Interaction, synergy and antagonism in prospective epidemiological studies

Abstract

In public health there is a growing appreciation for the advantage of the additive scale to better understand the impacts of factors involved in a health event. It is necessary to always remember that the concept of statistical interaction is scale dependent. In the causal relationship between a response and the presence of two or more factors, the concepts interaction, synergy and antagonism are the key ideas. The aim of this note is to show an application of the concepts interaction, synergy and antagonism in prospective studies from a public health perspective. We present three scenarios that illustrate analyses of interaction, independence, synergy and antagonism. Stata 12 software was used for fitting models (log-binomial model and Poisson) and estimating parameters. Appendixes are provided with concepts and Stata commands used in the processes of simulation and parameter estimation.