Estimates of the relative risk of mortality based on the ontario longitudinal study of aging

Data from the Ontario Longitudinal Study of Aging were analyzed to examine the associations of the independent variables income, income change, education, smoking and perceived health with the dependent variable mortality during a ten year follow-up beginning in 1969. The analyses investigate the associations of the independent variables with deaths, with other causes of attrition and with all causes of attrition. The results indicate that smoking is the strongest predictor of mortality, and income is the strongest socioeconomic predictor. The analyses also show that perceived health measured prior to the mortality follow-up masks the association between the independent variables and mortality. Since the exclusion of the perceived health variable did not appreciably reduce the fit of the models, it was omitted from further analyses. The distributions of mortality for the various independent variables differed appreciably between models using deaths and all causes, but the bivariate and multivariate associations between variables were relatively unaffected by the alternative methods of operationalizing the dependent variable.