SUICIDES IN THE BRAZILIAN MUNICIPALITIES: A SPATIAL APPROACH FOR CENSORED DATA

Abstract

Suicide is recognized internationally as a public health problem that ranks
among the ten most frequent causes of death among the general population, surpassing
the number of deaths from armed conflict. In Brazil, according to data from the Mortality Information System, for each registered suicide, it is estimated that there are over twenty attempts not reported. Under-registration, coupled with rarity and the inherent
characteristics of its distribution, makes the phenomenon even more complex, especially in the Brazilian context, where the continental dimensions carry socioeconomic
and cultural heterogeneities. In order to verify the influence of socioeconomic factors
in the presence of spatial interactions, a spatially distributed Tobit model is applied to
municipal data of 2010. The empirical approach adopted represents an important advance in the literature, since, besides considering the spatial autocorrelation a level of
aggregation that hitherto was not common to those surveyed regarding suicide, also
offers a treatment for censored variables. The main results corroborate previous studies
about the significant influence of socioeconomic factors and confirm spatial multiplier
effects, consistent with the Durkheimian postulates about the contagious effect. Therefore, it is desirable that public policies to contain and prevent suicide take into account
not only socioeconomic and demographic aspects, but also the territorial distribution of
occurrences and the spatial interaction between them.