ABSTRACT
Confounding by indication, or the lack of separation of the effect of a drug from the effect of the underlying condition for which the drug is prescribed, can confuse the interpretation of pharmacoepidemiologic studies. This issue has arisen when examining the effect of certain second-line drugs which are used in the treatment of severe rheumatoid arthritis. A number of large prospective epidemiologic studies have compared the risk of neoplasia in the general population with the risk in subjects with rheumatoid arthritis (RA) in an effort to determine if RA is related to elevated risk of neoplasia. The basic rationale of the study involves the comparison of cancer incidence between a cohort of rheumatoid arthritis patients and two sets of controls: general controls and osteoarthritis controls. In order to rule out people who were in the hospital because of early symptoms of undiagnosed cancer, anyone having a diagnosis of cancer up to one year after enrollment was excluded.
