Objective: The 1987 American Rheumatism Association (ARA) criteria for rheumatoid arthritis (RA) were developed to discriminate between patients with established RA and those with another rheumatological disorder. Their ability to determine which patients presenting with early synovitis have "true" RA is not known. We evaluated whether the 1987 ARA classification criteria for RA in patients newly presenting with inflammatory polyarthritis (IP) predict persistent, disabling, or erosive arthritis.
Methods: We studied 486 patients with early IP referred to the Norfolk Arthritis Register. The 1987 ARA criteria were applied at baseline, and assessed for their ability to identify (1) patients referred to hospital for whom the diagnosis of RA was recorded by the hospital physician; (2) patients at 3 years with (a) persistent synovitis; (b) moderate or greater disability; and (c) erosions.
Results: At baseline, 323 (67%) patients satisfied the ARA criteria in the classification tree format. Exactly 50% of those referred to hospital were given a diagnosis of RA. By 3 years, 76% of the 486 patients had persistent disease, 36% had a Health Assessment Questionnaire score > or = 1, and 40% had erosions. The sensitivity of the criteria was good, ranging from 77 to 87% depending on the outcome. The specificities were poor, and thus the overall discriminatory ability showed little improvement over random probability.
Conclusion: . Among patients newly presenting with IP, the 1987 ARA criteria for RA had a low ability to discriminate between patients who developed persistent, disabling, or erosive disease and those who did not. Alternative criteria are required for studies investigating early RA.