Validation of proposals for definitions of moderate and severe disease activity in SLE: impact on flares, quality of life, damage accrual, hospitalisations and mortality

Abstract

INTRODUCTION: Accurate assessment of disease activity in SLE is crucial but challenging due to its varied clinical manifestations and severity. Current tools like the SLE Disease Activity Index (SLEDAI) have limitations, including unvalidated cut-off points, low sensitivity to certain severe features and an overemphasis on serological markers. There is a need for improved definitions of disease activity. METHODS: We analysed data from 1463 patients with SLE in the prospective, multicentre RELESSER-PROS cohort (39 Spanish hospitals) over five annual visits. A panel of lupus experts used the Delphi method to develop new definitions for moderate disease activity state (MODAS) and severe disease activity state (SEDAS). These incorporated clinical SLEDAI (cSLEDAI), selected severe non-SLEDAI manifestations (eg, neuropsychiatric involvement, proteinuria, severe haematological features) and the Physician Global Assessment. We compared the predictive performance of MODAS/SEDAS with SLEDAI for mortality, organ damage, severe flares, hospitalisations and health-related quality of life, using receiver operating characteristics curves. RESULTS: At baseline, 20% of patients met MODAS criteria and 24.6% SEDAS criteria, versus 10.5% and 3.0%, respectively, by SLEDAI. MODAS/SEDAS reclassified 19.9% of patients considered mild by SLEDAI, and 53.3% of moderate cases. MODAS/SEDAS showed modest but consistent improvement in predictive accuracy for damage (area under the curve 0.570 vs 0.550), flares (0.609 vs 0.564) and hospitalisations (0.609 vs 0.565). These definitions were associated with worse outcomes and demonstrated a dose-response relationship, although the overall predictive ability remained moderate. CONCLUSION: MODAS and SEDAS offer an alternative framework for defining moderate and severe SLE activity, with modest but consistent improvements in predictive performance compared with SLEDAI. By integrating cSLEDAI, key severe features and physician judgement, they improve prognostic performance and support a severity-based approach to clinical management and research. Their clinical utility remains preliminary, and further external validation is required before routine implementation.