Background/Purpose:   Treatment guidelines recommend using disease activity measures (DAMs) to guide RA therapy, but DAM use in real-world treatment decisions is not defined.  We compared frequency of major therapeutic change (MTC) across disease activity categories for 3 DAMs (Disease Activity Score with 28 joints [DAS28], Clinical Disease Activity Index [CDAI], Routine Assessment of Patient Index Data 3 [RAPID3]); explored ability of DAMs to discriminate between patients with/without MTC; and identified DAM thresholds to optimize discriminant ability.

Methods:   US Veterans enrolled in Veterans Administration RA registry (VARA) had: visit with complete set of DAMs (index date); 2 additional visits ≥60 days apart during 18-month pre-index period; clinical data available 18 months before through 30 days after index date were eligible.  MTC was assessed 1 week before through 30 days after index, defined as: initiation of synthetic or biologic disease-modifying antirheumatic drug (sDMARD, bDMARD); DMARD dose escalation ≥25%; prednisone use (new agent or after 90-day gap); prednisone dose increased 25%; and/or corticosteroid injection in ≥2 joints.  Sensitivity and specificity for disease category to predict MTC was calculated for each DAM.  Youden’s Index (sensitivity + specificity – 1) was used to determine thresholds at which providers were likely to initiate MTC.  Receiver operating characteristic (ROC) curves were used to assess discrimination thresholds.

Results:   For 1,776 patients (12,094 visits), 89% were male, mean RA disease duration was 13.4 years (95% confidence interval [CI]: 12.8, 13.9), and mean age of 63.4 years (95% CI 62.9, 63.9).  Positive associations between disease activity level and MTC proportion were observed (Table 1).  Most MTCs occurred with sDMARDs (Table 2).  Empiric thresholds (highest Youden Indices) (95% CI) were DAS28: 4.02 (3.70, 4.36); CDAI: 12.9 (10.6, 15.1); RAPID3: 3.81 (3.32, 4.30) (Figure).

Conclusion:   Decision points for MTC per Youden Index thresholds were higher than moderate disease activity definitions, suggesting disease activity stratification is not aligned with physician practice.  MTC based on DAS28 and CDAI were driven by changes in sDMARDs. Physician-driven DAMs (DAS28, CDAI) had better discriminant abilities than patient-driven RAPID3.