Atrial fibrillation (AF) is the most common cardiac arrhythmia with estimated prevalence, in developed countries, of 1.5–2%, rising up to 4–5% in people over 65 years old.1 In Spain, a large population-based study found a prevalence of 4.4% in subjects older than 40 years old,2 whereas another study in primary care reported a prevalence of 6.1%.3 AF associated to high morbidity multiplies the stroke risk by 5 and the overall mortality risk by 2.4 Furthermore, cardioembolic stroke associated with AF have lower survival rates.5

Oral anticoagulation (OAC) therapy is prescribed to prevent arterial thromboembolism in AF patients and has shown to reduce the risk of stroke by about 60%.6 In non valvular AF (NVAF) patients, direct oral anticoagulant drugs (DOACs) have shown a similar or even higher rates of thromboembolic risk reduction than classical OAC7–9 and antiplatelet drugs with less bleeding risk.10 Whereas the 2012 European Society of Cardiology (ESC) guidelines only considered antiplatelet monotherapy for stroke prevention in AF patients who refused OAC or could not tolerate anticoagulants for reasons unrelated to bleeding,1,11 the 2016 ESC guidelines1,11 do not recommend antiplatelet monotherapy for stroke prevention in AF, regardless of stroke risk. Instead, 2016 ESC guidelines recommend that NVAF patients with intermediate to high embolic risk (CHADS2 score≥2 or CHA2DS2-VASc≥2) should receive antithrombotic treatment with OAC unless contraindicated, being DOACs the preferred option over Vitamin K antagonists (VKAs). NICE guidelines recommends to treat AF patients and CHA2DS2-VASC stroke risk score of ≥2 with new anticoagulants instead of aspirin.12 However, in the clinical practice there is an underuse of OAC in these patients with an elevated risk of stroke.13 In Spain, the FIATE study14 reported an utilization of 84.1% in the context of >2000 selected patients from primary care, decreasing to 40.3% in patients with non-permanent FA. In the Val-FAAP study,3 12.7% of patients with AF and CHADS2 score≥2 did not receive antithrombotic treatment and 19.3% were only treated with antiplatelet drugs.

The main objectives of the present study were: to describe demographic, clinical and treatment characteristics of the patients with NVAF and CHADS2≥2 not treated with OAC, as well as main reasons for the therapeutic decisions taken by the physicians, and prevalence of AF, NVAF and NVAF not treated with OAC in the primary care setting in Spain.

The present multicentre study allowed quantifying and characterizing the subset of patients with NVAF and CHADS2≥2 that did not receive OAC in the primary care Spanish clinical practice in 2014–2015. The overall estimated prevalence of AF in our study was 4.5%, similar to the 4.4% reported by the OFRECE study2 and slightly lower than the 6.1% reported in the VAL-FAAP study.3 However, the OFRECE study included general Spanish population older than 40 years and to whom an ECG was performed whereas our study included patients older than 18 years already diagnosed of AF. Approximately 2 in 10 patients with NVAF were not receiving OAC, similarly than in the FIATE study.14 Also, they seem to reflect a lower under-use of OAC in Spain compared to other European countries as Germany (39%)20 or Sweden (36%),21 and similar under-use that in United States (23%)22 despite of different patients populations included in these studies (CHA2DS2-VASc>1 for Wilke et al., AF patients without risk assessment for Andersson et al., and CHA2DS2-VASc≥2 for Hess et al.). In Italy, only 27% of acute stroke patients with known AF and without contraindications had received OAC prior to the stroke event.23 In this regard, a recent published study from ambulatory care sites in the United States22 stated that most of the patients not receiving OAC had multiple risk factors and a Class I guideline24 indication for OAC. In line, the Spanish ACTUA study concluded that a considerable number of NVAF patients in Spain who could benefit from DOACs were not treated with them.25

Different reasons have been suggested to explain this under-use of OAC in patients with AF. One of the reasons for not using ACOs was the associated haemorrhagic risk. In many cases, it has been shown to be similar to that of using antiplatelet agents. In the Spanish AFABE study,26 conducted in patients older than 60 years, the most common reasons for not administering or for discontinuing OAC in patients with AF were: cognitive impairment (15.1%), risk of haemorrhage (14.3%) and patient refusal/preference (13.6%). For instance, some studies have claimed that limitations associated to treatment with antivitamin K agents, to thromboembolic risk infraestimation, and to bleeding risk concerns, especially with elderly population with AF or NVAF could explain the underutilization of OAC.22,27,28 Moreover, there is also a reluctance to use OAC due to the need for continuous monitoring, and frequent interactions with other drugs or foods.29

According to the ESC guidelines at the time of study conduction30 all patients with AF and risk factors for stroke should receive a classical OAC or a DOACs, and only those who refused OAC or could not tolerate them were candidates to antiplatelet monotherapy. This recommendation is now abolished in current guidelines11 where aspirin is not recommended. In the analyzed cohort of 1310 NVAF patients with a CHADS2≥2, 82.4% were receiving antiplatelet agents generally perceived as a “harmless treatment”. Not negligibly, 17.6% of patients were not receiving any antithrombotic treatment. In this subgroup, the main reason for the therapeutic decision was patient refusal to monitoring, a high bleeding risk, frequent falls, uncontrolled hypertension, difficult access to monitoring, cognitive impairment, and unstable INR. These reasons were similar to those reported in the already mentioned Spanish AFABE study26 as well as in the Swedish study by Andersson et al.21 In our sample of moderate-to-high risk of stroke, physicians considered that 15% of subjects were at low thromboembolic risk.

Patient refusal to the continuous monitoring needed with the classical OAC can be arranged by the use of DOACs. The AVERROES study has shown that apixaban was clearly superior to aspirin without significantly increasing major bleeding events,31 supporting an alternative to antiplatelet for those NVAF patients not suitable for OAC treatment. It is expected that the growing evidence of favourable profile of safety and effectiveness of DOACs vs. VKA and the increasingly use in the clinical practice will enhance the use of anticoagulant therapy in patients with NVAF.32 For instance, patients with SAMe-TT2R2 score>2 have been proposed as good candidates for receiving DOACs instead of VKA.33 Nevertheless, therapeutic decisions in AF patients must be accompanied by regular reviews and efforts to correct the haemorrhagic risk factors that are reversible.1,11 In the present study, at least half of subjects had a HAS-BLED score≥3, and there was a high prevalence of modifiable bleeding risk factors, suggesting that more efforts are needed in the clinical practice for controlling the blood pressure and reduce the concomitant use of drugs increasing bleeding risk, mainly NSAIDs. For instance, our study showed that labile INR was one of the main reasons for therapeutic management decision even when this risk factor could be easily remedied with DOACs use.

Strengths of our study include an important number of participating centres geographically distributed across Spain and the consecutive inclusion of all patients who met selection criteria ensuring the representativeness of our sample. Regarding limitations, as the study was observational, it did not allow evaluating a causal relationship between thromboembolic events and treatments. Despite the high number of PCP and centres participating in the study (approximately 8 in 10 accepted to participate), we cannot discard a selection bias. However, the sociodemographic characteristics of our cohort were similar to those from previous large cohorts of NVAF patients collected in our country, thus supporting the external validity of our study.14,26 Another limitation was the inability to individually check NVAF diagnosis for each patient included in the prevalence questionnaire and, therefore, to check if the number of patients provided by physicians was exact. At last, (1) only patients with AF diagnosis registered in the medical history of primary health care were included leading to a possible underdiagnosis and underreporting of patients, as well as (2) only patients with a CHADs score≥2 were able to participate in the study, excluding patients with only one risk factor.

Approximately 1 in 5 patients diagnosed with NVAF and associated embolic risk in Spain do not receive OAC in the clinical practice, which are characterized by a mean age of 75 years old, with comorbidities, and a high prevalence of modifiable risk factors. Their therapeutic management is inadequate, since most of them are receiving antiplatelet agents, despite the bleeding risk associated.

These findings emphasize the need for optimize the therapeutic management of NVAF patients, according to the most recent recommendations.11

JPG and DVO participated in the design of the study, data compilation, data analyses and interpretation, and the writing, reviewing and acceptance of the final version of the manuscript. FF, IU, SFC and JC participated in the design of the study, data analyses and interpretation, and the writing, reviewing and acceptance of the final version of the manuscript.

This study was sponsored by Bristol-Myers-Squibb and Pfizer.

José Chaves, Irune Unzueta and Susana Fernández de Cabo are Pfizer employees. The other authors do not have any conflict of interest.