View article

There appears to be an emerging weight of evidence that the education-mortality association is most substantially accounted for by material rather than behavioural mechanisms. This study aims to determine whether this pattern will be observed in the Canadian population. Six-year survival data from the National Population Health Survey were analysed using Cox proportional hazards regression. To a base model of education, age and sex were added behavioural factors (smoking, drinking, physical activity, body mass index, and blood pressure), and material factors (household income adequacy, ownership and crowding, and unemployment). P-spline smooths were used to avoid linearity assumptions for continuous covariates. Among Canadians aged 25 and older, adjustment for individual variables in both the material pathway and the behavioural pahtway had generally no …