Survival analysis and iniquities in older Brazilians: a six year follow up survey in São Paulo, Brazil

Methods Data comes from a longitudinal survey – SABE Study (Health, Well Being and Aging) that began in 2000 with a sample of population aged 60+ living in São Paulo/Brazil (n=2,143 from a multi stage clustered sampling). A procedure with probability proportional to the size was carried out using census tracts with replacement. To achieve the desired oversampling for respondents aged 75+, additional households close to the selected census tracts were sampled. The second wave was done in 2006 when 1115 elders were re-interviewed. Descriptive statistics included tests for association using Rao Scott procedure with correction for sample-design. Multi variable analysis was done by adjusting Cox regressions with robust estimation, stratified by age and sex. Kaplan-Meier Survival Analysis was used to compare survival curves by social demographic conditions (sex, schooling, income, early conditions) and health (depression, comorbidities, disability, self-perceived health and falls).