Background/Purpose: Two versions of DAS28 are available, DAS28-4 comprising 4 variables [tender and swollen joint counts, acute phase reactant (APR), and patient global assessment] and DAS28-3 where patient global has been omitted. Despite the difference between DAS28-4 and DAS28-3 thresholds for remission and low disease activity (LDA) are the same. Additionally, the APR used to calculate the DAS may be either ESR or CRP. The aim of this analysis is to describe the agreement between these four possible indices, DAS28-4 ESR, DAS28-4 CRP,  DAS28-3 ESR and DAS28-3 CRP and to compare them in terms of classifying remission and LDA in a real-world, routine clinical care setting.

Methods: BioTRAC is an ongoing, prospective registry of patients initiating treatment for RA, AS, or PsA with infliximab or golimumab. In this analysis, data from RA patients who were treated with infliximab between 2002-2014 or with golimumab between 2010-2014 and had available information in all indices were used. The definitions for remission were as follows: DAS28-3/4 <2.6; LDA was defined as: DAS28-3/4 <3.2. Correlation between the different indices was assessed with the Pearson's correlation coefficient (r) and classification agreement was assessed with Cronbach's alpha (CA) and the kappa statistic.

Results: Eight hundred sixty nine RA patients who had 3,517 complete assessments were included in the analysis. Non-remission was classified by all indices in 61.4% of cases, while remission was achieved in one (5.9%), two (10.3%), three (5.3%), or all four (17.2%) indices. Similarly, non-LDA was classified by all indices in 46.1% of cases, while LDA was achieved in one (6.2%), two (10.9%), three (5.4%), or all four (31.3%) indices.

Overall, a strong linear positive correlation (r>0.8) was observed between all indices. When looking at the internal consistency in terms of classifying disease state, the CA was 0.905 for remission and 0.923 for LDA suggesting an overall high internal consistency. However, when looking at the individual inter-item correlations (Table 1), the agreement between indices was variable with DAS28-3 CRP and DAS28-4 CRP showing the highest correlation and DAS28-3 ESR and DAS28-4 CRP showing the lowest correlation. When comparing DAS28-4 ESR with DAS28-3 ESR, the latter categorized 16.5% of DAS28-4 ESR remission cases as non-remission and 3.0% of DAS28-4 ESR non-remission cases as remission. With respect to LDA, DAS28-3 ESR categorized 9.1% of DAS28-4 ESR LDA cases as non-LDA and 4.7% of DAS28-4 ESR non-LDA cases as LDA. Similar results were observed with DAS28 CRP.   

Conclusion: The results of this analysis show that, despite being highly correlated, variability exists in the classification of remission and LDA by the various DAS indices. These results suggest that decision making based on disease state achieved may vary significantly based on the type of APR used in the DAS index.