Background: The current paradigm for early undifferentiated arthritis suggests that persistent synovitis leads to erosive joint damage, which results in functional disability. Discriminating between different forms of early arthritis outcome is relevant for therapeutic decision-making and prevents the occurrence of circularity in diagnostic studies. Implementation of a prognostic model into clinical practice is highly required. Objective: To identify the prognostic factors at entry for the persistence of early inflammatory arthritis and to develop a scoring system to assess the outcome in patients presenting with early inflammatory arthritis. Methods: A cohort of 173 patients with early inflammatory arthritis, were assessed in a special early arthritis clinic. Case definition of synovitis was assessed clinically by the number of tender joint and swollen joint counts as well as symmetry of the affected joints and data were recorded. Baseline HAQ score, serum rheumatoid factor and anti-CCP titre were monitored. Patients satisfying the American College of Rheumatology classification criteria for RA and the European Spondyloarthropathy Study Group criteria for spondylo-arthropathry were excluded, as well as those with a specific rheumatic diagnosis. All the patients were reviewed every 3 months for 24 months period. Ultrasonography for both MCP and PIP joints, both hands, were performed at day of entry and 6 months after initial assessment. All patients had MRI both hands and wrist joints. The arthritis outcome recorded at 1-year follow up represented the gold standard for the diagnosis. Logistic regression analysis was performed to identify the independent factors (predictors) of persistent arthritis and accordingly a scoring system was invented that involved the predictors revealed from the analysis. A ROC curve was set to display the performance of the scoring system and a cut off point was selected taking into consideration a higher sensitivity than specificity, as this model will be mainly used as a screening tool. Results: During the follow up period, 93 patients showed evidence of self-limiting arthritis while 80 showed persistent arthritis. Duration of morning stiffness in minutes, percentage change in HAQ score after 3 months duration and anti-CCP positivity were the predictors of persistent arthritis. ROC curve analysis identified a cut off point of 121 above which the early arthritis patient would be more at risk of developing persistent arthritis. Conclusion: Application of a model (scoring system) to stratify patients presenting with early persistent inflammatory arthritis from those with self-limiting disease can be attained. The developed model was found to be valid on the studied cohort. This model is important for standard clinical practice as the value of prediction of persistent arthritis has its great impact not only for identifying the patients but also on the therapeutic outcome as well. (C) 2007 Elsevier Masson SAS. All rights reserved.
