Geographically Weighted Generalized Poisson Regression untuk Memodelkan Kasus Pneumonia Balita di Jawa Timur dengan Exposure

Pneumonia adalah bentuk infeksi saluran pernapasan akut yang menyerang paru-paru. Pneumonia balita menjadi salah satu penyebab kematian balita tertinggi di Indonesia, serta banyak ditemukan di Jawa Timur. Oleh karena itu, perlu dilakukan penelitian untuk mengetahui faktor-faktor penyebab tingginya pneumonia balita di Jawa Timur. Penelitian dilakukan dengan mempertimbangkan jumlah balita sebagai exposure karena tiap daerah memiliki jumlah balita yang berbeda. Penelitian ini menggunakan analisis Geographically Weighted Generalized Poisson Regression (GWGPR). Analisis GWGPR merupakan pengembangan dari Generalized Poisson Regression (GPR) namun menggunakan pembobot geografis yang dapat mengatasi keragaman wilayah/heterogenitas spasial sehingga menghasilkan model dan pendugaan parameter yang berbeda untuk setiap wilayah. GWGPR dipertimbangkan karena mampu (1) mendapatkan faktor melalui analisis regresi, (2) mengakomodasi efek spasial yang ada pada kasus pneumonia, dan (3) mengakomodasi perbedaan jumlah balita yang ada pada tiap daerah melalui exposure. Pada penelitian ini dilakukan pemodelan kasus pneumonia balita di Jawa Timur menggunakan GWGPR dengan variabel exposure yaitu jumlah balita dan pembobot Adaptive Bisquare Kernel. Jumlah kasus pneumonia tertinggi yaitu sebanyak 9308 kasus. Hasil pemodelan GWGPR exposure memiliki AICc sebesar 622,4103 membentuk dua kelompok kabupaten/kota berdasarkan variabel yang berpengaruh signifikan, yaitu persentase gizi kurang, persentase imunisasi dasar lengkap, persentase pemberian ASI eksklusif, persentase rumah tangga dengan jenis lantai tanah, dan persentase rumah tangga dengan bahan bakar kayu, arang, dan minyak tanah yang berpengaruh signifikan di semua kabupaten/kota. ======================================================================================================================== Pneumonia is a form of acute respiratory infection that affects the lungs. Under-five pneumonia is one of the highest causes of under-five mortality in Indonesia, and is prevalent in East Java. Therefore, it is necessary to conduct a study to determine the factors that cause the high rate of under-five pneumonia in East Java. The study was conducted by considering the number of under-fives as exposure because each region has a different number of under-fives. This study uses Geographically Weighted Generalized Poisson Regression (GWGPR) analysis. GWGPR analysis is a development of Generalized Poisson Regression (GPR) but uses geographic weights that can overcome regional diversity/spatial heterogeneity resulting in different models and parameter estimates for each region. GWGPR is considered because it is able to (1) obtain factors through regression analysis, (2) accommodate the spatial effects of pneumonia cases, and (3) accommodate differences in the number of children under five in each region through exposure. This study modeled under-five pneumonia cases in East Java using GWGPR with the exposure variable being the number of under-fives and Adaptive Bisquare Kernel weighting. The highest number of pneumonia cases was 9308 cases in Sidoarjo Regency while the lowest number of cases was 63 cases in Pasuruan Regency. The results of GWGPR exposure modeling with AICc of 622.4103 formed two groups of districts based on variables that had a significant effect, namely the percentage of undernutrition, the percentage of complete basic immunization, the percentage of exclusive breastfeeding, the percentage of households with soil floor type, and the percentage of households with wood, charcoal, and kerosene fuel that had a significant effect in all districts