Packyears:

product of the duration of smoking (in years) and the numbers of cigarettes smoked per day divided by a package size of 20 cigarettes, summarized for different smoking periods

What would justify to assume tobacco smoke as confounder or covariate?

Consider

(1) Tobacco smoke contains only small amounts of arsenic.

(2) Smoking is a habit which may not vary spatially such as air pollution with arsenic.

(3) How likely is a true ‚protective effect‘? A recall bias of the cases cannot be excluded. Whereas controls were mainly interviewed at home, cases were invited by their dermatologist to be interviewed at a dermatological outpatient service. Similar results for alcohol consumption support this potential bias (underreporting among cases).