Drug retention of biological DMARD in rheumatoid arthritis patients: the role of baseline characteristics and disease evolution

Objective To examine the association of the evolution in physician-reported and patient-reported outcomes with decision to stop biological DMARDs (bDMARDs) in RA. The contribution of baseline characteristics is well established, but little is known about how the disease evolution influences the decision to discontinue therapy. Methods RA patients who initiated a bDMARD treatment from 2009 and with information on date of visit were pooled from seven European RA registers. Each outcome was divided into baseline assessments (capturing the inter-individual differences at drug initiation) and changes from baseline at subsequent visits (capturing the individual evolution). Cox regression models were used to examine their association with drug discontinuation, adjusting for baseline patient and co-therapy characteristics and stratifying by register and calendar year of drug initiation. Results A total of 25 077 patients initiated a bDMARDs (18 507 a TNF-inhibitor, 3863 tocilizumab and 2707 abatacept) contributing an amount of 46 456.8 patient-years. Overall, drug discontinuation was most strongly associated with a poor evolution of the DAS28, with a hazard ratio of 1.34 (95% CI 1.29, 1.40), followed by its baseline value. A change of Physician Global Assessment was the next strongest predictor of discontinuation, then the Patient Global Assessment. Conclusions The decision to discontinue treatments appears to be mostly influenced by DAS28 and particularly its evolution over time, followed by Physician Global Assessment evolution, suggesting that the decision to stop bDMARDs relies more on the physician's than on the patient's global assessment.