Which variables best predict change in rheumatoid arthritis therapy in daily clinical practice?

Objective: To determine in clinical practice which clinical status variables for rheumatoid arthritis (RA) are most closely associated with a change in disease modifying antirheumatic drug (DMARD) therapy.

Methods: A prospective monocenter study was conducted in 204 consecutive patients with RA. Rheumatologists recorded patient characteristics, treatments, and disease activity data [tender and swollen joint count (28), morning stiffness, visual analog scale (VAS) for pain (0-100 mm), patient global assessment and physician global assessment, Westergren erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP)]. The rheumatologists decided whether or not to initiate or change treatment but were not informed that their decisions were part of the investigation. Logistic regression analysis was performed to evaluate which study variables best predict change in therapy. ROC analysis was used to obtain the cutoff value of the different composite indices (DAS28(ESR), DAS28(CRP), SDAI) for treatment change, as well as sensitivity and specificity.

Results: The variables that were predictive for a change in treatment were (in descending order): swollen joint count, morning stiffness, CRP, tender joint count, and patient global assessment. Composite index values associated with a decision to modify DMARD therapy were: DAS28(ESR) 4.2 (sensitivity 87%, specificity 70%); DAS28(CRP) 3.6 (sensitivity 86%, specificity 78%); and SDAI 15 (sensitivity 90%, specificity 86%). The discriminative ability of SDAI was better than that of DAS28(CRP) or DAS28(ESR).

Conclusion: In our study, swollen joint count was the variable with the greatest weight, which explains the observed better performance of SDAI.