Analysis of maternal mortality rate using multivariate adaptive generalized Poisson regression spline

Abstract

Maternal mortality is an essential indicator for assessing public health levels and the quality of healthcare services in a country. In South Sulawesi Province, maternal mortality reached 198 cases in 2021, increasing from 133 cases in 2020. This high mortality rate requires urgent attention to identifying contributing factors. Maternal mortality data is discrete categorical data that follows a Poisson distribution and experiences overdispersion. This nonlinear and random data pattern makes it more suitable for nonparametric analysis. The appropriate method for these data characteristics is Multivariate Adaptive Generalized Poisson Regression Splines (MAGPRS). This study aims to identify the factors influencing maternal mortality in South Sulawesi Province using the MAGPRS method. The best model was obtained with a combination of basis function (BF) of 20, Maximum Interaction (MI) of 3, and Minimum Observation (MO) of 2, yielding a Generalized Cross-Validation (GCV) value of 0.000046683 and an R-Square of 99.30%, indicating high model accuracy. The findings reveal that the most influential factors include the percentage of pregnant women in the K4 program (X₁), those receiving Td3 immunization (X₅), postpartum mothers receiving vitamin A (X₄), deliveries in healthcare facilities (X₃), and pregnant women in the K1 program (X₂).