Background/Purpose: The rheumatoid arthritis (RA) population starting tumor necrosis factor inhibitors (TNFi) today is much different in terms of accumulated and concurrent disease activity and comorbidity, than patients starting treatment 10 years ago. Still, few studies have investigated time-trends in infection risks and their determinants, on a clinically relevant scale. Our aim was to calculate the absolute risk of serious infection (SI) within 1 year of TNFi-initiation in patients with RA in Sweden and to examine whether the risk changes over calendar time and patient characteristics.

Methods: Patients with RA who initiated a TNFi from 2004 to 2011 were identified from the Anti-Rheumatic Therapies in Sweden (ARTIS) register. Patient demographics, concomitant medications and clinical measures were collected from the patient’s first visit (baseline). ICD-10 codes for SI hospitalizations were identified in the inpatient register (2004-2012) and linked to the study population using each patient’s unique personal identification number. We calculated the absolute risk of SI hospitalization within 1 year of treatment start using modified Poisson regression models stratified by risk factors previously identified by the German biologics register, RABBIT: concomitant corticosteroid use, total number of risk factors (age≥60, previous serious infection 1 year before TNFi-initiation, COPD or chronic kidney disease). We additionally stratified risks by calendar year of initiation (2004-4007 vs. 2008-2011) and investigated other comorbidities (cardiovascular disease, diabetes, number of hospitalizations, days hospitalized, outpatient visits) and disease activity measurements (DAS-28 and its components, HAQ, disease duration).

Results: We included 8562 biological-initiators with RA. A total of 344 (4.0%) individuals had at least one SI hospitalization within a year of biological initiation (2004-2007 4.4%; 2008-2011 3.7%). Serious infection risk decreased over time in age and sex-adjusted models (p value for calendar year 0.02) but when adjusted for comorbidities, year of TNFi-initiation was no longer significantly associated with a decreased risk of SI (p=0.15). Individuals who were older, male, with longer disease duration, a history of infection and higher disease severity were at an increased risk of SI. Our results from 2004-2007 were very similar to those reported by RABBIT. When examining risks from 2008-2011, the majority of the increased risk was observed in individuals with 3 or 4 risk factors (Figure).

Conclusion: Risks of SI within 1 year of TNFi initiation were similar to those reported by the German Biologics Register 2004-2007. Although the population starting TNFi has changed over time, the one-year risks of SI has only dropped modestly, but the relative importance of SI risk factors has changed, such that infection risk calculators need be updated using contemporary data.