PMS65 USE OF REGISTRY DATA AS A SOURCE OF REAL WORLD EVIDENCE IN BRIDGING DISCONNECTED NETWORKS OF FIRST AND SECOND LINES OF THERAPIES IN RHEUMATOID ARTHRITIS

Objectives: Building on our research as part of IMI GetREAL, we aim to develop methodology for incorporating real-world data (RWD) into decision-making in the drug development process. Using a case study in rheumatoid arthritis (RA) we aim to explore how RWD can be used to help demonstrate the relative effectiveness of new medicines and, in particular, to optimise an evidence base using first-line evidence to inform second-line effectiveness estimates in the evidence synthesis of biologic therapies in RA. Methods: We make use of data from the British Society for Rheumatology Biologics Register (BSRBR) to supplement randomised trial evidence. We do so by (i) deriving prior distribution for the correlation between treatment effects in first and second lines of biologic agents (ii) using treatment effects derived from the registry data directly. We use these estimates to inform a bivariate network meta-analysis (NMA) model of first and second line treatments. Estimates from registry data are used to bridge disconnected networks for the two lines of therapy. Disease activity score (DAS-28) and American College of Rheumatology response criteria (ACR20) were used as outcome measures. Results: Data were obtained from 44 trials of biologic therapies including 5 for second-line treatment. The bivariate second-line NMA results showed a decrease in uncertainty for individual treatments when compared to a second-line only univariate NMA. The bivariate NMA approach allowed for predictions of treatment effects that had not been evaluated in trials in a second-line setting. Conclusions: The bivariate NMA provides estimates for all treatments in first- and second-line, by using the correlation between them to predict estimates in second-line (or first line) where these estimates did not exist when data were synthesised in the univariate approach. This novel methodology may have further applications, for example for bridging evidence between networks of trials in children and adults.