ObjectivesPrevious studies demonstrated equivalence in terms of efficacy and safety of biosimilars (bsDMARDs) compared to original treatments (boDMARDs) and in switching situations. Less is known about what happens when initiating a bsDMARD in a molecule naïve patient. The objectives of our study were to compare the retention of treatment of subcutaneous boDMARDs and bsDMARDs globally, depending on the disease [rheumatoid arthritis (RA), spondyloarthritis (SpA), or psoriatic arthritis (PsA)], molecule [etanercept (ETN) or adalimumab (ADA)], line of treatment, or presence of citrate in the context of first use of each molecule (namely initiation) and to analyze treatment retention’s predictive factors.Materials and methodsThis multicenter retrospective study used data from shared medical records of the RIC-FRANCE network, encompassing the prescription of hospital rheumatologists and attached practitioners, of patients with RA, SpA, or PsA, with the starting ETN between 03/10/2016 and 31/07/2020, or ADA between 23/10/2018 and 31/07/2020. Clinical data were collected from medical records. Retention analysis was performed using Kaplan–Meier curves and the log-rank test. Retention’s predictive factors were analyzed using Cox proportional-hazard ratio.ResultsEight hundred forty-five prescriptions were analyzed: 340 boDMARDs and 505 bsDMARDs. About 57% of prescriptions concerned women. The mean age was 51.8 years. About 38% were prescriptions for RA, 16% for PsA, and 46% for SpA. An increase in the initiation over time was observed for both ETN and ADA. The retention rate of bsDMARDs was superior to boDMARDs’ one (39 vs. 23 months; p = 0.045). When molecules are compared, the difference was significant only for ETN (45 vs. 19 months for boDMARD; p = 0.0265). When comparing diseases, the difference in favor of bsDMARDs was significant in patients with RA only (p = 0.041). Citrated treatments displayed better retention compared to citrate-free treatments (p = 0.0137). Multivariable analysis of predictive factors for the cessation of treatment found shorter disease duration, boDMARD prescription, hospital practitioner prescription, late line of treatment, and female sex as significant. More side effects were observed with boDMARDs, especially more infections (17.8% vs. 7.8%).ConclusionEven if bsDMARDs’ prescription increases over time, its penetration rate is still below expectations. bsDMARDs displayed better retention compared to boDMARDs, especially for ETN, and in patients with RA. Citrated treatments had better retention. Prescription by a full-time hospital-based rheumatologist is associated with poorer retention.