• biologics
• real-world
• rheumatoid arthritis

Objectives: To test algorithms to select rheumatoid arthritis (RA) patients from the Tuscan healthcare administrative database (THAD). To identify and characterize trajectories and to investigate the relationship with the disease activity.
Methods: The PATHFINDER study is a retrospective population-based study conducted on THAD (EUPAS29263). Three analyses were performed during this Ph.D. program.
In the validation analysis, patients accessing the Rheumatology ward of Pisa University Hospital were extracted from THAD and the related medical charts represented the reference. The reference group included first-ever users of bDMARDs between 2014 and 2016 (index date) with at least a specialist visit. We tested four algorithms combining information on RA co-payment exemption codes, discharge record RA codes and emergency department access RA codes. We estimated sensitivity, specificity, positive and negative predicted values (PPV and NPV) with 95% confidence interval (95%CI).
In the drug-utilization analysis, we included RA patients identified by the best performing algorithm. We observed users for 3 years or until death, neoplasia, or pregnancy. We evaluated adherence quarterly, through the Medication Possession Ratio. Firstly, we identified adherence trajectories; then, we focused on the most populated trajectory to distinguish the related sub-trajectories.
In the effectiveness analysis, we extracted information on drug supplies from THAD and disease activity from medical charts. We measured discontinuations as the result of the coverage of each bDMARD dispensation based on DDD and the number of doses supplied plus a grace period amounting to 60 days and 30 days in the main and sensitivity analysis, respectively. We assessed DAS28 before (T0), after (T1) the first bDMARD supply, and before (TD0), after (TD1) discontinuation.
Results: In the validation analysis, among 277 reference patients, 103 had RA. The fourth algorithm identified 96 true RA patients, PPV 0.78 (95%CI 0.70-0.85), sensitivity 0.93 (95%CI 0.86-0.97), specificity 0.84 (95%CI 0.78-0.90), and NPV 0.95 (95%CI 0.91-0.98).
In the drug-utilization analysis, among 935 patients we identified 49 fully-adherent users, 829 continuous users, and 57 early-discontinuing users. Significant differences were observed among the index drugs. Three sub-trajectories were identified: continuous-steady users (556), continuous-alternate users (207), and continuous-declining users (66). No relevant differences emerged at the baseline.
In the effectiveness analysis, 33 users had at least 1 discontinuation and 3 DAS28 assessments recorded. At T0, 28 patients had DAS28 >3.2 and 13 had DAS28 ≤3.2. At T1, 28 patients showed DAS28 >3.2, while 38 patients DAS28 ≤3.2. In the subgroup of RA first-ever bDMARD users with at least one discontinuation, a significant distribution of DAS28 >3.2 was observed at T1. Users continuing treatment displayed most frequently a DAS28 ≤3.2. Overall, 91 discontinuations were estimated and 38 displayed a DAS28 ≤3.2 at TD0. Out of 37 discontinuations with DAS28 recorded both at TD0 and at TD1, 27 showed no improvement.
Conclusions: In conclusion, we identified the algorithm combining RA co-payment exemption code AND RA discharge record OR RA ED access record as the best performing. The majority of first-ever bDMARD users had a continuous adherence behavior. Disease control was achieved in half of RA patients starting bDMARDs, supporting treatment interruption.