The direct oral anticoagulants (DOACs: apixaban, dabigatran, edoxaban, and rivaroxaban) are first-line anticoagulants for many patients with nonvalvular atrial fibrillation (NVAF) and venous thromboembolism (VTE).[1] [2] This is due to their efficacy, safety profile, ease of administration, lack of a routine laboratory monitoring, and limited food or drug interactions. The landmark clinical trials that led to their approval showed generally high rates of mediation adherence.[3] Of course, taking anticoagulants as directed (medication adherence) is critical to realizing the beneficial results seen in studies.[4] However, it is uncertain if the high rates of adherence seen in a clinical trial population will be realized in actual clinical practice. Specific to DOAC therapy, it is uncertain if patients with low time in the therapeutic range (TTR) and high levels of international normalized ratio (INR) variability while on vitamin K antagonist (VKA) therapy are better served by transitioning to a DOAC, in large part due to questions of medication adherence. There have been concerns that adherence and persistence may be lower in this group with mixed study results to date.[5] [6] [7] [8]

In this issue of Thrombosis and Haemostasis, Elling et al reported a retrospective cohort study of 437 patients anticoagulated for any indication, switching from VKAs to DOACs from 2012 to 2019.[9] Medication adherence was evaluated by proportion of days covered by an anticoagulant prescription. Of this cohort, nearly 70% had VKA TTRs under 70%. However, this was not associated with subsequent good DOAC adherence (defined as >90% adherence), which was high for most patients (∼80%). Poor INR control (low TTR, high time under the therapeutic range, or high INR variability) while on a VKA was also not associated with lower persistence on DOACs compared to patients with better INR control. The results suggest that poor INR control should not dissuade clinicians from transitioning patients from a VKA to a DOAC. However, the study has notable limitations, including the potential for confounding and misclassification.[9] These issues are common in studies relying on prescription claims or fill data. The study was also small and may not capture important subgroups. Finally, the reason for poor INR control is not well described.

Received: 08 May 2024

Accepted: 13 May 2024

Article published online:
01 June 2024

© 2024. Thieme. All rights reserved.

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany