Pemodelan Spasial Bayesian dalam Menentukan Faktor yang Mempengaruhi Kejadian Stunting di Provinsi Sulawesi Selatan

Abstract

Indonesia merupakan negara dengan prevalensi balita stunting yang tinggi. Salah satu provinsi di Indonesia yang memiliki kasus stunting yang cukup tinggi adalah Provinsi Sulawesi Selatan. Penelitian mengenai kasus stunting dan faktor penyebabnya telah dilakukan. Namun, penelitian tersebut belum mengimplementasikan model Bayesian spasial Conditional Autoregressive (CAR). Penelitian ini bertujuan untuk mengetahui faktor yang mempengaruhi kejadian stunting di Provinsi Sulawesi Selatan dengan mengimplementasikan berbagai model Bayesian spasial CAR Leroux tanpa kovariat dan dengan memasukkan kovariat dalam model. Hasil penelitian menunjukkan bahwa model terbaik dalam memodelkan kasus stunting di Provinsi Sulawesi Selatan tahun 2020 adalah model Bayesian spasial CAR Leroux dengan hyperprior Inverse-Gamma IG(0,5;0,0005) dengan memasukkan kovariat persentase kemiskinan dan persentase balita 0-59 bulan gizi kurang. Persentase kemiskinan dan persentase balita 0-59 bulan gizi kurang berpengaruh positif terhadap kejadian stunting. Semakin tinggi persentase kemiskinan dan persentase balita 0-59 bulan dengan gizi kurang di suatu wilayah, semakin tinggi risiko stunting di wilayah tersebut. 50% kabupaten/kota di Provinsi Sulawesi Selatan berada dalam kategori risiko tinggi stunting. Kota Parepare merupakan kota dengan nilai risiko relatif (RR) tertinggi stunting, diikuti oleh Kabupaten Toraja dan Enrekang. Sebaliknya, Kabupaten Wajo merupakan kabupaten dengan RR terendah, diikuti oleh Kabupaten Luwu Timur dan Bone.Kata Kunci: Stunting, Bayesian, spasial CAR, Leroux  Indonesia is a country with a high prevalence of stunting. One of the provinces in Indonesia that has a fairly high number of stunting cases is South Sulawesi Province. Research on stunting cases and their causes has been done. However, these researches have not implemented the Bayesian Spatial Conditional Autoregressive (CAR) model. This study aims to determine the factors that influence the incidence of stunting in South Sulawesi Province by implementing various Bayesian spatial CAR Leroux models with and without covariates included in the model. The results showed that the best model for modeling stunting cases in South Sulawesi Province in 2020 is the Bayesian spatial CAR Leroux model with hyperprior Inverse-Gamma IG (0.5;0.0005) by including the covariates of the percentage of poverty and the percentage of children under five 0-59 months of malnutrition. The percentage of poverty and the percentage of children under five 0-59 months of malnutrition have a positive effect on the incidence of stunting. The higher the percentage of poverty and the percentage of children aged 0-59 months with malnutrition in an area, the higher the risk of stunting in that area. 50% of districts/cities in South Sulawesi Province are in the high-risk category of stunting. Parepare City is the city with the highest Relative Risk (RR) value for stunting, followed by Toraja and Enrekang Regencies. On the other hand, Wajo Regency is the district with the lowest RR, followed by Luwu Timur and Bone Regencies.Keywords: Stunting, Bayesian, spatial CAR, Leroux