In a population of patients with non-valvular atrial fibrillation (NVAF), and treated with oral anticoagulants (OAC) in routine clinical practice in England, this study will describe patients treated with each OAC, and to estimate and compare event rates of treatment discontinuation, bleeding, and healthcare resource utilisation (HCRU) across OAC treatments.

Inclusion Criteria:

1. Have a prescription for apixaban, rivaroxaban, dabigatran, or VKA during the study period
2. Are ≥ 18 years of age at index date
3. Have ≥ 12 months of computerised medical data prior to index date (date of OAC initiation)
4. Have a record of AF on or ever prior to index date

Exclusion Criteria:

1. Have a record for a valvular condition that is considered causal for AF on or ever prior to index date
2. Have a history of the OACs which are prescribed during the study period (ie, history of index OAC, as patients need to be newly prescribed either apixaban, rivaroxaban, dabigatran, or VKA)
3. Have more than 1 OAC exposure which starts on the same date

Adults with NVAF recorded in the UK Clinical Practice Research Datalink(CPRD) and Hospital Episode Statistics(HES) database, and who were newly prescribed with apixaban, rivaroxaban, dabigatran or VKA from 1st December 2012 and 31st December 2015 (or up to the most recent data available).