The scientific assessment of combined effects of risk factors: different approaches in experimental biosciences and epidemiology

The analysis of combined effects of substances or risk factors has been a subject to science for more than a century. With different goals, combined effect analysis was addressed in almost all experimental biosciences. The major theoretical foundation can be traced back to two distinct origins. First, to the work by the pharmacologist Loewe on the concept of concentration additivity and second to the biometrician Bliss and the concept of independent action. In the search for a general solution and a unified terminology the interrelations of the concepts have extensively been studied and experimental findings reviewed. Meanwhile there seems to be consensus in experimental sciences that each concept has its role in predicting combined effect of agents and both are used for hazard und risk management. In contrast, epidemiologists describe combined effects mainly in terms of interactions in regression models. Although this approach started from a probabilistic model equivalent to the concept of independent action this origin is rarely acknowledged and effect summation is usually the preferred concept nowadays. Obscure biological meaning, the scale dependency of interaction terms as well as unavoidable residual confounding are taken as reasons why no new insights in combined effect analysis are likely to occur from epidemiology. In this paper we sketch the history of ideas and the state of the arts in combined effect analysis. We point to differences and common grounds in experimental biosciences and epidemiology.