How to include the characteristics of the distritos of the Municipality of São Paulo in epidemiologic studies?: an income inequality analysis using the propensity score matching approach

Abstract

OBJECTIVES: The spatial pattern of income distribution in the Municipality of São Paulo, considered to be of a "radial" type, has been challenged by recent studies due to the complex distribution of social and demographic characteristics between its distritos. This demands an in-depth analysis that takes into consideration a multitude of variables in order to control for local heterogeneity. This study aims to identify the distritos of São Paulo that can be defined as "comparable" to another one, by using a statistical methodology known as propensity score matching. METHODOLOGY: The 96 distritos of the Municipality of São Paulo were analyzed separately. 16 variables were included in the model, and the Gini coefficient was used to define "exposure" (high inequality) and "non-exposure" (low inequality). Of the distritos, 27 were considered "comparable". RESULTS: Of the 16 variables inserted in the model, nine presented a statistically significant difference between included and excluded distritos, which is expected by this methodology. Of the 17 pairs of distritos considered to be comparable, only three were composed of distritos situated in the same administrative region, and only one was composed of bordering distritos. CONCLUSION: The complex spatial distribution of the propensity score in the Municipality of São Paulo indicates that it is very difficult to divide the city according to its geographical regions. In order to understand how the distritos of São Paulo affect the health of its residents, it is important to take into consideration its many particularities and how they are spatially distributed.