Study population
From the base cohort (see above) we restricted the dataset to patients with at least two INR measurements, at least one recorded vitamin K antagonist (VKA) dispensing, and at least one statin dispensing. Within this cohort we identified all incident prescriptions for simvastatin, atorvastatin and rosuvastatin, defined as a prescription with no preceding prescription for a statin within the last 2 years. If one individual had two such incident prescriptions for the same statin, only the first was included. We further restricted to those with at least one INR measurement 8 weeks before statin initiation, as well as at least one INR measurement within 12 weeks after statin initiation. Finally, we excluded those with no VKA prescription within 8 weeks before statin initiation and those aged <18 years at the time of statin initiation.