In my opinion, the greatest recent innovation in the field of epidemiology has been the use of advanced statistical techniques like proportional hazards regression to analyse epidemiologic data. This innovation has made possible the rapid and efficient analysis of large complex databases involving many independent variables and numerous chronic disease outcomes. Strong, etiologically plausible relationships have been detected. Weak, etiologically implausible relationships have also been detected. Some of these weak relationships have been deemed to be ‘causal’ simply because they are ‘significant’ according to a statistical test. Unfortunately, these weak relationships do not satisfy the established epidemiologic criteria for causality and yet they have been used to implement major public health regulations.
Perspective must be given to that fact that the strongest and most valid relationships in epidemiology typically involve infectious diseases, where there is a clear causal agent related to a specific disease. For instance, malaria parasites transmitted by female Anopheles mosquitoes cause malaria and the Vibrio cholerae bacterium transmitted by contaminated water causes cholera.
Modern epidemiology needs to use the innovation of advanced statistical techniques to identify and examine relationships that are strong and plausible, not weak and implausible. Otherwise, it risks damaging its credibility as a scientific discipline.