THE USE OF STRUCTURAL EQUATIONS MODELING IN THE ANALYSIS OF THE INFLUENCE OF SKIN COLOR/RACE AND SOCIOECONOMIC STATUS ON THE HEALTH OF BRAZILIAN ELDERLIES

Abstract

Introduction: The use of the variable skin color/race in quantitative studies has been increasing. However, these studies are subject to criticism due to the difficulty in distinguishing the actual effects on color/race induced by other variables. The application of Structural Equations Modeling (SEM) in the health field can be useful for the analysis of correct estimates of the impact of this variable, as well as to improve the quality of the adjustment for confounding covariates. Objective: To report an example of SEM in the analysis of direct and indirect effects of color/race and Socioeconomic Status (SS) on the Health of Brazilian elderlies. Material and Methods: This was a cross-sectional study including 28,769 elderlies aged ≥60 years surveyed in the PNAD (National Survey by Household Sampling), 2008. The variable health is a construct which was measured by the variables self-assessment of health condition, functional disability and the number of chronic conditions. The exploratory variables were: self-reported color/race and the SS. This construct was measured by the variables: educational level, household income per capita, and number of people per household. Results: The direct effect of color/race on health was found to be small and negative. The total (direct and indirect) effect was positive and significant, but smaller than the indirect one mediated by SS, which was the greater effect of color/race on health. The largest effect observed on health was the SS. This effect was found to be negative and moderate. Conclusion: The skin color/race partly explains the health of older people, but this effect is lower than that caused by the SS. The SEM allowed to bypass some methodological problems related to the use of the variable color/race in health. DESCRIPTORES Ethnicity and Health. Aged. Statistical Analysis. Health Inequalities.