Introduction: Regardless of the benefits of stroke prevention, only 50% of atrial fibrillation (AF) patients recommended for oral anticoagulation (OAC) actually use these medications. We identified groups of patients with similar OAC adherence patterns, and evaluated what patient characteristics affect group membership. Hypothesis: We hypothesized that patients continuously adherent to OAC would be more likely to reside in the Midwest and Northeast and have higher socioeconomic score (SES) than patients who never used OAC. Methods: Using Medicare claims data, we identified continuously enrolled beneficiaries with a first diagnosis of AF in 2014-2015 (n=36,185). We calculated the proportion of days covered with OAC in the first 12 months after first diagnosis. Using group-based trajectory models, we identified trajectories of OAC adherence. We constructed multinomial logistic regression to evaluated how a comprehensive list of demographics, system-level factors and clinical characteristics (list in figure) affected group membership. Results: We identified 4 trajectories of OAC adherence: patients who never used OAC (group 1, 43.6%), late OAC initiators (group 2, 7.6%), early OAC initiators who discontinued OAC (group 3, 8.9%), and continuously adherent patients (group 4, 39.9%). Black race, eligibility for low income subsidy, residence in segregated metropolitan areas (measured by index of dissimilarity), residence in the Southeast, Southwest or West, higher HAS-BLED score, and chronic kidney disease decreased the odds of belongings to the continuously adherent group. Conclusions: Trajectories of OAC adherence were highly variable, and were determined not only by demographics and clinical characteristics, but also system-level factors including income level, segregation measures, and region of residence. These trajectories may be useful in the implementation of targeted strategies to mitigate OAC underuse in stroke prevention.