Objective: To compare the diagnostic properties of the Rome 1961 criteria for active rheumatoid arthritis (RA) and the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 criteria for RA with regard to their ability to classify, diagnose, and predict outcome in RA.
Methods: Analysis of cross-sectional and longitudinal data from repeated health examinations and review of clinical records of 3,509 Pima Indians followed up from January 1966 to December 1990.
Results: The ACR 1987 criteria identified approximately 50% of the cases identified by the Rome 1961 criteria, in both cross-sectional and longitudinal analyses. The ACR 1987 criteria were better predictors of subsequent development of a clinically supported diagnosis and treatment with slow-acting antirheumatic drugs (both P < 0.001), but were less sensitive than the Rome 1961 criteria for detecting cases for which there already was a clinically supported diagnosis (P < 0.001).
Conclusion: In a population-based analysis, the ACR 1987 criteria are less sensitive for detecting clinical disease, but predict a clinically more severe prognosis, compared with the Rome 1961 criteria. The sensitivity of both sets of criteria to identify clinical disease is improved if multiple examinations or inactive disease are taken into account.