OBJECTIVE: To develop and validate a modified version of the Disease Activity Score with 28 joint count (mDAS28), for use in epidemiological research, when acute-phase reactant values are unavailable.

METHODS: In a cross-sectional development cohort (5729 patients), statistically significant predictors of the logarithm of erythrocyte sedimentation rate (lnESR) were identified. After computation of the mDAS28, a cross-sectional validation cohort (5578 patients) was used to evaluate internal, criterion, and construct validities. The ability of the mDAS28 to discriminate between disease states was also assessed. A second validation cohort (longitudinal, 336 pairs of patient visits) was used to assess sensitivity to change.

RESULTS: Significant predictors of lnESR included tender and swollen joints with 28 counts, patient's and physician's assessments of global health, and patient's assessment of pain (visual analog scale 0-100 mm) and a physical function (modified Health Assessment Questionnaire 0-3; mHAQ). Satisfactory internal validity (alpha = 0.72) and strong criterion validity compared to the DAS28, the Simplified Disease Activity Index (SDAI), and the Clinical Disease Activity Index (CDAI) (r = 0.87-0.96) were found. Predictive validity was demonstrated by good correlation with the mHAQ (r = 0.58). The mDAS28 showed substantial agreement with the DAS28, SDAI, and CDAI in discriminating between disease states (kappa = 0.70-0.77) and moderate to substantial agreement between response levels (kappa = 0.52-0.73). Both mDAS28 and DAS28 measures classified patients similarly in remission compared to the SDAI and CDAI. The mDAS28 was superior in detecting change (standardized response mean = 0.58) followed by the DAS28, CDAI, and SDAI.

CONCLUSION: The mDAS28 is a valid and sensitive tool to assess disease activity in epidemiological research, as an alternative to the DAS28, when acute-phase reactant values are unavailable.