Epidemiology of cancer in rheumatoid arthritis: methodologic pitfalls.

Studies that try to associate immunoinflammatory disease, typified by rheumatoid arthritis, and malignancy have been limited by several important methodologic difficulties. Usually the hypothesis examined has been one of an increase in incidence of one or more specific neoplasms, and sometimes this hypothesis has been generated after examining the data. Some of the more common methods for assessing risk, including case reports and small series, case-control studies, hospital-based studies, and animal studies, and the competing risk fallacy can create problems of interpretation and can skew the results. The ideal study is one in which a random sample of 1,000 or more patients with rheumatoid arthritis are followed prospectively from disease onset to death and the data compared with total population data over the same period. Ideally, there would be perfect follow-up and autopsy information available on all patients and all control patients. Such studies of course will not be available. However, those studies that are designed best fail to associate rheumatoid arthritis with malignancy, although certain specific malignancies do appear associated with alkylating agent therapy for rheumatoid arthritis.