|
 |
| BRIEF RESEARCH ARTICLE |
|
| Year : 2022 | Volume
: 66
| Issue : 4 | Page : 501-503 |
|
|
Spatial Modeling of leprosy disease in east java province with spatially varying regression coefficients models
Husnul Chotimah1, IG N. Mindra Jaya2
1 Master Scholar, Departemen of Statistics, University Padjadjaran, Bandung; Staff, Statistics Indonesia, Probolinggo City, East Java, Indonesia
2 Master Scholar, Departemen of Statistics, University Padjadjaran, Bandung, Indonesia; Doctor Scholar, Faculty of Spatial Science, Groningen University, Groningen, Netherlands
| Date of Submission | 04-Oct-2021 |
| Date of Decision | 07-May-2022 |
| Date of Acceptance | 16-Oct-2022 |
| Date of Web Publication | 31-Dec-2022 |
Correspondence Address:
Husnul Chotimah
Statistics Indonesia, Probolinggo City, East Java
Indonesia
 Source of Support: None, Conflict of Interest: None  | Check |
DOI: 10.4103/ijph.ijph_1887_21
 Abstract | | |
Indonesia ranks third with the most leprosy cases globally. East Java is the province that has the highest leprosy cases. The Provincial Government socialized the East Java Leprosy Eradication Program, which targets a maximum of one leprosy case per 10,000 residents. We propose spatially varying regression coefficients models to evaluate the effects of risk factors on of leprosy cases in East Java, use Geographically Weighted Generalized Poisson Regression and Geographically Weighted Negative Binomial Regression (GWNBR) models. The best models GWNBR categorize municipalities into six groups based on variables that have a significant impact on leprosy cases. The percentage of households with access to adequate sanitation is a significant factor in determining leprosy cases in all municipalities in East Java. We can conclude that clean and healthy living behavior, health facilities, and health workers significantly affect the number of leprosy cases in East Java.
Keywords: Geographically Weighted Generalized Poisson Regression, geographically Weighted Negative Binomial Regression, leprosy
How to cite this article:
Chotimah H, N. Mindra Jaya I G. Spatial Modeling of leprosy disease in east java province with spatially varying regression coefficients models. Indian J Public Health 2022;66:501-3 |
How to cite this URL:
Chotimah H, N. Mindra Jaya I G. Spatial Modeling of leprosy disease in east java province with spatially varying regression coefficients models. Indian J Public Health [serial online] 2022 [cited 2023 Feb 4];66:501-3. Available from: https://www.ijph.in/text.asp?2022/66/4/501/366579 |
According to the WHO, Indonesia ranks third in the world in terms of leprosy cases, and the Indonesian Ministry of Health reports that East Java has the highest rate of leprosy cases. Cumulatively, East Java accounts for 24% of all leprosy cases in Indonesia.[1] Leprosy is dubbed the “most feared disease” due to the fact that it can result in physical disability and, of course, economic, social, and environmental problems. The Indonesian Ministry of Health's strategic plan targets the eradication of leprosy in all municipalities by 2024.[2] As part of its commitment to reducing leprosy prevalence in East Java, the East Java Provincial Government socialized the East Java Leprosy Eradication Program, which targets a maximum of one leprosy case per 10,000 residents.[3] As a result, proper analysis is required to ensure that the program is on target and operates optimally.
As count data, the number of leprosy cases follows the Poisson distribution, and Poisson regression is extremely appropriate in the equidispersion condition (mean of response variable equal to its variance).[4] Leprosy cases are prone to spatial instability, spatial data analysis can be used to overcome it.
This study will analyze using the Geographically Weighted Generalized Poisson Regression (GWGPR) method and Geographically Weighted Negative Binomial Regression (GWNBR). The purpose of this study is to compare the two methods. In essence, both methods are capable of overcoming the overdispersion problem associated with Poisson regression with spatial effects. The distinction between the two methods, however, is in the distribution function.
The data used is the number of new leprosy cases in 2019 as a response variable (Y) derived from the East Java Health Profile 2019. The predictor variables are the percentage of households with access to adequate sanitation per municipalities (X1), the percentage of households with access to adequate drinking water sources per municipalities (X2), all of which are sourced from BPS East Java (https://jatim.bps.go.id/), the number of health facilities per municipalities (X3), and the percentage of doctors in health facilities (X4), and the percentage of food management places that have health requirements (X5) are sourced from the 2019 East Java Health Profile (https://dinkes.jatimprov.go.id/).
Tobler's first law states that “everything is related, but that which is close has more influence than that which is far away,” implying spatial dependence.[5] Moran's I have the ability to detect spatial dependence. Each location, however, has its own distinct characteristics. Each location has a unique structure, parameters, and functional forms, demonstrating the spatial heterogeneity detectable by the Breusch–Pagan Test.
GWNBR is the most frequently used method for modeling spatially varying regression coefficients. It is a negative binomial regression extension. One method that is quite effective when dealing with spatial heterogeneity in overdispersion count data.[6] Maximum Likelihood Estimation (MLE) is used in conjunction with the following probability function:[7]
Testing the significance of the GWNBR model parameters consists of simultaneous and partial tests. Simultaneous significance test using Maximum Likelihood Ratio Test.
The GWGPR model is a model that uses geographic weighting in its parameter estimation and produces different parameters for each location. The method used in the estimation is MLE, which requires a Newton Raphson numerical iteration procedure. The form of the GWGPR equation is as follows:[8]
East Java has 8 new cases discovered for every 100,000 residents. The majority, municipalities with a high rate of new leprosy cases are located on Java's north and south coasts. Coastal areas are places where fishers inhabit the majority with low health standards. The results of the National Socioeconomic Survey of Indonesia show that the coastal areas have a percentage of households with access to adequate sanitation relatively inadequate than others.[9] In addition, the results of the GWNBR and GWGPR models indicate that the variable percentage of households with access to adequate sanitation is a significant factor affecting the number of people living with leprosy in all municipalities in East Java. Clean and healthy living behavior is a factor to prevent the spread of leprosy.[10]
The existence of spatial effects in the case of leprosy in East Java resulted in precise and effective spatial data analysis be used. In detail, the GWNBR model can classify 38 municipalities in East Java province into six groups in [Figure 1]. Each group contains municipalities that are close to each other and have neighborly relationships. Households with access to adequate sanitation, households with access to adequate drinking water sources, health facilities, doctors in health facilities, and food management places with health requirements are variables that significantly effect on many municipalities. In comparison, the GWGPR method classified municipalities into three groups [Figure 2] based on variables that significantly effect on the number of leprosy cases. The majority of the municipality have households with access to adequate sanitation and doctors in health facilities that significantly affect. GWNBR and GWGPR shows that health facilities and health workers significantly affect the number of leprosy cases in addition to clean and healthy living behavior. | Figure 1: Thematic map of the municipalities grouping in East Java based on significant variables of the GWNBR model. GWNBR: Geographically Weighted Negative Binomial Regression.
Click here to view |
 | Figure 2: Thematic map of the municipalities grouping in East Java based on significant variables of the GWGPR model. GWGPR: Geographically Weighted Generalized Poisson Regression.
Click here to view |
GWNBR and GWGPR have different results in modeling and municipalities classification. Therefore, the best selection criteria are needed to determine the most appropriate models. The minimum Akaike information criterion (AIC) result of 369.17 indicates that the GWNBR is more appropriate for modeling the factors that affect the number of leprosy and classifying districts based on the variables that influence them. GWNBR can accommodate differences in municipalities characteristics, environment, and socioeconomic conditions. So that each municipality has different factors that significantly affect the number of leprosy cases.
This research only involves clean and healthy living behavior factors and the availability of health facilities. Hopefully, further research will involve other factors such as socioeconomic factors.
Leprosy is “most feared disease” due to the impact to be one of the most critical health problems in East Java province. The high number of cases requires the right strategy to solve it. The following conclusions are drawn from the analysis. The GWNBR has a minimum AIC and is therefore the best model for determining the factors influencing leprosy cases in East Java. Based on the model, the factors influencing the number of leprosy cases in East Java are clean and healthy living behavior, health facilities, and health workers.
Financial support and sponsorship
Nil.
Conflict of interest
There are no conflicts of interest.
| References | |  |
| 1. | |
| 2. | East Java Provincial Health Office. East Java Province Health Profile 2019. Surabaya: Health Office; 2020.  |
| 3. | |
| 4. | Agresti A. Categorical Data Analysis. 2 nd ed. New York: John Wiley & Sons; 2002. |
| 5. | Anselin L. Spatial Econometrics: Methods and Models. The Netherlands: Kluwer Academic Publishers; 1988. |
| 6. | Fitriani R, Gede Nyoman Mindra Jaya I. Spatial modeling of confirmed COVID-19 pandemic in East Java Province by geographically weighted negative binomial regression. Commun Math Biol Neurosci 2020;2020:1-17. |
| 7. | Silva AR, Rodrigues TC. Geographically weighted negative binomial regression-incorporating overdispersion. New York: Springer; 2013. |
| 8. | Berliana SM, Purhadi, Sutikno, Rahayu SP. Parameter estimation and hypothesis testing of geographically weighted multivariate generalized poisson regression. Mathematics 2020;8:1523. |
| 9. | Statistics Indonesia. Housing and Environmental Health Indicators 2020. Jakarta: Statistics Indonesia; 2020. |
| 10. | |
[Figure 1], [Figure 2]
|
