Background Composite indexes in RA are widely validated in clinical trials and have been adapted to everyday practice for therapeutic judgment, monitoring and determining remission. Our aim was to investigate the accuracy of the main composite scores (DAS28, SDAI, CDAI) in evaluating disease activity and remission in RA, in real life conditions.
Methods Data from a nationwide observational cohort of RA patients, recruited from October 2010 to March 2011, in 11 centers. We investigated agreement between DAS28, SDAI and CDAI in different disease states using κ statistics, and agreement between DAS28, SDAI, CDAI and ACR/EULAR Boolean-based definition for remission. ROC curves were constructed to determine cut-off points for treatment interventions (DMARD changing or dosage increasing) and assess discriminative ability for every index. All statistic computation was done using SPSS ver 17.0.
Results The mean age of the 249 patients was 50.1±14.5 years, mean disease duration 8.4±7.8 years, 85.5% were female. At inclusion, the mean DAS28 was 3.9±4.8, mean SDAI 22.5±21,7 and mean CDAI 11.1±10,2. Agreement between various activity states was good between DAS28, modified DAS28 and SDAI cut-offs (κ =0.41 to 0.67), but moderate with CDAI (κ =0.20 to 0.37) (see table). With the ACR/EULAR remission criteria set as a gold standard, the best performance was with the SDAI and modified DAS28. The optimal cut-off for treatment change was 4.25 for DAS28 (sensitivity 73.4%; specificity 69.8%), 16 for SDAI (sensitivity 76.8%; specificity 68.9%) and 8.75 for CDAI (sensitivity 72.7%; specificity 68.0%); areas under the curve were roughly the same with a slightly better discriminative ability with DAS28 (0.76 Vs 0.74 for SDAI and 0.73 for CDAI).
Conclusions According to the various indices performed on an Algerian population, good agreement is noted between different states, except for CDAI. For remission, SDAI shows the best agreement with the ACR/EULAR remission criteria. Finally, DAS28 has the best ability for treatment changes.
Disclosure of Interest None Declared