(69) Evaluating Brazil's National Administrative Claims Data for Real-World Pharmacoepidemiologic Studies: Trends in Rheumatoid Arthritis Using DATASUS (2010-2020)

Publication Number: 400

Background: Real-world evidence on the treatment trends and dynamics for rheumatoid arthritis (RA) in Brazil is limited. DATASUS is a large national administrative claims dataset of the public health care system in Brazil that integrates ambulatory and hospital databases with drug dispensation and all procedures performed in SUS (Sistema Único de Saúde) for 150 million individuals who are exclusively dependent (~75% of population).

Objectives: To evaluate DATASUS as a potential data source in pharmacoepidemiological (PE) studies.

Methods: We used DATASUS to identify patients who had at least one claim of RA and ≥2 claims of disease-modifying anti-rheumatic drug (DMARD) after 01-Jan-2010; conventional synthetic (cs), biologic (b) or targeted synthetic (ts) DMARD. Patients were excluded if had < 6-month follow-up or use of DMARD prior to index. Index date was the first DMARD claim, follow up was until end of the study period (31-Dec-2020) or last information available. DMARD starts, duration, discontinuation, and switching were measured and stratified for SUS-exclusive and non-SUS exclusive cohorts. Multivariate logistic and cox regression evaluated determinants of initiating or switching to a new tsDMARD, respectively.

Results: We identified 250,251 RA patients with mean age (SD) 58.4 (15.3) years. Majority were female (83%), white (58%), and resided in the Southeastern states (53%). Of those, 62% were SUS-exclusive and 38% non-SUS exclusive. b/tsDMARDs were used by 118,135 (47%) patients; among them, adalimumab was most common (47%, 32%; in 2010, 2020), followed by etanercept (34%, 19%; 2010, 2020), and others. Age (adjusted odds-ratio 1.78 [50+]; 95%CI 1.57-2.01), SUS exclusive (0.53; 0.47-0.59), distance to clinic (0.57 [160+ km]; 0.45-0.72), and pre-index immunosuppressant csDMARD claims (1.23; 1.08-1.41) were independent predictors of initiating a novel oral tsDMARD (JAK inhibitor). For switching from bDMARD to tsDMARD, associations were similar, except directionality of SUS exclusive (1.10; 1.03-1.18) and distance to clinic (1.18; 1.03-1.35), and number of previous bDMARD (0.15; 0.14-0.16).

Conclusions: DATASUS is a valid source for treatment-related analyses in RA reflecting the public health system in Brazil. Distance to clinic, number of previous bDMARDs, and SUS exclusive were predictive of patients switching to a tsDMARD, which may relate to the convenience of an oral DMARD. Future investigations should consider misclassification bias as many individuals (38% of our cohort) have private health insurance and may seek SUS to access high-cost medication while not seeking additional care within SUS.