Inflation in epidemiology: "The proof and measurement of association between two things" revisited