Portal de Periódicos da CAPES

Determining causation in epidemiology

1999; Wiley; Volume: 27; Issue: 3 Linguagem: Inglês

10.1111/j.1600-0528.1999.tb02006.x

1600-0528

Flemming Scheutz, Sven Poulsen,

Advanced Causal Inference Techniques

Epidemiology uses many methods to identify the causes of disease, although it remains impossible to provide proof that any specific factor is a cause. We are only able to present supporting evidence. We subscribe to the pragmatic view that a factor is indeed a cause if its elimination improves health. We describe a deterministic causal model, where disease develops when the necessary component causes exist in one of several possible constellations of causes, each of which constitutes a sufficient set of component causes. If we accept this concept of disease causation, the notion of multifactorial disease becomes meaningless. We should rather turn our attention to component causes that we can eliminate in order to improve public health. The complex nature of diseases and the fact that it is only possible to present supporting evidence for a causal relationship are some of our reasons for basing the identification of causes on a theoretic model or framework. The purpose is to construct and present a model that is complex enough to formalize basic intuitions concerning cause and effect. Finally, conceptual frameworks provide guidance for the use of multivariable statistical techniques and may assist in the interpretation of the results.