Background: Use of glucocorticosteroids and NSAIDS for the treatment of RA has been associated with an increased risk of gastrointestinal (GI) perforations. The introduction of disease modifying agents, such as methotrexate or tumour necrosis factor inhibitors (TNFi) provided a seemingly safer treatment option, but the safety of other types of biologics relative to TNFi remains unclear.

Objectives: To estimate the incidence of gastro-intestinal perforations among Swedish RA patients treated with TNFi and non-TNFi biologics, and compare it with the incidence among bionaïve patients with RA and a matched general population comparator group.

Methods: We performed a register-based cohort study, including all Swedish RA patients, with follow-up between 2010 and 2015. For these, all treatment initiations with biologic disease modifying anti-rheumatic drugs were identified through the Swedish Rheumatology register (SRQ), and grouped by class into TNFi and non-TNFi drugs (10 857 and 5823 treatment starts). Biologics naïve patients with RA were identified by recorded diagnosis in the Swedish National Patient Register (NPR), since 2001 (n=54,732). Five general population controls were matched to each biologic-treated patient with RA (n=53,214) by sex, age, and geographical region. A patient could participate to several of these exposure cohorts. For each patient, follow-up under a cohort ended either with an outcome event (GI perforation identified in the NPR as an ICD10 code from a predefined list) or with the first of any of the following censoring events: emigration from Sweden, death, transition to another cohort, discontinuation of treatment (+90/+180 days lag time) or end of study period. Crude incidence rates were tabulated for each cohort and adjusted hazard ratios (HR) and 95% confidence intervals were estimated in multivariable Cox regressions, controlling for baseline differences. The final adjusted model included the following covariates: sex, age, line of biologic treatment, disease characteristics, co-medication at treatment start, co-morbidities and a history of GI perforation.

Results: We found 31 GI perforations among 18 604 person-years (pyr) exposed to TNFi, and 31 GI perforations among 10 947 pyr exposed to non-TNFi, corresponding to crude incidence rates of 1.67 and 2.83 per 1000 pyr, respectively. The crude incidence rate among the biologics-naïve was 2.54 while among the general population comparators it was 0.94. The rate of GI perforations remained higher in patients with RA compared to the general population after adjustment for patient characteristics, HR of 1.78 (95% CI: 1.44 to 2.17), whereas the seemingly increased rate among bionaïve and non-TNFi users vs TNFi was largely explained by differences in age and disease history at start of follow-up, with adjusted HRs of 1.10 (0.68–1.78) for TNFi vs bionaïve and 1.10 (0.63–1.91) for TNFi vs non-TNFi, respectively.

Conclusions: Although patients with RA had a higher rate of GI perforations than matched general population comparators, no significant differences in risk remained between bionaïve, TNFi or non-TNFi treated RA patients after adjusting for baseline patient characteristics.

Disclosure of Interest: None declared

DOI: 10.1136/annrheumdis-2018-eular.6831