Chinese Registry of Rheumatoid Arthritis: IV. Correlation and Consistency of Rheumatoid Arthritis Disease Activity Indices in China
Rheumatoid arthritis (RA) is a chronic inflammatory joint disease affecting approximately 1% of the global population. Characterized by persistent synovitis, uncontrolled inflammation can lead to joint destruction, deformities, and functional impairment. Accurate assessment of disease activity is critical for guiding treatment and achieving remission. Composite indices such as the Disease Activity Score 28 (DAS28), Simplified Disease Activity Index (SDAI), and Clinical Disease Activity Index (CDAI) are widely used in clinical practice and research. However, discrepancies in disease activity categorization across these indices can lead to inconsistent treatment decisions. This study, leveraging data from the Chinese Registry of Rheumatoid Arthritis (CREDIT), evaluates the correlation and concordance among these indices in a large cohort of Chinese RA patients, providing insights into their interchangeability and applicability in clinical settings.
Study Design and Methodology
The CREDIT database, established in 2016, is a nationwide, multicenter registry encompassing over 500 hospitals across 26 Chinese provinces. This cross-sectional analysis included 30,501 RA patients enrolled between November 2016 and August 2018. Participants met the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) classification criteria for RA and were aged over 16 years. Exclusion criteria included coexisting autoimmune diseases or acute infections.
Disease activity was assessed using four indices:
- DAS28-ESR: Incorporates erythrocyte sedimentation rate (ESR), tender joint count (TJC28), swollen joint count (SJC28), and patient global assessment (PtGA).
- DAS28-CRP: Replaces ESR with C-reactive protein (CRP).
- SDAI: Combines TJC28, SJC28, PtGA, physician global assessment (PhGA), and CRP.
- CDAI: Excludes acute-phase reactants (CRP/ESR) from SDAI.
Statistical analyses included Spearman correlation coefficients for continuous variables, quadratic weighted kappa (κ) for ordinal agreement, and Bland-Altman plots to assess measurement differences between DAS28-ESR and DAS28-CRP. Discordance rates were calculated to quantify classification mismatches. Receiver operating characteristic (ROC) curves determined optimal cutoff values for DAS28-ESR, DAS28-CRP, and CDAI corresponding to SDAI remission (≤3.3).
Key Findings
Patient Characteristics
The cohort comprised 80.46% females, with a mean age of 52.6 ± 13.1 years and median disease duration of 4 years (interquartile range [IQR]: 1–10). Most patients exhibited moderate (MDA) or high disease activity (HDA) across all indices. Rheumatoid factor (RF) positivity was observed in 84.70% of patients. Median ESR and CRP levels were 32 mm/h (IQR: 17–56) and 9.4 mg/L (IQR: 3.0–25.0), respectively. Mean scores for DAS28-ESR and DAS28-CRP were 5.1 ± 1.7 and 4.5 ± 1.6, while median SDAI and CDAI scores were 24.6 (IQR: 14.9–39.7) and 22.5 (IQR: 13.5–36.8).
Correlation Among Indices
All indices demonstrated strong pairwise correlations (Spearman’s ρ > 0.9, P < 0.001). The highest correlation was between SDAI and CDAI (ρ = 0.989), reflecting their structural similarity. DAS28-ESR and DAS28-CRP also correlated closely (ρ = 0.951), as did DAS28-ESR with SDAI (ρ = 0.920) and CDAI (ρ = 0.911).
Concordance and Discordance
Despite high correlations, classification discordances were significant. Weighted κ values ranged from 0.637 (DAS28-CRP vs. CDAI) to 0.895 (SDAI vs. CDAI). Approximately 30% of patients were classified into different disease activity categories (remission, low [LDA], moderate [MDA], or high [HDA]) depending on the index used. Key observations included:
- DAS28-ESR vs. DAS28-CRP: 28.96% discordance, with DAS28-ESR classifying 25.94% of patients into higher activity categories than DAS28-CRP.
- SDAI vs. CDAI: 8.70% discordance, the lowest among pairwise comparisons.
- DAS28-ESR vs. SDAI: 25.70% discordance, with SDAI categorizing 13.40% into higher and 12.30% into lower activity groups.
Bland-Altman analysis revealed a mean difference of 0.54 (95% CI: 0.51–0.58) between DAS28-ESR and DAS28-CRP, with DAS28-ESR yielding higher scores in 83% of cases. Absolute differences exceeded 1.2 in 10.09% of patients, indicating clinically significant discrepancies.
Optimal Cutoffs for Remission
Using SDAI remission (≤3.3) as the reference standard, ROC curves identified optimal cutoff values for other indices:
- DAS28-ESR: 3.06 (AUC: 0.9572, sensitivity: 91.5%, specificity: 91.1%).
- DAS28-CRP: 2.37 (AUC: 0.9768, sensitivity: 95.4%, specificity: 93.0%).
- CDAI: 3.20 (AUC: 0.9967, sensitivity: 98.9%, specificity: 98.3%).
These thresholds diverged from established ACR/EULAR cutoffs, underscoring the need for region-specific adjustments.
Clinical Implications
The study highlights critical limitations in the interchangeability of RA disease activity indices. While high correlations and κ values suggest statistical agreement, substantial discordance in categorical classification (≈30%) poses challenges for clinical decision-making. For example, a patient classified as HDA by DAS28-ESR might be labeled MDA by DAS28-CRP, potentially delaying aggressive treatment. Similarly, SDAI and CDAI, though more concordant, still exhibit discrepancies due to the exclusion of acute-phase reactants in CDAI.
The tendency of DAS28-CRP to underestimate disease activity compared to DAS28-ESR aligns with prior studies, possibly due to CRP’s shorter half-life and rapid response to inflammation. This discrepancy is particularly relevant in populations with elevated ESR levels, such as older patients or those with comorbidities. The derived optimal cutoffs for DAS28-CRP (2.37 vs. the conventional 2.6) and DAS28-ESR (3.06 vs. 2.6) suggest that lower thresholds may better align with SDAI-defined remission in Chinese patients.
Strengths and Limitations
This study’s strengths include its large, nationally representative sample and rigorous methodology. The CREDIT database’s standardized data collection minimized intercenter variability, and the inclusion of real-world patients enhances generalizability. However, the cross-sectional design precluded longitudinal assessment of index performance over time. Additionally, the analysis did not account for ethnic heterogeneity within China or potential confounders like treatment regimens.
Conclusion
This comprehensive evaluation of RA disease activity indices in China underscores their high correlation but significant classification discordance. DAS28-ESR, DAS28-CRP, SDAI, and CDAI are not interchangeable, and their use may lead to divergent therapeutic strategies. The derived optimal cutoffs for DAS28-ESR, DAS28-CRP, and CDAI highlight the need for region-specific adaptations to improve alignment with SDAI remission criteria. Clinicians should prioritize indices that incorporate acute-phase reactants (e.g., SDAI) for conservative disease activity assessment, particularly in populations with elevated inflammatory markers. Future research should explore longitudinal discordance patterns and validate proposed cutoffs in diverse cohorts.
doi.org/10.1097/CM9.0000000000001517
Was this helpful?
0 / 0