Mortalidade adulta por níveis de escolaridade no estado e município de São Paulo: uma proposta de estimação a partir do Censo Demográfico de 2010

Abstract

The aim of the thesis was to investigate differentials in adult mortality based on educational levels from the demographic sample data of 2010 Census. This is a draft estimate from an alternative database to the main source of mortality studies in Brazil, a Mortality Information System (SIM) of the Ministry of Health. Among the highlighted advantages of using the Census, are: the availability of data (micro data) on both the living and the dead within a reference period and variety of information on cohabiting households and surviving individuals with those who suffered death. On the other hand, the only personal information on the dead are age and sex. Given this limitation, a secondary proposal of the thesis was to present an allocation methodology on educational levels of individuals who have suffered death by using multiple imputation ordered logistics regression and imputation by normal multivariate model (MVN) using Expectation-Maximization (EM) algorithm and simulations. To achieve this, a sample was selected from the State of Sao Paulo to develop and validate the methodology so as to make it applicable in Brazil in the future. The MVN model was more appropriate in the consistency and validation analysis of the methods; this phase depended on mortality data from SIM to be conducted. After obtaining information on the educational levels of the dead, rates of mortality by age, sex and education were estimated using coefficients obtained by Poisson regression. The results show significant educational gradient in mortality, especially for men and the city of São Paulo as compared to the state. An important point was the differential behavior in mortality by educational level with age. From the literature review, one of the hypotheses was a decreasing differential with age. However, the difference was almost constant by age. To further analysis, it is necessary to understand how transition of mortality has occurred in the state and how periodic measures may influence the results. It is important to understand the relationships between period, age and cohort to analyze the differential behaviors of educational mortality. Furthermore, the increased access to education which has elevated educational level of younger cohorts in Brazil should be taken into account. It should however be pointed out in these discussions, that they were not privileged in the thesis. Regarding the use of census to study differentials in mortality, results give an optimistic perspective. Meanwhile, there is an expectation to improve the coverage and quality of information on the educational levels of SIM. Census can be an important source for the studies of socioeconomic and mortality differentials offering other features that can complement the analysis of the effect of education.