Atrial fibrillation (AF) is a cardiac arrhythmia that is associated with a higher risk of thromboembolic events, e.g. stroke. The prevention of thromboembolic events relies on the initiation of antithrombotic therapy, including oral anticoagulants (OAC). According to guidelines, patients at high risk of thromboembolic event, as defined by a CHADS₂ score≥2, must receive anticoagulants. Until recently, vitamin K antagonists (VKA) were the only OAC available on the market. In 2012, the first direct oral anticoagulant (DOAC) entered the Belgian market. DOACs have several advantages over VKAs that could increase the use of OACs in these patients, and especially in older ones in whom anticoagulation is often underused. Our objectives were to (1) describe the evolution of the underuse of anticoagulants in older people with AF at high risk of thromboembolic event since the marketing of DOACs and (2) describe factors associated with this underuse.

We conducted a retrospective cross-sectional study including geriatric patients admitted during the pre-DOAC (2008–2011) and post-DOAC (2013–2015) periods in an academic hospital in Belgium. Five inclusion criteria were met: age≥75years, diagnosis of AF, indication for anticoagulation (CHADS₂ score≥2), risk of functional decline (Identification of Seniors at Risk [ISAR] score≥2) and comprehensive geriatric assessment. The use of anticoagulants and antiplatelets at home before admission was recorded. Risks of stroke and bleeding were calculated using CHADS₂ and HEMORR₂HAGES scores, respectively. Factors associated with the underuse of anticoagulants were assessed through a multivariable logistic regression. To assess if the pre-DOAC vs post-DOAC marketing period was associated with anticoagulation underuse, three different models were used. First, the multivariable logistic regression described above was used (model 1). The two other models included propensity score (PS) techniques to reduce bias due to confounding variables that could be associated with the time period and/or the underuse of anticoagulants. The second model (model 2) was a mixed-effect logistic regression with PS matching to take into account he matched nature of the data. Variables that were associated with the underuse of anticoagulation in the univariate logistic regression (first model), and those that are clinically relevant (e.g. bleeding risk, antiplatelet use) were introduced in a logistic regression model, with the time period being the outcome, to assess the PS for each patient. The last model (model 3) consisted of a multivariable logistic regression using the inverse probability of treatment weighting.

Anticoagulant underuse, present in 34% of geriatric patients with AF, was lower in patients with a history of stroke (OR (95%CI): 0.55 (0.35; 0.85), P=0.008) or congestive heart failure (OR (95%CI): 0.67 (0.37; 0.86), P=0.008) but higher in those receiving antiplatelets (OR [95%CI]: 8.45 [5.67; 12.85], P<0.001) and in older individuals (P=0.003). Anticoagulant underuse decreased significantly from the pre-DOAC (37.3%) to the post-DOAC (29.7%) era, as shown by the two analyses using PS. Indeed, time period 1 was significantly associated with OAC underuse in model 2 (OR [95%CI]=1.58 [1.03; 2.41], P=0.035) and model 3 (OR [95%CI]=1.34 (1.05; 1.70), P=0.017).

In older patients with AF, anticoagulant underuse was mainly associated with anti-platelet use. Anticoagulant underuse and antiplatelet use have both decreased since DOAC marketing. We cannot ascertain whether this was solely a DOAC effect or the effect of other factors, e.g. the increased awareness of clinicians of the importance of anticoagulation in these patients coming along with information campaigns and published updated guidelines that accompanied DOAC marketing. Underuse of anticoagulants was still a concern for three in ten geriatric patients with AF at high risk of thromboembolic event.