Factors associated with oral glucocorticoid use in patients with rheumatoid arthritis: a drug use study from a prospective national biologics registry

Background: Glucocorticoids (GCs) are used in similar to 60% of patients with rheumatoid arthritis (RA). Although diseasemodifying, they also have significant adverse effects. Understanding factors associated with GC use may help minimise exposure. The aims of the present study were to describe oral GC use in RA; determine any change in use over time; and determine factors associated with oral GC use, commencement or cessation. Methods: Adult patients with RA were identified in the Australian Rheumatology Association Database (ARAD), a national Australian registry that collects long-term outcome data from patients with inflammatory arthritis. Patients were categorised by their ARAD date of entry (DOE), with population-averaged logistic regression and transition state analysis used to determine any change in GC use over time. Fixed-effects panel regression was used to examine whether GC current use was associated with pain/arthritis activity/Health Assessment Questionnaire (HAQ) scores or medication use. Transition state analysis was used to assess whether these factors influenced the commencement or cessation of GCs. Results: A total of 3699 patients with RA completed a baseline ARAD questionnaire (73% female, mean age 57 years). The probability of GC use decreased over time according to ARAD DOE: September 2001 to March 2005, 55% (95% CI 52-58%); March 2005 to September 2008, 47% (45-49%); September 2008 to March 2012, 42% (39-45%); and March 2012 to October 2015, 39% (34-43%) (p < 0.001). Conventional synthetic disease-modifying anti-rheumatic drugs (OR 10.13; 95% CI 8.22-12.47), non-steroidal anti-inflammatory drugs (1.18; 1.02-1.37) and opioids (2.14; 1.84-2.48) were associated with GC current use, as were lower pain scores (0.94; 0.90-0.98), higher arthritis activity scores (1.09; 1.05-1. 14) and poorer HAQ scores (1.52; 1.30-1.79). Use of biologic disease-modifying anti-rheumatic drugs (bDMARDs) was not associated with GC current use (0.98; 0.83-1.15) or GC cessation (HR 0.87; 95% CI 0.75-1.01), but it was associated with GC commencement (0.54; 0.47-0.62). Conclusions: The probability of oral GC use decreased over time, with reduced commencement and increased cessation of GCs. The modest effect of bDMARDs on GC cessation was not statistically significant.