Defining Difficult-to-Treat Rheumatoid Arthritis in Routine Care: Comparative Performance of Published Criteria and Description of Early and Late D2T-RA Phenotypes

Abstract

Objectives Despite advances in treat-to-target strategies, a subset of rheumatoid arthritis (RA) patients remains refractory to therapy, described as difficult-to-treat RA (D2T-RA). The European Alliance of Associations for Rheumatology (EULAR) has proposed formal criteria for D2T RA. Yet, their application in real-world settings varies, leading to inconsistent prevalence and limited comparability across studies.[1,2] Moreover, the timing of refractoriness—whether patients become D2T early or later in their course—remains largely unexplored. This study compared several published D2T-RA definitions within a real-world cohort and introduced a temporal framework distinguishing early and late D2T-RA. Methods Using data from the Ontario Best Practices Research Initiative (OBRI), a real-world RA registry, we evaluated 4 adapted versions of the EULAR D2T-RA definition among patients who initiated their first advanced therapy after enrollment. These adaptations differed by the timeframe allowed for treatment failure and the disease activity measures applied. We compared their prevalence, overlap, and agreement, and explored temporal anchors to identify balanced thresholds for distinguishing early vs late D2T-RA. Results Among 1121 eligible patients, 215 (19%) met D2T-RA criteria according to ≥1 definition (Table). Their mean (±SD) age was 54.8 ± 11.5 years, and most were female (85.1%). The mean disease duration was 7.7 ± 8.6 years. The prevalence of D2T-RA ranged from 5.5% (definition 4) to 12.7% (definition 2). Definitions 1 and 2, which incorporated multiple disease activity indicators, identified the largest proportion of patients, whereas definition 4—based solely on initiation of a third b/tsDMARD without timeframe restriction—was most restrictive. Agreement between definitions ranged widely (κ = −0.67 to 0.87), with the highest concordance between definitions 1 and 2 (κ = 0.87). Temporal analyses across multiple anchors showed that defining D2T onset from the initiation of the first advanced therapy yielded the most balanced early-to-late distribution, approximately 45% vs 55%—when applying a 2-year threshold, compared with more skewed ratios (>70% late) observed with other temporal anchors. Table. Components of Adapted Difficult-to-Treat RA Definitions and Patient Distribution Across Definitions and Subcomponents N=1121, total D2T RA 215 . Abbreviations: b/tsDMARD, biologic or targeted synthetic disease-modifying antirheumatic dryg;. MoA, mechanism of action; CDAI, Clinical Disease Activity Index; DAS28-ESR/CRP, Disease Activity Score in 28 joints using erythrocyte sedimentation rale or C-reactive protein; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein; HAQ, Health Assessment Questionnaire; MD, physician; PT, patient; AT, advanced therapy. Conclusion Our findings highlight substantial heterogeneity across adapted D2T-RA definitions, indicating they are not interchangeable. More inclusive definitions identify broader groups of patients with active disease and incorporate clinical and patient-reported measures, whereas more restrictive definitions—focused mainly on the number of advanced therapies—may underestimate disease impact. A 2-year threshold from the first advanced therapy offers a balanced distinction between early and late D2T-RA. These findings emphasize the need for harmonized multidimensional definitions and temporal stratification to improve comparability and guide clinical decision-making. References [1.] Nagy G. Ann Rheum Dis 2021;80:31-5. [2.] Hofman ZLM. Rheumatology 2025;64:65-73.