Objective: To identify baseline variables that predict remission at 1 year in patients with recent onset inflammatory polyarthritis (IP).
Methods: We prospectively studied 167 patients aged >or=16 years with a 4-week to 12-month history of swelling of >or=2 joints. At baseline, no patient had previously received corticosteroids or disease-modifying anti-rheumatic drugs (DMARDs). To adjust for differences in baseline variables associated with the type of treatment given (a surrogate marker of disease severity), we used regression analysis. The classification probability of treatment thus obtained was entered, along with other significant baseline variables, in a second separate regression analysis to identify variables that predicted remission (no swollen joints).
Results: Frequency of remission was 50.9% at 1 year. In the first regression analysis, variables associated with treatment with DMARDs or DMARDs and corticosteroids versus corticosteroids alone included age, morning stiffness, swollen joint count (SJC), disease severity according to the patient, and rheumatoid factor (RF) level; the strongest association was for higher SJC. In the second regression analysis, the model that best predicted remission (correct in 70.1% of cases) included age, tender joint count (TJC), erythrocyte sedimentation rate (ESR), RF, total Sharp score, disease severity according to the physician, and the 1987 American Rheumatism Association (ARA) criteria for rheumatoid arthritis (RA); the strongest association was for failure to meet these criteria. The model's sensitivity, specificity, and area under the receiver operating characteristic (ROC) curve were 70.6%, 70.9%, and 75.4%, respectively.
Conclusion: Although we identified some predictors of remission, no model accurately predicted remission at 1 year in this cohort.