Introduction The aim of this study was to examine to which degree the often demonstrated association between educational status and mortality is confounded by family characteristics, that is environment in childhood and genetic setup. To disentangle an independent causal effect of educational status on mortality from selection mechanisms, we used a discordant sibling design making it possible to control for the shared early life environment and to a lesser degree genetic setup per design.

Methods The study was register-based and the study population consisted of all individuals born in Denmark between 1950 and 1979 who had at least one full sibling born in the same period (N=1 381 436). All individuals were followed from age 28 years until the age of death, emigration, or December 2009. Using Cox' regression analyses, we estimated HRs for mortality according to educational status.

Results Both conventional cohort analyses and intra sibling analyses were carried out. Educational differences observed in the cohort analyses were attenuated in the intra- sibling analyses.

Conclusion The attenuation of the association in the intra sibling analyses indicates that environment in childhood and/or genetic setup explain some of the association between educational status and mortality. However, significant associations still persisted in intra-sibling analyses, supporting an independent effect of education on mortality.