Text

Introduction: Systemic glucocorticoids (GCs) remain integral to rheumatoid arthritis (RA) management; however, their long-term safety thresholds remain controversial. Clinically meaningful toxicity is generally assumed at doses exceeding 5 mg/day (prednisone equivalent), whereas the safety profile of very low-dose exposure remains uncertain. Moreover, the combined influence of dose intensity and treatment duration on adverse event (AE) risk has not been comprehensively quantified. We therefore examined dose–response relationships and temporal exposure dynamics of GC-associated AEs in RA using nationwide real-world data.

Methods: German statutory health insurance claims data (2007–2022) were analyzed. Adults with confirmed RA initiating systemic GCs were matched 1:1 to GC-free controls based on age, sex, comorbidity burden, region, and index quarter. Incident AEs comprised cardiovascular, metabolic, infectious, neuropsychiatric, and musculoskeletal outcomes. Incidence rate ratios (IRRs) were estimated across predefined average daily dose strata using Poisson regression. To disentangle dose–time interactions, piecewise exponential additive mixed models (PAMMs) jointly modeled mean daily dose and time since initiation, enabling classification of exposure–response patterns as dose-driven, time-driven, or balanced.

Results: Among 47,693 RA patients, 8,011 initiated systemic GCs; 5,531 were matched to controls. Composite AE risk increased progressively at doses >5 mg/day, with marked escalation at ≥7.5 mg/day. Cardiovascular events rose from an IRR of 1.83 at 5–<7.5 mg/day to 6.88 at ≥7.5 mg/day. Acute mental disorders increased from 0.95 at 2.5–<5 mg/day to 6.80 at ≥7.5 mg/day. In contrast, very low-dose exposure (>0–<2.5 mg/day) was associated with a lower composite AE incidence compared with controls (IRR 0.59 [0.54–0.63]). Within the 2.5–<5 mg/day range, risks were largely neutral during shorter treatment durations. PAMM analyses indicated pneumonia risk as predominantly dose-driven, whereas cardiovascular and psychiatric outcomes demonstrated balanced dose–time dynamics.

Conclusion: In RA, GC-associated harm is strongly dose-dependent, with substantial risk escalation beyond 5 mg/day. Very low-dose exposure (≤2.5 mg/day) was not associated with measurable excess risk and was linked to lower composite AE incidence, suggesting a favorable net clinical balance between inflammation control and GC-related toxicity. These findings support minimizing cumulative exposure and avoiding higher-dose regimens while recognizing distinct safety profiles at very low doses.

Disclosures: The study was funded by GSK (ID 222292) and is registered as clinical trial (ID: NCT06488703).