Indeed, a problem with observational studies (although well-done studies adjust for smoking status). Thank you for reminding us of the perils of observational studies and confounding factors.

Michael