Personalized benefit‐risk assessments combining clinical trial and real‐world data provide further insights into which patients may benefit most from therapy: Demonstration for a new oral antiplatelet therapy

PURPOSE: Quantitative benefit-risk (B-R) assessments are used to characterize treatment by combining key benefits and risks into a single metric but have historically been done for the "average" patient. Our aim was to conduct an individualized assessment for the oral antiplatelet vorapaxar by combining trial and real-world data to further personalize the treatment profiles. METHODS: Using linked UK health care databases, we developed risk prediction equations for key ischemic and bleeding events using Cox proportional hazards models. Trial hazard ratios, relative to placebo, were applied to baseline risk estimates to compute expected attributable risks, summed to derive a per-patient net clinical benefit (NCB). High risk subgroups were defined a priori, and Gaussian mixture models (GMM) were fit to characterize the NCB distribution and identify subgroups with similar NCBs. RESULTS: NCB was consistently positive for all subgroups, likely due to the outcome correlation, and would remain positive with a 12-fold increase in bleeding risk. GMMs identified three distinct NCB subgroups. Compared with the middle/lower NCB subgroups, those with a higher NCB tended to be older, female, and have higher CV disease burden. CONCLUSIONS: Personalized B-R assessments are feasible and clinically valuable and can be used to better predict who would benefit most from therapy.