Analysis of the spread of dengue hemorrhagic fever with the moran index (Case study of Sukoharjo Regency in 2019)

One of the endemic diseases that can be found in tropical areas is DHF or Dengue Hemorrhagic Fever (DHF). Dengue fever can be transmitted through Aedes aegypti mosquitoes that have been infected with the dengue virus. It is estimated that the level of dependence on DHF is always related to the infected area around it. DHF can spread to other people through mosquito bites. Sukoharjo Regency in 2018 recorded 35 cases of DHF sufferers and zero patients died. Increased up to 90% in 2019, there were 317 cases of dengue fever and 10 patients died. Therefore, the purpose of this study is to determine the autocorrelation of the spread of DHF in Sukoharjo Regency in 2019. This study used data on the number of DHF sufferers per subdistrict in Sukoharjo Regency in 2019. The research method used the calculation of the Moran Index. Calculation of Moran’s Index produced a value I = -0.181. Moran’s index value is in the range - 1≤I< 0 so that it indicates a negative autocorrelation. The correlation is classified to be a weak relationship. Negative spatial autocorrelation values indicate locations that are close to each other have values that are not close together. Subdistricts with a high number of sufferers tended to be irregular.


Introduction
Dengue Hemorrhagic Fever (DHF) is one of the endemic diseases that can be found in tropical areas [1].Dengue fever is caused by the dengue virus, which is transmitted to humans primarily through the bites of infected female mosquitoes, mainly the Aedes aegypti mosquito [2].The dengue virus does not develop at the base of the mosquito's proboscis [3].Instead, the virus replicates and multiplies in the mosquito's midgut after it feeds on the blood of an infected person [4].The virus then moves to the salivary glands of the mosquito, from where it can be transmitted to another person during a subsequent mosquito bite [5].
The level of dependence on DHF is related to the presence and abundance of the Aedes mosquito population in a given area [6].The more prevalent these mosquitoes are, the higher the risk of dengue virus transmission and subsequently the development of DHF cases in the affected population [7].When applied to the spread of diseases like DHF, Tobler's first law helps us understand that the risk of transmission is higher in areas close to the location where the disease is already present [8].Mosquitoborne diseases like dengue usually spread locally through the movement of infected mosquitoes from one area to another, with a higher likelihood of transmission occurring in nearby regions [9].
The number of DHF cases in Sukoharjo Regency based on the data reported in 2015-2019 is more than 100 cases per year (Table 1).The data shows a fluctuating pattern in the number of DHF cases and deaths over the five-year period.The highest number of DHF cases and deaths were recorded in 2016, 1314 (2024) 012071 IOP Publishing doi:10.1088/1755-1315/1314/1/012071 2 with 545 cases and 13 deaths.However, there was a significant decrease in cases and deaths in 2017, with only 115 cases and 2 deaths.The year 2018 saw the lowest number of cases with 35 reported, and fortunately, there were no deaths recorded in that year.However, in 2019, there was a notable increase in cases, with 317 individuals affected by DHF, and 10 deaths were reported [10].Hence, it indicates the importance of addressing and controlling the spread of the disease in this region.This fluctuating trend in the number of cases and deaths over the years suggests that the transmission of DHF in Sukoharjo Regency is subject to various factors that may vary from year to year.It also highlights the importance of continuous monitoring and surveillance to identify outbreaks and promptly implement appropriate control measures for 12 existing subdistricts.
Geographic information system (GIS) and remote sensing can be used to study spatial aspect of health [11][12][13][14][15][16].The use of the Moran Index to determine the spatial pattern of the spread of DHF in Sukoharjo Regency is a valuable approach in understanding the geographical distribution and identifying possible clusters of cases [17].The Moran Index is a common method in spatial statistics that helps to assess the presence of spatial autocorrelation, which means the tendency of similar values to cluster together in geographic space [18].
The Moran Index measures the degree of correlation between the values of a variable at each location and the values at neighboring locations [19].In the context of DHF in Sukoharjo Regency, it can help for identifying whether areas with high DHF cases are more likely to be surrounded by areas with high DHF cases (positive spatial autocorrelation), or if areas with high DHF cases are surrounded by areas with low DHF cases (negative spatial autocorrelation) [20].Overall, the use of the Moran Index in studying the spatial pattern of DHF cases in Sukoharjo Regency can provide valuable insights into the geographical distribution of the disease and help inform evidence-based public health interventions [18].
The studies carried out by Hernawati [21] on the spatial distribution of DHF in the city of Bandung and Habinuddin's research [22] on the identification of spatial autocorrelation in DHF spread in Bandung City reveal different results regarding the presence of autocorrelation.While Hernawati found both Cluster quadrant I (High-High) and quadrant III (Low-Low) patterns, indicating spatial autocorrelation in DHF cases, Habinuddin's study did not detect autocorrelation using Gaery's C method [21,22].
Additionally, the study conducted by Mailanda et al. [23] on the Autocorrelation Analysis of COVID-19 positive cases using the Moran Index and LISA (Local Indicators of Spatial Autocorrelation) found positive spatial autocorrelation with clustering patterns and characteristic similarities at adjacent locations.This indicates that COVID-19 cases tended to cluster in specific geographic areas.Another study on the spatial distribution of malaria parasite prevalence in Ethiopia used the Queen contiguity second order type of spatial weight matrix to formalize spatial interaction among subdistricts.The study found positive spatial autocorrelation in malaria prevalence rate and identified hot spot areas for malaria parasite prevalence in the eastern and southeast parts of the region [24].
Based on the aforementioned background, this study aims to calculate the Moran Index and create a distribution map using the data from the Sukoharjo Regency Health Office.By employing the Moran Index, researchers can gain insights into the spatial pattern of DHF cases, identifying whether they exhibit clustering (high-high or low-low) or dispersion (high-low or low-high) patterns [25].
Understanding the spatial distribution of DHF cases in Sukoharjo Regency through the Moran Index can help public health authorities to target resources and interventions more effectively.By identifying areas with high DHF cases and clustering patterns, authorities can implement focused strategies to control the spread of the disease, reduce its impact, and prevent further outbreaks.

Research areas and data used
Fig. 1 shows the research study area is Sukoharjo Regency were located in Central Java Province, adjacent to Gunung Kidul and Daerah Istimewa Yogyakarta.The regency is divided into 12 subdistricts and comprises a total of 167 villages.The administrative area of Sukoharjo Regency is 46,666 hectares, with Sukoharjo subdistrict being the largest among the other subdistricts [10].
As of 2020, the population of Sukoharjo Regency is 907,587 people, and it has shown consistent population growth over the past 10 years.Kartasura is the most densely populated subdistrict in Sukoharjo, with a population density of 6,035 people per km 2 , while Bulu Subdistrict has the lowest population density, with 778 people per km 2 , compared to the other subdistricts [10].

Tools and materials
Table 2 and Table 3 is a list of tools and materials used in this study.

Data processing and analysis
This study utilized secondary data sourced from the Sukoharjo Subdistrict Health Office.Table 4 shows data on the number of DHF sufferers per subdistrict in Suko-harjo Regency in 2019.

Weighting Matrix.
The immensity of the relationship between locations is determined based on the weighting matrix (W) [26].The Queen Contiguity method is one of the common approaches used in spatial analysis and spatial statistics [27].It is a type of spatial weighting matrix used to represent the spatial relationships between different locations or spatial units [28].In this method, the weighting matrix is constructed such that neighboring locations have equal weights, while non-adjacent locations have zero weights [24].The observation area is seen based on the closeness between the sides and corners.The matrices are then standardized to produce Table 5 which shows the queen matrix continuity standards [28].
Table 5. Matrix of queen contiguity standardized Moran's Index.Moran's Index is a measure of spatial autocorrelation, which quantifies the degree of spatial clustering or dispersion in a dataset.It helps to identify whether the spatial pattern of a variable exhibits clustering (positive correlation) or dispersion (negative correlation) [20].The Moran index is measured using the following equation below.
Under the conditions: I is Moran's Index n is the number observation locations xi is the observed value at the location i xj is the observed value at the location-j  is the average value of observations wij is weighting matrix element between locations i dan j The formula for Moran's Index involves comparing each data point's value with the average value of its neighbours.The resulting index ranges between -1 and 1, with specific interpretations.First, a positive Moran's Index value (close to +1) indicates a spatial pattern of positive autocorrelation.In other words, similar values tend to cluster together on the map.For example, in a regional study, high-income areas might be surrounded by other high-income areas.Second, a negative Moran's Index value (close to -1) indicates a spatial pattern of negative autocorrelation.This means that dissimilar values cluster together, while similar values are dispersed.For instance, in an environmental study, areas with high pollution levels might be surrounded by areas with low pollution levels.Third, a Moran's Index value close to 0 indicates spatial randomness, implying that there is no significant spatial correlation present in the data.The values are scattered across the map without a discernible pattern (Kun, 2007).
Based on this explanation can be formulated the Moran Index parameters that can be seen below.H0 : I = 0 (There is no spatial autocorrelation) H1 : I ≠ 0 (There is spatial autocorrelation) 2.3.3.Moran Scatterplot.The Moran Scatterplot is a useful visualization tool used to explore the spatial autocorrelation of geographic data.It helps to visualize the relationship between the values of a variable at different locations and their spatial interactions [20].In a Moran Scatterplot: (1) Each data point on the horizontal axis corresponds to a specific location or spatial unit.(2) The vertical axis displays the spatially lagged values of the same variable.Spatial lag refers to the average value of the variable in the neighbouring locations surrounding each data point on the map.(3) Each data point on the scatterplot represents a spatial unit and is placed based on its value on the horizontal axis and the corresponding spatially lagged value on the vertical axis [20].
The Moran Scatterplot provides an intuitive and visual way to assess the spatial autocorrelation and understand how the values of a variable are related to their neighbouring locations on the map.It complements the quantitative analysis using Moran's Index and helps researchers gain insights into the spatial patterns present in the dataset [19].Fig. 2 explains the quadrants in Moran Scatterplot.The research flowchart for this study is presented in Figure 3.It can be explained that the initial data for the study consists of two main datasets: (a) Admin shp in Sukoharjo Regency: It refers to administrative boundary data in shapefile format, containing the boundaries of subdistricts or other spatial units within Sukoharjo Regency.(b) Data on DHF sufferers per subdistrict: This dataset contains information on the number of DHF cases reported in each subdistrict of Sukoharjo Regency.Second, the process of determining the neighbourliness of each subdistrict involves creating a spatial weight matrix.This matrix defines the spatial relationships between the sub-subdistricts based on their geographical proximity or some other criterion.Common approaches for constructing the spatial weight matrix include distance-based weights.Third, the standardized spatial weight matrix is used in the calculation of Moran's Index.The standardization process helps to ensure that the index values are not affected by the scale of the data or the number of neighbours for each subdistrict.

Figure 3. Research Flowchart
Moran's Index is computed to quantify the spatial autocorrelation of DHF cases across Sukoharjo Regency.The index value will indicate whether DHF cases are clustered, dispersed, or randomly distributed across the subdistricts.Fifth, the results of Moran's Index can be further visualized using a Moran's Scatterplot.This plot helps to visualize the relationship between the DHF case counts in each subdistrict and their spatially lagged values (i.e., the DHF cases in neighbouring subdistricts).The pattern observed in the scatterplot will provide insights into the spatial autocorrelation and clustering tendencies of DHF cases.The final step involves visualizing the spatial distribution of DHF cases in Sukoharjo Regency on a map.This map can help researchers and stakeholders better understand the spatial patterns and hotspots of DHF cases across the region.
By following this research flowchart (Fig. 3) and conducting the spatial analysis with Moran's Index and related techniques, researchers can gain valuable insights into the spatial characteristics of DHF occurrences, identify areas with higher risk, and potentially aid in designing targeted interventions or public health strategies to tackle the disease in the region.

Results Weighting Matrix
In spatial analysis, a neighborhood relationship refers to the spatial connectivity or adjacency between different spatial units or regions.Queen Contiguity is a type of spatial weighting method that defines neighboring relationships based on shared boundaries, including both shared edges and shared corners.The weight between locations in this context indicates the degree of spatial proximity or connectivity between them [30].
Figure 4. is neighborhood matrix histogram representing the distribution of the number of neighbouring groups based on their respective number of neighbours.The x-axis displays the number of neighbouring groups, and the y-axis shows the count (number) of neighbouring subdistricts falling into each group.For instance, Group 2 and Group 4 both have four neighbours, meaning they are surrounded by four neighbouring subdistricts.On the other hand, Group 5 has only one neighbour, implying that it is surrounded by just one neighbouring subdistrict.The histogram is generated using GeoDa to visualize the neighborhood relationships based on the Queen Contiguity method.GeoDa is a popular software program commonly used for spatial data analysis, including spatial statistics and spatial visualization [27].
Based on the information in the Table 6, it is known that in group 1, represented by the light blue color, consists of four neighboring subdistricts: Weru, Mojolaban, Kartasura, and Gatak.These subdistricts are connected to each other, forming a group with 2 neighbors each.Group 2, represented by the dark blue color, includes three closest neighboring subdistricts: Bulu, Polokarto, and Baki.Each of these subdistricts is connected to the other two, forming a group with 3 neighbors each.Group 3, represented by the light green color, has four neighbors, which are Tawangsari, Nguter, Sukoharjo, and Bendosari subdistricts.Group 4, represented by the dark green color, has the highest number of neighbors, consisting of only one subdistrict, Grogol.This subdistrict is surrounded by five neighboring subdistricts.The table summarizes the neighbourhood relationships and their corresponding spatial units or regions, providing insights into the spatial connectivity and clustering of neighboring subdistricts based on the Queen Contiguity method and the resulting histogram.

Results Moran's Index
Moran's Index measures the degree of spatial autocorrelation, which indicates the similarity or dissimilarity of attribute values among neighboring locations in a spatial dataset.The index ranges from -1 to +1, where positive values indicate positive spatial autocorrelation (similar values cluster together), negative values indicate negative spatial autocorrelation (dissimilar values cluster together), and a value close to zero indicates a lack of spatial autocorrelation [19].
In this case, the Moran's Index value of -0.181 suggests a weak negative spatial autocorrelation among the regions or subdistricts being analyzed.This means that neighboring subdistricts tend to have dissimilar values for the attribute being studied.The information provided in Figure 5. and the interpretation of Moran's Index and the quadrant patterns help understand the spatial autocorrelation and the distribution of attribute values across the studied regions or subdistricts.Based on the study results and the distribution pattern, it is suggested that there is a negative spatial autocorrelation for DHF in Sukoharjo Regency.Negative spatial autocorrelation means that the occurrences of DHF are not clustered in Hot-Spots or Cold-Spots, but rather distributed across spatial outlier areas.These results can refer to previous research, likely provides more detailed insights into the methodology and implications of the results [31,32].The Moran Scatterplot is a graphical representation of spatial autocorrelation, and by categorizing the subdistricts into these four categories based on their spatial relationships and the number of DHF cases, it helps to identify the different spatial patterns of disease occurrences in Sukoharjo Regency.This information is crucial for public health officials to understand the distribution of the disease and plan targeted interventions to address areas with high occurrences or clusters [33].

Discussions
The Moran Index value obtained is -0.18, which falls within the range of -1 ≤ I < 0. This negative value indicates a negative spatial autocorrelation.In the context of this study, negative spatial autocorrelation means that adjacent subdistricts (locations) have dissimilar values of DHF sufferers [25].
IOP Publishing doi:10.1088/1755-1315/1314/1/01207111 Furthermore, subdistricts with a high number of DHF sufferers are scattered and do not exhibit strong spatial clustering.The concentration of subdistricts in Quadrants II and IV, as spatial outliers, suggests that areas with different characteristics (high occurrences of DHF) are adjacent to areas with low occurrences.This confirms the weak negative spatial autocorrelation observed in the study [18,25].
Overall, the negative spatial autocorrelation and dispersed spatial pattern indicate that the distribution of DHF in Sukoharjo Regency does not exhibit strong spatial clustering.Instead, it shows irregular occurrences across the subdistricts.Understanding such spatial patterns is crucial for public health planning and intervention strategies, especially when dealing with diseases like DHF that can have varying impacts on different areas [18,25].
Indeed, DHF is a significant public health concern in tropical and sub-tropical regions, including Indonesia, where Sukoharjo Regency is located.The presence of two seasons (wet and dry) in tropical climates contributes to the yearly occurrence of DHF cases.The rainy season, in particular, creates favourable conditions for the breeding of Aedes aegypti mosquitoes, the primary vector responsible for transmitting the virus that causes DHF [34].
Within the five-year period from 2015 to 2019, there were 1,327 cases of DHF reported in Sukoharjo Regency, and out of those cases, 48 individuals unfortunately died due to the disease.Various environmental factors, such as humidity, wind, and rainfall, play a crucial role in the development and spread of Aedes aegypti mosquitoes.The accumulation of standing water in containers and other areas provides breeding sites for mosquitoes, increasing the risk of DHF transmission.Additionally, population density, mobility, and behaviour can influence the spread of the disease, especially in densely populated and unplanned urbanized areas [34][35][36].
While the spatial autocorrelation analysis revealed a weak negative spatial autocorrelation for DHF in Sukoharjo Regency, it is essential to continue efforts to control and prevent the spread of the disease.Early awareness and education about DHF are crucial to promote preventive measures among the community.Eradicating mosquito breeding sites and controlling mosquito populations remain key priorities for DHF control [35,37].
Fogging, which involves the use of insecticides to control adult mosquitoes, is one method that can be used, although its effectiveness may be limited.Thus, more sustainable approaches, such as source reduction (eliminating breeding sites), using larvicides, and community participation in maintaining hygiene and environmental cleanliness, are essential for long-term DHF control [38,39].
Public health authorities, in collaboration with the community, need to implement comprehensive strategies to combat DHF, considering the various factors that contribute to its spread.Monitoring and surveillance of the disease, as well as the implementation of integrated vector management programs, are essential components of an effective DHF control strategy [40].

Conclusions
Based on calculations, the Moran Index value I= -0,181.Moran's index value is in the range -1≤I < 0 so that it indicates a negative autocorrelation but the correlation can be said to be weak because it is close to zero.Negative spatial autocorrelation which can be interpreted if adjacent locations have dissimilar values, in this case the subdistricts with a high number of DHF sufferers are scattered and irregular.Subdistricts with a high number of sufferers also tend to be irregular.The spread of DHF can be influenced by several factors, including environmental factors, the host and the virus itself.

Figure 6 .
Figure 6.Map of the distribution DHF in Sukoharjo Regency in 2019

Table 2 .
Tools used in research

Table 3 .
Data used in research

Table 6 .
Subdistrict and number of neighbors