Document resource
Objectives To identify predictive correlates of discordance between the Physician’s Global Assessment Score (PGA) and remaining Lupus Low Disease Activity State (LLDAS) criteria, among patients who attain low disease activity at individual visits.Methods Patient visit data from the Asia-Pacific Lupus Collaboration (APLC) Cohort (2013-2020) were collected, with prior ethical approval. We included all visits where patients attained LLDAS criteria, minus the PGA criterion. PGA (0-3) concordance was defined as PGA<1, discordance defined as PGA>1, per LLDAS criteria. Data including demographics, clinical disease activity (SLE Disease Activity Index (SLEDAI-2K)), serologic disease activity (C3, C4, anti-DNA antibodies), inflammatory markers (C-reactive protein and Erythrocyte Sedimentation Rate) and treatments were collected. Statistical analysis was performed using descriptive statistics and mixed-effects logistic regression models.Results We included 19632 visits of 2973 patients (92% female, 88% Asian, median follow-up 2.5 years). Four hundred and fifty-six visits (2.9%) were not in LLDAS due to PGA discordance; the remainder met all LLDAS criteria. During most visits (n=18328, 93%), patients had only serologic or no disease activity and PGA discordance was 1.5-3.9%. Visits with arthritis (n=156) had the highest frequency of PGA discordance (14.7%). Demographic factors negatively associated with PGA discordance included age, tertiary education, high-income country residence. Rash, alopecia, arthritis, thrombocytopenia, leukopenia, and serologic activity were independent predictors of PGA discordance. Across visits with only serologic activity, higher SLEDAI-2K score at previous visit, raised anti-DNA antibodies, and addition of prednisolone were predictive of PGA discordance. Across visits with no disease activity, higher SLEDAI-2K score at previous visit, prednisolone dose intensification and addition of azathioprine were predictive of PGA discordance; hydroxychloroquine use was negatively associated with discordance.Conclusions PGA discordance with other LLDAS criteria was infrequent in the APLC cohort (2.9%), due to most eligible visits having no documented clinical disease activity. However, among visits with clinical disease activity, PGA was discordant in up to 15% of cases. PGA discordance in visits with serologic only or no disease activity was predicted by higher SLEDAI-2K score at previous visit or treatment intensification. PGA may therefore operate as an indicator of disease severity or disease activity not captured by SLEDAI-2K, within the LLDAS definition.