Lumpur, Malaysia.

Figure 3
Flow diagram of respondent selection The risk map produced showed localities with high risk areas and low risk areas.

Figure 9
The risk map produced showed localities with high risk areas and low risk areas.