Is age a cause?

 This is Joseph.

Inspired by this twitter thread. 

This question is amazingly complicated when you look at it closely. Age is an odd cause for discussion in public health because chronological age is not something you can directly intervene on. When you look closely, age is actually a bundle of concepts:

• There are direct causal age effects due to social rules based on age. Criminal justice (e.g. minor versus adult sentencing), age limits on movies or alcohol purchases, eligibility for school or daycare, medicare eligibility, and other such exposures
• Age is a proxy or marker for biological aging processes (e.g., puberty, baldness, biological entropy)
• Age is also a convenient way to adjust for cohort effects (e.g., exposure to lead gasoline, educational changes)

So it is super clear that age is a cause of the first set of exposures, in that it changes the frequency or level of some consequent outcome (e.g., length of jail sentence, access to alcohol). 

It is also pretty clear that the second definition of age is important to de-confounding observational associations. For example, without adjusting for age (or sex), would baldness be associated with death? It would, but not in any causal way (as older males have both less hair and a higher risk of death). The idea that age is not a common cause of the exposure and outcome seems odd, even if the measure (chronological age) is imperfect. But I don't know of a generally better measure of biological entropy, and certainly not one that is so widely available. 

The last (cohort effects), it seems to me, is a case where a lot of the cohort effects would be best directly measured (if feasible). Unlike biological aging, there is a huge risk that there is effect measure modification by factors like geography. It is quite clear that exposures to leaded gasoline may vary based on car ownership of an area and the proximity to roads. 

So the most surprising thing is how age is actually a vague concept when discussing causality.