Adult obesity in African regions: an analysis by beta regression models

• Autor(a) principal: Lirio, Valentina Wolff
• Data de Publicação: 2024
• Outros Autores: Loose, Laís Helen
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Saúde (Santa Maria)
• Texto Completo: https://periodicos.ufsm.br/revistasaude/article/view/70075
• topic: Modelo de regressão beta; Obesidade; África; Beta regression model; Obesity; Africa

Objective: to model the proportion of obese adults on the African continent in 2016 using covariates. Method: beta regression model, the data set used in the study was collected on the World Health Organization (WHO) website, from 43 countries belonging to the African continent, divided into five regions of Africa. The variables used in the study were the proportion of obese adults (y) as the response variable, and the variables life expectancy at birth (in years) (x1) were used as explanatory variables. alcohol (registered per capita consumption) (x2); the prevalence of insufficient physical activity (x3); Africa regions (x4); mean BMI (x5); overweight among children (5-9 years old) (x6); and estimated prevalence of depression (x7). The analyzes were performed in the R software, using the betareg package. Results: The beta regression model with variable dispersion proved to be adequate. The covariates that influence the proportion of obese adults are life expectancy at birth (in years) (x1), the prevalence of insufficient physical activity (x3), mean BMI (x5), and overweight among children (5-9 years) (x6), for the mean model, and the covariates mean BMI (x5) and estimated prevalence of depression (x7), for the precision model. All covariates were significant at the 10% significance level. All covariates for the mean model, except x1, had a positive effect on the response variable (y), and in the precision model, both x5 and x7 had a negative effect. Final considerations: this study is expected to present an adequate approach for modeling data on the proportion of obese adults, the dissemination of the beta regression model, and the identification of risk factors for obesity.