Background Oral glucocorticoids (GCs) are commonly used to treat rheumatoid arthritis (RA) with GC prescriptions often issued in primary care. Some general practitioners (GPs) may have a greater tendency to prescribe GCs, which could affect the likelihood of a patient receiving a GC prescription. Patient characteristics such as age and comorbidities may influence GC prescribing differently, depending on GC prescribing tendency.

Objectives To a) determine whether GC prescribing tendency influences GC prescriptions for RA patients and b) whether baseline patient characteristics affect GC prescribing differently in patients seen by prescribers with high vs low GC prescribing tendencies.

Methods Patients with incident RA during a defined follow up period were identified from the Clinical Practice Research Datalink, a large UK general practice database. The primary (i.e. most frequent) prescriber of all medications was determined for each patient. For each primary prescriber, the mean proportion of time their patients spent on GCs during follow up was calculated. The median of these values was used as a cut-off for “low” or “high” GC prescribers. Logistic regression was used to determine the effect of prescriber tendency on the likelihood of receiving a GC prescription during follow up and to test the interaction between prescriber tendency and a range of baseline patient characteristics, including patient demographics, comorbidities that are also indications for GC use and known steroid-related comorbidities.

Results 16 536 patients with RA were identified and 3270 GPs were assigned as primary prescribers. The mean proportion of time their patients spent on GCs ranged from 0 to 100% (median 10.2%, IQR 0.1%– 24.4%), and GPs were categorised as ‘high’ prescribers if their patients spent a mean of ≥10.2% of follow-up on GCs. 6,372 (38.5%) patients were assigned a “low” GC prescriber and 10,164 (61.5%) were assigned a “high” GC prescriber. The odds of a patient receiving a GC prescription during follow up was 3.1x greater if they were seen by a “high” GC prescriber compared to a “low” GC prescriber (95%CI 2.88-3.29). The probability of receiving a GC prescription increased with age, but the effect differed significantly between the prescriber groups: OR 1.14 (95%CI 1.10, 1.18) per decade in the “low” GC prescriber group, OR 1.27 (95% CI 1.23-1.30) in the “high” GC prescriber group. Baseline comorbidities that influenced the probability of receiving a GC prescription and that were significantly different between GC prescriber groups were asthma (“low” OR 2.65 CI 2.29-3.06, “high” OR 1.76 CI 1.57-1.98) lower respiratory tract infection (“low” OR 1.75 CI 1.53-2.00, “high” OR 1.45 CI 1.31-1.61), hypertension (“low” OR 1.21 CI 1.03-1.42, “high” OR 1.30 CI 1.18-1.43), and inflammatory bowel disease (“low” OR 0.85 CI 0.46-1.56), “high” OR 1.92 CI 1.21-3.06).

Conclusions Prescriber tendency towards high GC use was associated with an increased likelihood of RA patients receiving GCs. The effect of certain patient characteristics on GC prescription differed according to prescriber tendency.

Disclosure of Interest None declared