- Top of page
- Conflicts of interest
Abstract. Andersson P, Löndahl M, Abdon N-J, Terent A (Hudiksvall Hospital, Hudiksvall; Lund University, Lund; and Uppsala University, Uppsala; Sweden). The prevalence of atrial fibrillation in a geographically well-defined population in Northern Sweden: implications for anticoagulation prophylaxis. J Intern Med 2012; 272: 170–176.
Objectives. The aims of this study were to evaluate the community-based prevalence of atrial fibrillation (AF) in a western society using a geographically well-defined population in the northern part of Sweden as a reference and to estimate the proportion of patients eligible for oral anticoagulation (OAC) prophylactic therapy according to the stroke risk indices CHADS2 and CHA2DS2-VASc. Bleeding risk was assessed using the HAS-BLED score.
Design. The study population was recruited from AURICULA, a Swedish national quality register for patients receiving anticoagulation treatment. All patients with the diagnosis AF in the catchment area are registered in AURICULA.
Results. Of the 65 532 inhabitants in the catchment area, 1616 were diagnosed with AF (1200 cases were characterized as chronic AF). Thus, the overall prevalence of AF was 2.5%. The prevalence increased with age from 6.3% in patients over 55 years of age to 13.8% in those over 80 years. The prevalence was higher in men than in women in all age groups. Overall, 56.3% and 85.1% of the population were at high risk of stroke (≥2 points) according to CHADS2 and CHA2DS2-VASc, respectively. In addition, 26.9% had an increased bleeding risk according to HAS-BLED.
Conclusion. Within this large Caucasian population, we identified the highest community-based prevalence of AF to date. The prevalence was strongly associated with increasing age and male gender. Using CHA2DS2-VASc instead of CHADS2 widened the indication for OAC prophylactic therapy of AF in this population.
- Top of page
- Conflicts of interest
The study protocol was approved by the Regional Ethical Committee in Uppsala (application number Dnr 2008/090). Our study included all inhabitants of the Swedish county of Northern Hälsingland. According to the Swedish National Registry (Statistics Sweden), 65 532 people were living in this county on 31 December 2010. All OAC prescriptions in this catchment area are administered from Hudiksvall Hospital, the only hospital in the county, or from general practitioner centres via the web-based data system AURICULA , which is a Swedish national quality registry for AF. Patient characteristics, including risk factors according to CHADS2 and CHA2DS2-VASc, as well as OAC prescriptions were prospectively registered in AURICULA as part of daily clinical routine. International normalized ratio (INR) values were classified retrospectively as labile or stable according to HAS-BLED criteria in the majority of patients.
Through close collaboration with the general practitioners in the catchment area, as well as with colleagues at the hospital, patients identified with AF but not receiving OAC were also prospectively registered in AURICULA. To ascertain the accuracy of the register and to minimize the number of unregistered AF patients, data searches have also been conducted using the Central Diagnosis Register, and chart and ECG audits have been performed.
Study inclusion began on 28 March 2008 and concluded on 31 December 2010. Prevalence data are given for 31 December 2010, thus registered patients who died or moved out of the catchment area before this date are not included in the analysis. Diagnosis of AF was based on either 10-s 12-lead resting ECG or Holter registration. ECGs were interpreted by a general practitioner, an internist or a cardiologist. AF was defined as paroxysmal if spontaneous recovery occurred within 48 h. Atrial flutter was considered as AF.
Patients were classified according to the CHADS2 risk score at the time of inclusion. CHADS2 is a validated score system for stroke risk estimation  and is based on a scoring system in which two points are assigned for a history of stroke or transient ischaemic attack and one point for age ≥75 years, history of hypertension, diabetes mellitus or recent heart failure. A score of ≥2 is considered to identify patients at high risk of stroke. CHA2DS2-VASc includes the same risk factors as CHADS2 plus vascular disease, age between 65 and 74 years and female gender . CHA2DS2-VASc score ≥2 also identifies patients at high risk  but it was not used clinically during the study period.
The HAS-BLED score estimates the risk of bleeding on the basis of seven variables: hypertension, abnormal liver/renal function (1 or 2 points), stroke, bleeding, labile INR, age ≥65 years and abuse of drugs/alcohol (1 or 2 points). The maximum score is 9, and ≥3 indicates increased risk of bleeding after initiation of antithrombotic therapy . The scoring systems are shown in detail in Table 1. The data are given as mean ± standard deviation or frequency.
Table 1. Scoring according to CHADS2 , CHA2DS2-VASc  and HAS-BLED 
| ||Characteristics||Points|| ||Characteristics||Points|| ||Characteristic||Points|
|C||Cardiac failure||1||C||Cardiac failure||1||H||Hypertension||1|
|H||Hypertension||1||H||Hypertension||1||A||Abnormal liver/kidney||1 or 2|
|A||Age||1||A||Age ≥65 or ≥75||1 or 2||S||Stroke||1|
|D||Diabetes mellitus||1||D||Diabetes mellitus||1||B||Bleeding||1|
|S||TIA/stroke||2||S||TIA/stroke||2||L||Labile international normalized ratio||1|
| || || || ||Female gender||1||E||Age ≥65||1|
| || || || ||Vascular disease||1||D||Alcohol/drug use||1 or 2|
|Max score|| ||6|| || ||9|| || ||9|
- Top of page
- Conflicts of interest
On 31 December 2010, the prevalence of AF in the entire population in the county of Northern Hälsingland was 2.5%. AF was only identified in 2 of 24 243 inhabitants younger than 35 years. The overall prevalence in the rest of the population (≥35 years of age) was 3.9%. Chronic AF was more prevalent than paroxysmal AF, accounting for almost four-fifths (79.6%) of all cases (Fig. 1). The prevalence of AF increased with age (Table 2); 6.3% of the population aged 55 and above had AF compared to 13.8% of those above 80 years. The prevalence was higher in men than in women in all age groups (Fig. 2).
Table 2. Prevalence of atrial fibrillation (AF) at baseline by age and gender
|Age group (years)||All||Men||Women|
| n ||Cases of AF||Percentage of AF|| n ||Cases of AF||Percentage of AF|| n ||Cases of AF||Percentage of AF|
|≤34||24 243||2||0.0||12 678||2||0.0||11 565||0||0.0|
|All||65 532||1616||2.5||32 868||928||2.8||32 664||688||2.1|
The mean age of the 1616 (42.6% women) patients with AF was 74.3 ± 10.8 years. Hypertension was the most common concomitant medical condition (57.8%) followed by ischaemic heart disease (24.3%), congestive heart failure (20.0%) and cerebrovascular disease (17.7%). Diabetes mellitus was present in 16.3% of the population with AF. An angiotensin-converting enzyme inhibitor or an angiotensin II receptor blocker was prescribed to 52.6% and a statin to 35.1% of all patients with AF. Beta-blockers (74.4%) were the most commonly used rate-controlling medication followed by digoxin (20.0%). According to CHADS2, 43.7% of the population were at a low to intermediate risk (score 0/1) and 56.3% were at a high risk (score ≥2) of stroke; the corresponding rates for CHA2DS2-VASc were 14.9% and 85.1% (Fig. 3).
Warfarin was used by 63.7% of all AF patients, and by 55.2% and 41.2% of patients with a low stroke risk (grade 0) according to CHADS2 and CHA2DS2-VASc, respectively (Fig. 4). According to the HAS-BLED score, 26.9% of the AF population was classified as high risk (score ≥3). A high CHADS2 score was associated with a high risk of bleeding according to HAS-BLED (Fig. 4).
- Top of page
- Conflicts of interest
Within a large geographically well-defined Caucasian population, we found that the prevalence of AF was 2.5% in the whole population and 3.9% amongst those older than 34 years of age. The prevalence of AF was strongly associated with increasing age; more than 6% of individuals aged 55 years and older had AF compared to almost 14% in those older than 80 years. Men had a higher prevalence of AF at every age compared with women.
From previous reports, the prevalence of AF has ranged from 0.9% (≥50 years of age)  to 5.5% (≥55 years of age)  in primary care-based studies or overall populations (Table 3). Several problems associated with the assessment of AF may cause these differences between studies. First, many patients are unaware of AF (or periods of AF) and therefore receive no medical care [16, 17]. Second, prevalence figures might be biased by denial of informed consent in trials, as well as by study inclusion and exclusion criteria that might exclude whole groups of patients because of, for example, age or sex [9, 18]. Therefore, assessment of AF prevalence ought to be geographically defined and based on ECG measurements, information from general practitioners and hospital charts.
Table 3. Review of published reports of prevalence of atrial fibrillation
|Author||Published||Location||Population Age|| n ||Women (%)||Prevalence (%)|
|Ref. study||This study||Ref. study||This study||Ref. study||This study|
|Schmutz ||2010||Geneva, Switzerland||≥50||3 285||29 048||48||52||Overall: 0.9 m: 1.3; w: 0.4||Overall: 5.4 m: 6.4; w: 4.5|
|Heeringa ||2006||Rotterdam, Netherlands||≥55||6 808||24 678||59||52||Overall: 5.5 m: 6.0; w: 5.1||Overall: 6.3 m: 7.4; w: 5.2|
|Furberg ||1994||USA||≥65||5 201||14 737||57||54||Overall: 5.4 m: 6.2; w: 4.8||Overall: 8.9 m: 10.3; w: 7.7|
|Hobbs ||2005||W Midlands, UK||≥65||14 781||14 737||57||54||Overall: 7.2||Overall: 8.9 m: 10.3; w: 7.7|
|Murphy ||2007||Glasgow, Scotland||All ages||362 155||65 532||51||49||Overall: 0.9 m: 0.9; w: 0.8||Overall: 2.5 m: 2.8; w: 2.1|
In the Framingham study, the prevalence of AF more than doubled between 1968 and 1989 in patients aged 75–84 years of age . As compared to the 1985–1989 cohort in the Framingham study, our age-matched prevalence is significantly higher (12.7% vs. 7.4%), thus indicating a further increment in the presence of AF in the elderly population.
In a recent large (n = 362 155) cross-sectional general practitioner study conducted in Scotland , using a centrally administered morbidity register, the reported overall AF prevalence of 0.9% was considerably lower than in the present study. This difference is not related to differences in age, as we found that the prevalence was higher in every age group compared to the Scottish study.
The overall prevalence of AF in people older than 55 years of age living in a suburb of Rotterdam was reported to be 5.5% , which is more similar to the finding of the present study. Interestingly, we found that the prevalence of AF was higher in younger patients but lower in the oldest age group of patients compared with the Dutch study. This might be owing to differences in study design. In the Dutch study, ECG and health history were recorded for all study participants, whereas only patients admitted to the healthcare organization were included in our study. Consequently, in the present study, the presence of AF might have been missed particularly in elderly people living in assisted care homes. Furthermore, in the Dutch study only 78% of all inhabitants in the area consented to study participation, which might have had an impact on prevalence figures in general and in young patients in particular. However, despite using a more extensive nominator than in the Dutch study (number of inhabitants vs. number of study participants), our reported prevalence of AF was not lower.
The similarity in outcome between these two studies with their different methodological designs strengthens the probability that they represent the actual prevalence of AF in western society.
The risk stratification index used in the present study to predict stroke in patients with AF was CHADS2 , which is the current standard clinical index. According to this standard, patients with a CHADS2 score ≥2 are classified as at high risk of stroke, and guidelines recommend OAC treatment for these patients . However, there is increasing evidence that risk factors other than heart failure, hypertension, age ≥75 years, diabetes and prior transient ischaemic attack/stroke should be considered in refining stroke risk stratification for AF [22–27]. In the latest stroke risk index, CHA2DS2-VASc, female gender, age 65–75 years and presence of vascular disease are also included in the thromboembolic risk estimation. CHA2DS2-VASc is accepted as the risk stratification index in the latest European guidelines . Applying CHA2DS2-VASc instead of CHADS2 in our population increases the number of individuals who are at high risk of stroke and therefore direct candidates for OAC from 56.3% to 85.1%. The observed rate of OAC use was 63.7% in our AF population. A similar prescription rate has been reported from the AFFECTS Registry describing warfarin usage amongst practising American cardiologists . Although these prescription rates reflect the estimated frequencies of patients at high risk of stroke according to CHADS2 stratification, warfarin use is not yet in line with current guidelines . Patients considered at low risk of stroke appear to have been overprescribed OAC treatment in several studies, including the present study, whereas a significant proportion of those at high risk according to CHADS2 classification did not receive warfarin . In the present study, patients waiting for cardioversion only constituted a minority of warfarin-treated patients in the lowest-risk group. Thus, treatment tradition seems to have a larger impact on warfarin prescription in these low-risk patients than newer risk-scoring models, such as CHADS2. In our post hoc analysis of bleeding risk according to HAS-BLED, a substantial number of patients with a high CHADS2 score also had a high bleeding risk. HAS-BLED was not used in our clinical setting, but most clinicians worldwide are aware of the major risk factors in this stratification scheme and probably include these factors in individual risk calculation, which might explain the fact that CHADS2 alone had little effect on prescription rate (Fig. 4a). Approximately a quarter of our patients had an increased bleeding risk according to the HAS-BLED score.
Some limitations of this study should be considered. Patients who receive medical care in Northern Hälsingland have been screened for AF during the study inclusion period, and through careful auditing we estimate that the number of missed patients in this population is very low. However, not all inhabitants of Northern Hälsingland have been screened for AF and thus, although a very high prevalence rate has been reported, patients with asymptomatic AF or those who have not sought healthcare for other reasons might not have been included in our prevalence estimation. In particular, elderly people living in assisted care homes might have been missed.
CHADS2 and CHA2DS2-VASc risk score calculations were prospectively registered but only CHADS2 data were calculated prospectively and applied in clinical practice. Data regarding liver function and INR stability were registered retrospectively, and HAS-BLED scores were calculated post hoc using the recorded medical history data. As such the HAS-BLED risk assessment may have been underestimated.
In conclusion, the prevalence of AF in this large population-based study from a geographically well-defined area of northern Sweden was 2.5%. AF was rarely seen in individuals below 35 years of age. The prevalence of AF was strongly associated with increasing age (6.3% in individuals of 55 years and older, and 13.8% in those above 80 years) and men had a higher prevalence at every age compared with women. The new stroke risk index, CHA2DS2-VASc, increases the number of patients regarded as high risk compared to CHADS2 and therefore markedly widens the indication for OAC prophylactic therapy of AF.