What Is a Cause?

Keren Xu

2018/09/06

What is the definition of cause in the context of epidemiology? Is it correct to say that smoking causes lung cancer? These questions started to bother me again when I started my phd courseworks in epidemiology. I revisited my notes of a epidemiology course taught by Dr. Sharon Schwartz, and found some interesting examples there:


You have placed a teakettle filled with water on the stove. Several minutes pass and the phone rings. Immediately after, the teakettle whistles.

You are sitting in the living room with your dog. A stranger approaches the house. The dog starts barking.

If the phone had not rung, would the teakettle have whistled?

If the stranger had not approached the door, would the dog have barked?


Definition of a cause: X and Y occurred and Y would not have occurred had X not occurred.
Exceptions? Are these circumstances where we would say that X caused Y but we would not say X and Y occurred and Y would not have occurred had X not occurred? Are there circumstances where we would say that a stranger approached the house (X) and caused the dog to bark (Y) but we would not say a stranger approached the house (X) and the dog barked and the dog would not have barked (Y would not have occurred) had a stranger not approached the house (had X not occurred)?
Definition of a cause: X and Y occurred and in the circumstances, Y would not have occurred if X had not occurred.
Exceptions? Are there circumstances where we would say that X and Y occurred and in the circumstances Y would not have occurred had X not occurred but we would not say that X caused Y?


Jone was born and John died. If John had not been born he would not have died. Therefore John’s birth is a cause of his death.


Definition of a cause: X and Y occurred and, within a causal field, in the circumstances, Y would not have occurred if X had not occurred.
Exceptions? Are there circumstances where we would say that X caused Y but we would not say X and Y occurred and, within a causal field, in the circumstances, Y would not have occurred if X had not occurred.


A man sets out on a trip across the desert. He has two enemies. One of them puts a deadly poison in his reserve can of drinking water. The other (not knowing this) makes a hole in the bottom of the can. The poison water all leaks out before the traveler needs to resort to this reserve can; the traveler dies or thirst. What caused his death?


Definition of a cause: X and Y occurred and Y would not have occurred had X not occurred. X and Y occurred and, in the circumstances, Y would not have occurred if X had not. X and Y occurred and [within a causal field], in the circumstances, Y would not have occurred if X had not. X and Y occurred and [within a causal field], in the circumstances, Y would not have occurred if X had not at least not when and how it did.


Let’s see how the cause was defined in those Epidemiology textbooks 📚:

Mackie’s Counterfactual Definition of a Cause: X and Y occurred and [within a causal field], in the circumstances, Y would not have occurred if X had not at least not when and how it did. - From Mackie’s cement of the universe 1966 http://www.oxfordscholarship.com/view/10.1093/0198246420.001.0001/acprof-9780198246428

Rothman’s Definition:- From Rothman Modern Epidemiology 1998

Mackie’s INUS causes: insufficient but necessary components of unnecessary but sufficient causes


Thus, it is not correct to simply say that smoking causes lung cancer. It is not true that lung cancer would not have occurred if smoking was not the exposure, since there are other causal pies for lung cancer. In terms of lung cancer, smoking is neither a sufficient cause nor a necessary cause. Smoking may be a component of a sufficient cause that leads to lung cacner. However, smoking itself is not a sufficient cause for lung cancer. Neither is smoking a necessary cause of lung cancer, because lung cancer may occur in never-smokers.