THU0158 Which Patient Characteristics Influence The Choice of Biological Therapy after First TNF Inhibitor Therapy in RA? A Nationwide Study of Channeling in Sweden 2010-2012

Background The past decade has seen the introduction of many biological DMARDs (bDMARDs) for treating RA, but there is only limited evidence from comparative studies to guide the choice of therapy for specific patients. Whilst a TNF inhibitor (TNFi) is typically used as the first bDMARD, many patients will switch from this initial therapy. The choice of the next bDMARD is, however, largely a matter of patient and physician preference. Perceived or established differences between available bDMARDs may lead to non-random allocation of treatment, a quantification of which is essential for a correct assessment of comparative effectiveness and safety in observational studies. Objectives To describe baseline patient characteristics at initiation of different bDMARDs after a switch from a TNFi as first bDMARD Methods Patients with RA who initiated a second bDMARD in 2010–2012 within one year of discontinuing initial TNFi treatment were identified using the Swedish Rheumatology register (SRQ). Data on disease history, defined as ever diagnosed with relevant co-morbidities, were retrieved by linking to the nationwide and virtually complete Swedish Patient Register. Results The most common bDMARDs after initial TNFi therapy were rituximab and etancercept; the least common was infliximab. Across drugs, there were only modest differences in demographics and RA characteristics, and moderate differences in disease history. Those initiating a second TNFi were slightly younger, less often RF+, and had slightly lower DAS28 than those initiating a non-TNFi bDMARD. Initiators of rituximab had longest disease duration, highest proportion RF+ and with history of malignancy or COPD, while etanercept-initiators had the lowest proportion with history of malignancy. Those initiating a second TNFi had more often switched due to adverse events, though lack of effect was the most common reason for all drugs. Conclusions We observed some evidence for channeling regarding disease history, both by class and by drug within class. These differences were small compared to previously reported channeling to non-TNFi as first bDMARD, but should still be taken into account in observational studies of comparative effectiveness and safety of different biological therapies. Disclosure of Interest None declared