The contribution of health behaviours in explaining the social gradient in mortality by occupational position, education, and income in the British Whitehall II and the French GAZEL study.

In Whitehall II, education categorized as university, secondary, and primary education was collected at phase 5 (1997–1999) and was available on 6,776 participants. The remaining participants, n = 2,978, were imputed using multiple imputation. Income was not available at study baseline (1985–1988). We thus use a proxy measure composed of measures of car ownership and type of accommodation. The highest category represents participants owning a car and their house, the lowest represents participants not owning a car and living in rented accommodation. The intermediate category represents other combinations of car ownership and type of accommodation. In the GAZEL study education and income were collected at study baseline (1989). Education was categorized as university, secondary, and primary education. For income, the following three categories (based on quintiles of income, converted in Euro from French Francs) were used in the analysis: aHR for lowest versus highest occupational position adjusted for age at baseline and sex.bPercent attenuation in log HR = 100× (βModel 1 − βModel 1+ health behaviour(s))/(βModel 1), where β = log(HR).