Healthy user bias in comparative safety studies for brand‐name versus generic products; the example of warfarin

: Warfarin was selected as a case study to examine confounding when comparing a product across different manufacturers because it is a narrow therapeutic index drug with prevalent beliefs for brand-name superiority. Medicare beneficiaries aged ≥65 years with atrial fibrillation and an incident outpatient warfarin prescription from July 2006 through July 2015 were included in the study population (N = 746,098). Substantial imbalances were observed between brand-name warfarin and generics for (i) clinical comorbidity, (ii) socioeconomic status, (iii) prescriber specialty, (iv) recent ambulatory and emergent care, (v) drug adherence, (vi) pharmacy setting (e.g., retail, mail-order), and (vii) risk scores for bleeding and thrombosis. Patients receiving brand-name warfarin were healthier than patients receiving generic manufactured warfarin. Utilization of generic warfarin products also differed by geographic region and pharmacy setting. Manufacturer-level comparative-safety studies for causal inference should carefully consider the presence of these imbalances and their potential for introducing healthy user bias.