Abstract of the chapter


Causality: old ideas and new challenges

Juan José Granizo


Formally causality is the part of epidemiology that explores the etiological relationships between an exposure and a health effect. This definition has the practical difficulty of measuring the actual exposure to a risk factor and the effect that is being investigated. Epidemiology has used several theoretical models to define a cause. The first scientific model is the one applicable to infectious diseases (Koch's model), although it is outdated, as well as being useless for chronic diseases. Multicausal models apply new ideas such as disjunctive plurality (an effect can have multiple causes), conjunctive plurality (the causes must be linked to cause an effect) and the multiplicity of effects of a single cause. To overcome the limitations of both models, a "modified determinist" (Rothman) model has been proposed that saves the theoretical foundations of determinism by admitting the fact that the disease appears according to random patterns. The Rothman model details sufficient and necessary causes, defines risk as that condition that increases the likelihood of an effect occurring, and includes ideas with great impact such as interaction and confounding factors. Inferring is a subjective process, so it is convenient to be sceptical with any claim of causality. In contrast to the Rothman model, a probabilistic model has been proposed, based on the random appearance of an effect, which may be useful for large populations or when the risks or effects are not well outlined. One step further we have the "models of complex causes" that assume that the causes are related to each other, are dependent on each other and with relationships that change over time, and in turn are modified by the effects. These models are in the line of analysis of chaotic phenomena such as climatology, for which they rely on the ideas of Lorenz and the butterfly effect (small changes in the conditions of the initial model cause great changes in the final effect), the models of fractals, network theory and catastrophe theory. They are very useful for explaining infectious outbreaks or safety incidents that may be caused by the accumulation of small failures that cause considerable effects. Among the challenges of causality is demonstrating that an exhibition is not a real risk (demonstrating the innocence of an exhibition) what constitutes a social demand not answered by science. The tools that allow us to identify the protection factors that block the effect of known risk factors are pending. A third challenge would be that the qualitative methodology will surpass the quantitative one, not for rigor, but for simplicity since in the future complex mathematical techniques will be used, which will be beyond the reach and understanding of most clinical researchers. And a fourth challenge will be to combine clinical data, lifestyles, environmental factors and genetics using big data, something that may change our view of epidemiology.