Multiplicative and additive interactions between risk factors for coronary heart disease

There are a series of well-established risk factors of coronary heart disease (CHD): hypertension, high total cholesterol, smoking, diabetes, older age and male sex [1]. Some studies have paid attention to interactions between them, but have mainly looked at multiplicative interactions with age and/or sex. For example, relative risks associated with many risk factors are larger at younger compared to older ages [2]. The dominant approach to quantifying the association of risk factors with disease is the use of multiplicative models, such as Cox regression. They allow estimation of the association between risk factor and disease as a ratio in hazard between exposed and unexposed groups as well as estimation of the multiplicative interactions between risk factors. An alternative approach is to fit additive hazards model that provides the excess risk due to the presence of risk factor and opportunity to quantify interactions on additive scale [3]. The examination of interactions on the additive scale is rarely done, despite calls for the wider use of absolute measures in epidemiology and public health practice [4], [5].