An assessment of bleeding risk should be part of a patient assessment before starting anticoagulation ther-
apy. Our analysis did not include bleeding risk, result ing in a potential overestimate for undertreatment.
Furthermore, the data captured in this analysis cover only a short period of patients’ medical histories; the
information was limited to prescribing patterns and outcomes during the 1 year after the diagnosis of AF. The
data may have omitted relevant risk-related events that occurred before the diagnosis of AF (eg, a previous
stroke), and this might have led to underestimations of stroke risk scores.