The Implementation of Hierarchical Linear Spline Regression Model to Maternal Mortality Rate Data in Indonesia

Abstract

Hierarchical spline nonparametric regression analysis is one regression method to identify the correlation pattern among the variables with unidentified curves and nested data structure using a spline truncated regression approach on its function. On the maternal mortality rate, there is a nested data structure between the regional/ city level and provincial level. This study aimed to identify the correlation between the response variable and predictor variable from the maternal mortality rate data in 2020. There are two analyzed levels consisting of the city/ regional level (level-1) with 256 cities/ regencies and the provincial level (level-2). The level-1 predictor variable is total pregnant women taking blood booster supplements and total pregnant women attending K4 antenatal care and the level-2 predictor variable is poverty rates and school enrollment rates. A spline truncated approach has high flexibility and this study employs Maximum Likelihood as its estimator. The finding suggested that if a province has a lower poverty rate than 7.24%, maternal mortality rates tend to increase. If a province has lower enrolment rates than 67.8 %, the maternal mortality rate tends to decrease. If a regency or a city has fewer pregnant women taking blood booster supplements than 2258, maternal mortality rates tend to decrease. If a regency or a city has fewer pregnant women attending K4 antenatal care than 2258, the maternal mortality rate tends to decrease.