BACKGROUND: Direct-acting oral anticoagulants (DOACs) are recommended to reduce risk of stroke and systemic embolism in patients with non-valvular atrial fibrillation (NVAF). However, DOAC dosing inconsistent with FDA-approved product labels is common and associated with poor clinical outcomes.

OBJECTIVES: Identify DOAC dosing inconsistent with FDA-approved product labels in ambulatory care patients with NVAF; identify variables associated with dosing lower and higher than label.

DESIGN: Retrospective analysis using electronic health records from nine US healthcare systems.

PATIENTS: Adults with NVAF receiving DOAC therapy in 2022.

MAIN MEASURES: Rates of label-inconsistent dosing; multivariable regression analysis to identify demographic and clinical variables associated with dosing lower and higher than label.

KEY RESULTS: Among 51,128 NVAF patients (56.1% male, 94.3% White, mean [SD] age 73.5 [10.5] years), 5008 (9.8%) were prescribed label-inconsistent doses of DOACs (6.8% lower and 3.0% higher than label). Age ≥ 75 years, renal impairment, and hypertension were significantly associated with inconsistent dosing both higher and lower than label. Female sex and higher weight were significantly associated with dosing lower than label, as were heart failure, vascular or liver disease, and bleeding history. Dosing higher than label was significantly associated with male sex, race (African American/Black), weight < 60 kg, and use of drugs with potential drug-drug interactions. When prescribed by primary care physicians, DOAC doses were 37% (95% CI, 27-49%) more likely to be lower than label and 30% (95% CI, 16-46%) more likely to be higher than label than when prescribed by cardiologists or electrophysiologists. Label-inconsistent dosing varied (6.7 to 15.8%) across participating systems.

CONCLUSIONS: DOAC dosing inconsistent with label varied by demographics, clinical characteristics, prescriber specialty, and healthcare system, suggesting a need to monitor and assess dosing decisions in NVAF. Identification of variables associated with dosing inconsistencies may enable targeted interventions to ensure label-consistent dosing in vulnerable populations.