2.3 Atrial fibrillation

Atrial fibrillation (AF, an abnormal heart rhythm with rapid and irregular beating) is estimated to increase the risk of stroke 3 to 5-fold and to be associated with around a quarter of all ischaemic strokes[117-119]. Additionally, AF is associated with more severe strokes leading to higher mortality and disability[120].

2% of patients attending the emergency room with AF had a stroke within 1 year in Western Europe and 4% in Eastern Europe, according to a large international cohort study[121]. The 2016 European Society of Cardiology (ESC) Guidelines for AF Management[122] recommend that anyone over 65 years or at high risk of stroke is screened for atrial fibrillation.

AF is often asymptomatic and screening is not routinely undertaken in Europe. Accurate information on its prevalence in the general population is therefore not widely available. Some European countries have undertaken screening studies to estimate the proportion of the population affected (Appendix 1, Table 2). They reported rates in the general population ranging from 1.3% (UK, Italy) to 3.9% (Greece), with rates being highly dependent on age. Less data is available from Central and Eastern European countries.

Screening studies also found that between 10% and 66% of people with AF were previously unknown cases (Belgium[95], Portugal[123], UK[124], Spain[125]). This implies a significant under-diagnosis in Europe. A UK trial showed that opportunistic screening with simple pulse palpation resulted in significantly improved detection rates[126]. A major screening study has been launched in Sweden to detect AF and to see whether screening reduces stroke incidence and is cost effective[127].

AF is significantly more common in people who have had a stroke than in the general population. Reported prevalence rates are as high as 31-38% in a Greek study (Appendix 1, Table 1). Variations are also high within countries and studies with similar methodologies (e.g. Italy[128], Appendix 1, Table 1).

AF is often only diagnosed after someone has had a stroke: studies from Ireland[129], Iceland[16], Croatia[130] and Norway[131] reported that between a quarter to over half of AF diagnoses known after stroke were unknown before. A recent metaanalysis also reported that 24% of stroke patients are newly diagnosed with AF after their stroke[132]. These reports again suggest a significant under-diagnosis of AF in the general population. Better detection rates of AF could lead to improved primary prevention.

Due to Europe’s ageing population and AF’s strong association with age (0.7% in those aged 55-59 vs. 17.8% in those aged ≥85 years,[133]), the prevalence of AF is expected to rise. Using data from the communitybased Rotterdam Study and population projections from Eurostat, it was estimated that the number of adults over 55 with AF will more than double between 2010 and 2060 from 8.8 million to 17.9 million[117]. An Icelandic study projected the prevalence of AF to rise from 1.9% in 2008 to 3.5% in 2050[134].

In the UK, the number of AF related ischaemic strokes has trebled in the last 25 years in adults over 80 years and is predicted to treble again by 2050, with AF-related embolic events costing the UK around £374 million per year[135].

AF is therefore an important part of European[136, 137] and most national stroke guidelines, both for primary and secondary prevention. However, some European countries have not developed national guidelines covering AF treatment in relation to stroke primary prevention (Czech Republic, Greece, Latvia, Estonia,[9])

The ESC Guideline recommends treating AF patients following a structured risk assessment with oral anticoagulants for those with high risk scores[136]. Several recent, international European studies have assessed treatments rates of AF patients with oral anticoagulants (Table 8).

Table 8: International oral anticoagulants treatment  studies




Countries included and treatment rates*

PREFER-AF study[138] 2012: France: 90%, Germany, Austria, Switzerland: 87%, Italy:

72%, Spain: 88%, UK: 78%, CHA2DS2-VASc score: score≥2:

86%, score=0: 63%

EORP-AF study[139, 140] 2012/13: Western Europe (Belgium, Denmark, Netherlands, Norway): 72%, Eastern Europe (Poland, Romania): 74.7%, Southern Europe (Greece, Italy): 76%, CHA2DS2-VASc>=2: East (93%, South: 95%, West 81%)
BALKAN-AF survey[141] 2014/15: Bulgaria: 73%, Croatia: 84%, Romania: 76%, Serbia: 79%. High treatment rates but poorly related to CHA2DS2

-VASc score: score≥2 74%, score=0 57%

Euro Heart Survey[142] 2003/4: 67% of eligible patients compared to 40-50% of low risk patients (according to CHA2DS2-score)


Global register, including Poland, Finland, Norway, Sweden, UK, Denmark, France, Germany, Netherlands, Belgium, Austria, Italy, Spain, Czech Republic, Hungary: overuse of OAC in low risk, underuse in high risk patients
RE-LY Atrial Fibrillation Registry[144] International study, AF patients attending ER: CHA2DS2-Vasc score≥2: 63% in Western Europe, <40% in Eastern Europe, proportion of INR values between 2.0 and 3.0: 67% in Western Europe, 59% in Eastern Europe
Gloria-AF[145] Global register, including Austria, Belgium, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, France, Germany, Greece, Ireland, Italy, Latvia, Netherlands, Norway, Poland, Portugal, Romania, Slovenia, Spain, Sweden, Switzerland, UK:

High treatment rates (90%), but over-use in low risk patients


*The CHA2DS2-VASc-score is a validated risk stratification scoring system, including congestive heart failure, hypertension, age, diabetes mellitus, stroke/TIA/ thromboembolism, vascular disease, and sex.

Generally, these studies found relatively high oral anticoagulant treatment rates, but treatment was often not in accordance with the guidelines and patients’ specific risk profiles. Under- use of oral anticoagulants in elderly patients[146] or those with high risk scores and over-use in those with low risk scores was observed.

Also, the majority of these studies recruited patients registered with cardiologists (PREFERAF, EORP-AF, BALKAN-AF, Euro Heart Survey, Gloria-AF). So the generally high treatment rates might not be representative for patients in primary care. Several smaller studies, particularly those using primary care data or data from screening studies, discovered much lower rates (Table 9). These figures might be more representative of real-life AF anticoagulant treatment. Again, a mismatch between treatment and patients’ risk scores was observed, showing poor adherence to guidelines.

Table 9: Oral anticoagulant treatment rates in national/regional studies




Oral anticoagulant treatment rates

Denmark 66%, with 76% treated according to guideline (2011 primary care data)[147]; 67% in 2015 for newly diagnosed AF patients (was 40-50% in 2010)[148]
Germany 71% treated according to guideline (2004-6 data)[149]
Greece 41% of eligible AF patients on oral anticoagulants, 34% on antiplatelets, 25% no therapy (screening study, rural Greece)[150];

>55% of intermediate risk and 67% of high risk AF patient not on oral anticoagulants[151]

Italy 84% at time of diagnosis (2% had low CHA2DS2-VASc score), but only 30% 2 years after diagnosis (2009-11 data, primary care)[152])
Poland 41% of eligible AF patients on oral anticoagulants (2006/10 data, cardiology/ internal medicine/ neurology wards)[153]
Spain 24% of AF pat >60 years with CHA2DS2-VASc≥2 not on oral anticoagulants, female gender, older age, cognitive impairment related to lower rates[154]
Sweden 53% of eligible AF patients on oral anticoagulants (2005-10 data)[155]
UK 53%, with 8% of very high risk patients no treatment, 38% of low risk patients on oral anticoagulants, lower treatment rates in elderly (2003 primary care data[156]
Some studies have investigated oral anticoagulant treatment rates for AF in a stroke population (Table 10). Before stroke, rates were found to be very low. After stroke, rates were slightly higher, but still low and varied significantly between studies and countries. Older age was a significant predictor of lower treatment rates[157]. There is a trend towards higher treatment rates in more recent studies.

Table 10: Anticoagulation rates in AF patients before and after stroke (primary and secondary prevention)





Before stroke Austria 16% patients diagnosed with AF prior to or at admission, 1999/2000 stroke register data[158]
Finland 55% patients with CHA2DS2-VASc-score ≥2, 49% in 2003 rising to 65% in 2012[159]
Ireland 39% 2015[129]
Poland 40% in 2010/3, compared to 6% in 1995/9[104]
Sweden 16% known AF, prior to ischaemic stroke, 2005-10 stroke register data[160]
UK 23% in 2011, compared to 12% in 1995[105]
  25% poor correlation with CHA2DS2-Vasc score (1999-2008
  46% primary care data[106]
    those in AF on admission[161]
After stroke Austria 33% at discharge in 1999/2000[158]
France 91% 2004-6, local stroke register data[162]
Germany 55% 2008/9 national audit data[163]
Ireland 84% 6 months after ischaemic stroke, 2011 data[116]
Italy 74% 2004-6, local stroke register data[162]
Lithuania 37% 2004-6, local stroke register data[162]
Poland 21% 2008/9 national audit data[163]
  22% 2004-6, local stroke register data[162]
Spain 59% pat with embolic infarction in 2009[164]
  23% 2004-6, local stroke register data[162]
Sweden 63% / in 18-64 / in>85a, 2005 data[157]
  9% within the first 3 months after stroke, 2005-10, national stroke
  35% register data[160]
  37% 2008/9 national audit data[163]
UK 34% 2008/9 national audit data[163]
  40% 2007-2012 local stroke register data; was 18% in 2004-6[162]


The spread of novel oral anticoagulants (NOAC) (now known as non-vitamin K antagonist oral anti-coagulation) will overcome some obstacles to the use of traditional oral anticoagulants (the need for frequent monitoring, for example), and might improve treatment rates. However, their uptake has been reported to be slow[138]: use of NOAC in Italy: <1%, UK: 4%, France: 6%, Spain: 11%, Germany: 12%. The National Health Service in England reported that most areas had a NOAC uptake below 20% of oral anticoagulants with wide variations (4%  to  69%)[165]. In Poland, 19% of anticoagulated AF patients discharged from cardiology were using NOAC[166]. More encouragingly, a German study of four tertiary stroke centres reported that half of ischaemic stroke patients discharged with anticoagulation were prescribed NOAC[167] and recent European data from GLORIA-AF show that 52.3% of all oral anticoagulant prescriptions for newly diagnosed AF patients were NOACs[145]. A recent Danish study found 73% of all prescribed oral anticoagulants being NOAC in 2015[148].

In conclusion, the prevalence of AF is expected to rise significantly over the next few decades due to an ageing population and there is evidence of severe under-diagnosis. Anticoagulation rates reported from large surveys using specialist cardiology set-ups are encouraging but studies using primary care data show much lower treatment rates, particularly in the older age group[156, 157]. Retrospective analyses of treatment rates in AF patients prior to stroke are lower still (16-39%).  Adherence to available treatment guidelines and the recommended use of risk stratification is still insufficient[168].

Treatment rates after stroke were higher, but still showed much room for improvement.