Risk taking disease.

rence of this newly defined disease, it would be difficult to detect if outcomes were determined only according to the presence or absence of injury events, as opposed to this more refined measure. Further, in the development ofmultivariable etiological models, one guiding principle is never to adjust for variables unless we understand how they relate to both the exposure and the outcome. If one accepts our proposition (that we are not measuring this outcome appropriately), this means that we may not be developing appropriate etiological models. We might be considering variables as risk factors contributing to the occurrence of disease when, in fact, they are actually part of the disease process. Adjustments for these variables could lead to biased estimates of risk being attributed to the underlying risk factors. More importantly, this would lead to inappropriate recommendations for preventive intervention.