Epidemiology: a science for justice in health†

The relationships between social environment and health, a time honoured topic of epidemiological research, have been the object of renewed interest particularly in the last 10–15 years. The scientific importance of socially relevant variables, namely those characterizing either a person as an agent in society (e.g. roles, such as sexual or professional, education, income) or a social institution (e.g. type of family, a health delivery service), emerges in several ways. First they are omnipresent as potential, and often actual, confounding factors to be measured and controlled when investigating the associations of other exposures with health outcomes. For instance the association, possibly causal, between levels of blood vitamin C and reduced mortality from several causes has been recently challenged on the grounds of uncontrolled confounding by socioeconomic factors. In general these factors tend to be underrated, namely left unmeasured or inadequately measured, not least because of the received (non)wisdom that for variables of interest only as potential confounders crude and cheap procedures of measurement may be all that is needed. Second whatever effect they may have on health outcomes may be partly direct at the level of each individual and partly contextual at the level of the aggregate to which he/she belongs, an aspect not captured by common methods of single-level statistical analyses and which may pose delicate issues of modelling in multi-level analyses. Third socioeconomic conditions accompany each individual from birth and can exercise shortand long-term effects: ideally they would need to be measured at different points of the whole life course.