Background Biotherapies such as subcutaneous tumour necrosis factor-alpha inhibitors (SC-TNFis) have transformed the management of inflammatory joint diseases such as ankylosing spondylitis (AS). The assessment of SC- TNFis persistence and its determinants is needed.

Objectives The objective of this study was to describe treatment persistence in real-world settings, and identify the determinants of persistence among AS patients initiating treatment with an SC-TNFi.

Methods The French national health insurance scheme database lists all outpatient and inpatient healthcare consumption for individuals covered by the general health insurance scheme. Using French claims data, AS was diagnosed using Long Term Disease status and hospital admission, based on ICD-10 codes. Patients were then identified through prescription filled for adalimumab (ADA), etanercept (ETA), certolizumab pegol (CZP) and golimumab (GLM) between 2012/07/01 and 2013/12/31. A patient was considered as non-persistent in the event of a prolonged interruption of the therapy lasting 91 days or more. Persistence was estimated with Kaplan Meier analysis. Determinants of persistence in the 12 months before initiation were identified using Cox models.

Results A total of 9,098 patients with AS were identified. In the descriptive analyses of the 12 months persistence, differences were observed for AS patients, with raw/non-adjusted persistence rates of 33.2% for CZP, 49.3% for ETA, 52.4% for ADA and 54.5% for GLM. Results of the Cox model are presented, including hazard ratio for biotherapy, adjusted on sex, age, socio-economic status, and criteria on disease severity. The variables biotherapy, socio-economic status and hospital admission for IRMD did not meet the proportionality hypothesis of risks, and were corrected by the addition of a variable integrating the interaction with time.

Conclusions Non-persistent patients were more likely female, with deprived socio-economic status, multiple comorbid conditions, and multiple line of biotherapy. Age, hospital admission for IRMD and treatment with GLM (compared to CZP, ETA and ADA) decreased the risk of non-persistence. Further analyses are needed to assess the impact of non-persistence.

Disclosure of Interest B. Fautrel Grant/research support from: AbbVie, MSD, Pfizer, Consultant for: AbbVie, Biogen, BMS, Celgène, Hospira, Janssen, Lilly, MSD, NORDIC Pharma, Pfizer, Roche, SOBI, UCB, M. Belhassen Employee of: PELyon, C. Hudry Consultant for: ABBVIE, MSD, BMS, PFIZER, ROCHE, CELGENE, NOVARTIS, BIOGEN, UCB, SANDOZ, AMGEN., Employee of: COCHIN Hospital, M.-C. Woronoff: None declared, N. Gouyette Employee of: MSD France, A. Clément Employee of: MSD France, E. Van Ganse Consultant for: PELyon; ALK ABELLO; AstraZeneca; Bayer; BMS; BIF; GSK; IMS; LASER; MSD, F. Tubach: None declared