OBJECTIVES: To determine whether and how patient characteristics impact physicians’ decision to prescribe anticoagulants (AC) in elderly patients with non-valvular atrial fibrillation (NVAF), and the preference for type and dosing of AC.

METHODS: A clinical vignette study was conducted among cardiologists in the United Kingdom, Italy, and Germany. Attributes for vignettes were selected using literature, expert opinion and a survey. Seven patient characteristics were included in twelve vignettes and varied using an experimental design. Physicians were asked whether they would prescribe AC, which AC type to prescribe, and whether to make dose adjustments. Responses were analyzed using logistic regression.

RESULTS: 120 physicians completed the survey. AC was not prescribed in 20% of the vignettes in which it was recommended and in 24% where it should be considered. The likelihood of AC prescription decreased with higher age (OR=0.43; p<0.001), lower renal function (OR=1.54; p<0.001), dual antiplatelet therapy use (OR= 0.54; p<0.001), history of falls (OR=0.35; p<0.001), and history of minimal (OR=0.22; p<0.001), minor (OR=0.31; <0.001), and major (OR=0.04; p<0.001) bleeding. Underdosing was seen in 25% of the vignettes and overdosing in 30%. The likelihood of lower AC dosing increased with higher age (OR=1.35; p=0.02), lower renal function (OR=0.80; p<0.001), dual antiplatelet therapy use (OR=1.57; p=0.002), history of falls (OR=1.45; p=0.008), and history of minor (OR=1.69; p=0.006) and major (OR=2.01; p=0.001) bleeding.

CONCLUSIONS: There was variation in physicians’ therapeutic decisions regarding AC in elderly patients with NVAF. We identified underuse and under- and overdosing of AC. Some motivations for dose lowering were well-established, but others were not. Elderly patients with a complex profile are at higher risk of receiving a non-recommended dose of DOACs, which might be influenced by complicated dosing reduction criteria.