Spatial linear regression models in the infant mortality analysis

Abstract

Infant mortality is one of the main concerns for governments in programs of public health. It is also an important measure used to evaluate the quality of life in several countries. The aim of this paper is twofold: first, study the spatial distribution of infant mortality in a Brazilian city using spatial lattice methods. Secondly, propose a new method based on the square root transformation in the response variable of the spatial regression models in order to reach residuals with constant variance or normality. The response variable is “the number of deaths of infants under one-year-old”, while the independent variables are “the number of women in fertile age”, “the number of women in gestational risk age”, “the number of illiterate women”, “the monthly income of the woman and the men”, “the number of residences with more than six inhabitants” and “the demographic density”. All these variables are available for each census sector of a Brazilian city. The spatial dependence of the number of deaths of infants under one-year-old has been assessed through the global and local Moran indexes. Furthermore, three models have been fitted, namely, the classic regression model, the spatial autoregressive model (SAR) and the conditional autoregressive model (CAR). The Akaike information criterion (AIC) has indicated SAR model as best goodness of fit. The variables “number of women in fertile age” and “monthly income of the women” have been shown to be statistically significant to predict the number of deaths of infants under one-year-old inside the census sectors.