THE AUTHORS REPLY.

Abstract

We appreciate the letter by Ashley-Martin et al. (1) on our framework for descriptive epidemiology (2). They make an excellent point that when it comes to descriptive epidemiology, we should not restrict ourselves to describing the distribution of health states but should also consider the distribution of exposures, risk factors, or other health threats known to be associated with deleterious health states. We appreciate their noting that our framework could just as easily be applied to study the distribution of these health threats as to study the distribution of health states. The idea that health threats are not randomly distributed throughout the population has been previously advanced by many other scholars under the headings “fundamental causes” (3) or “environmental justice” (4), among other things. In these cases, the focus on the health threat or exposure represents a move further back “up” the chain of causation, and we might want to describe who is most likely to be exposed in terms of person, place, and time. To frame this another way, one person’s exposure is another person’s outcome. The goal of such a description of the distribution of health threats may be to draw attention to their inequitable distribution or to target an intervention to mitigate the health threats (shall we call this primary prevention, or is it pre–primary prevention?). If we are interested in an intervention to shift the distribution of that exposure, the “target trial” framework (5) referenced by Ashley-Martin et al. (1) could again become relevant, in that the descriptive work could serve as the basis for identifying a cause of the exposure distribution in a population. However, we reemphasize that descriptive epidemiology does not need to be an input to any future causal inference work to have value.