Despite the availability of treatment guidelines and effective treatments, real-world effectiveness remains suboptimal partly due to poor patient medication adherence. We evaluated a comprehensive set of sociodemographic, health insurance, and disease-related factors for association with patient decision to discontinue anti-rheumatic medications (ARMs) in a large observational RA cohort in Ontario, Canada. Patients from the Ontario Best Practices Research Initiative registry were included. The following predictors of ARM discontinuation were evaluated with cox-regression: patient age, gender, education, income, smoking, health insurance type/coverage, RA duration, erosion presence, RF positivity, DAS28-ESR, physician global, HAQ-DI, comorbidity number, ARM types, and physician characteristics (gender, academic position, urban vs. rural, distance from patient's residence). Patients (1762) were included with a mean (SD) age of 57.4 years (13.0). Approximately 80% were female, 29% had early (≤ 1 year) RA, and 70% were RF-positive. Mean (SD) baseline DAS28-ESR and HAQ-DI were 4.5 (1.5) and 1.2 (0.76), respectively. In multivariate analysis, married status (HR [95%CI] 0.73 [0.56-0.96]), RF positivity (0.73 [0.56-0.96]), and higher comorbidity number (0.92 [0.85-0.99]) were significant predictors of ARMs continuation while higher physician global (1.10 [1.04-1.15]), NSAID use (1.75 [1.29-2.38]), and number of ARMs (1.23 [1.07-1.40]) were associated with ARMs discontinuation. In a subset analysis assessing conventional or biologic DMARD discontinuation, higher HAQ-DI and biologic use over time were associated with lower hazard for discontinuation. Several sociodemographic, disease, and treatment parameters were identified as independent predictors of patient discontinuation of ARMs. These results should be considered when developing patient adherence support programs and in the choice of treatment regimens.
Keywords: Adherence; Anti-rheumatic medication; DMARDs; Registry; Rheumatoid arthritis; Treatment discontinuation.