Effect of different approaches to treatment of smoking as a potential confounder in a case-control study on occupational exposures