High Prevalence of Atrial Fibrillation Among Patients With Ischemic Stroke
Background and Purpose—Atrial fibrillation (AF) is a common cause of devastating but potentially preventable stroke. Estimates of the prevalence of AF among patients with stroke vary considerably because of difficulties in detection of intermittent, silent AF. Better recognition of AF in this patient group may help to identify and offer protection to individuals at risk. Our aim was to determine the nationwide prevalence of AF among patients with ischemic stroke, as well as their use of oral anticoagulation.
Methods—Cross-sectional study of unselected patients in cross-linked nationwide Swedish health registers. All 94 083 patients with a diagnosis of ischemic stroke in the nationwide stroke register Riks-Stroke between 2005 and 2010 were studied. Information about previously diagnosed AF, and comorbidity, was obtained from the nationwide Patient Register and cross-referenced with the national Drug Register containing data on all dispensed pharmacological prescriptions in Sweden.
Results—Combination of data from Riks-Stroke and from the Patient Register showed that 31 428 (33.4%) patients with ischemic stroke had previously known, or newly diagnosed, AF. Of those, only 16.2% had received warfarin in a pharmacy within 6 months before stroke onset. After hospital discharge, only 35.0% of the survivors received warfarin within the first 3 months after discharge. The likelihood for underlying AF was strongly correlated to the CHA2DS2-VASC score, which is a point based scheme for assessment of stroke risk in AF but which also predicts likelihood of AF. In this scheme points are given for age, previous stroke or transient ischemic attack, hypertension, heart failure, diabetes, vascular disease and female sex.
Conclusions—Access to nationwide register data shows that AF is more common among patients with ischemic stroke than those previously reported. Few patients with stroke and AF had anticoagulant treatment before the event, and few got it after the event. CHA2DS2-VASc could be a useful monitoring tool to intensify efforts to diagnose AF among patients with cryptogenic stroke.
Independent of other risk factors, atrial fibrillation (AF) confers a 5- to 6-fold increased risk for ischemic stroke.1 The prevalence of AF increases markedly with increasing age2 and is more common among patients with concomitant cardiovascular disease.3 The prevalence of diagnosed AF in Sweden is currently 3.2% of the adult population (≥20 years).2 Estimates of the prevalence of AF among patients with ischemic stroke vary between 15% and 38%4–8 and the true proportion may be even higher because of difficulties in detection of intermittent, silent AF.9–11 Therefore, the effect of AF on stroke occurrence may not be fully appreciated. A recent meta-analysis of 5038 patients in 32 studies reported that, on average, 11% of patients with ischemic stroke have newly detected AF when examined by cardiac rhythm monitoring.12 Most of these studies have, however, been limited by small sample sizes and selected inclusion of patients.12
In the present study, we used a different approach to estimate the prevalence of AF in unselected patients with acute ischemic stroke. Using record linkage, we combined clinical information in a nationwide stroke register with AF diagnoses recorded in national administrative registers before and after the index stroke event. This permits access to more complete data on the occurrence AF in ≥94 000 patients with ischemic stroke, thus providing estimates of AF prevalence at a population level.
The aim of the study is to determine the nationwide prevalence of previously known or newly detected AF among patients with ischemic stroke.
The study population consists of all patients in the nationwide Swedish stroke register Riks-Stroke between July 1, 2005, and July 1, 2010. Only ischemic strokes were considered, transient ischemic attack was not. In patients with ≥1 stroke event during this period, only the first event was counted. Additional information about each patient was obtained by cross-matching data, using individual civic registration numbers, with information from nationwide Swedish registers: the Patient Register, the Dispensed Drug Register, and the Population Register.
Riks-Stroke13 carries detailed information about ≈96% of all stroke events that result in a hospital contact in Sweden.14 Riks-Stroke provided information about type of stroke and if there was previously known or newly diagnosed AF. A validation study of Riks-Stroke has previously shown that diagnoses of AF in Riks-Stroke was accurate, as adjudicated by scrutiny of a sample of individual medical records.15
The information in Riks-Stroke was complemented by data from the national Patient Register, which carries detailed information about hospitalizations and visits to hospital-affiliated open clinics, including dates and diagnoses, for every contact in Sweden. For description of previous and current disease, we used diagnoses given after 1997 when the current version of the International Code of Diseases-Tenth Revision (ICD-10) was introduced in Sweden. The codes included are listed in Table I in the online-only Data Supplement. A diagnosis of AF at hospital discharge is correct in 97% of cases according to a validation study.16 For most other diagnoses, the positive predictive values are in the range 85% to 95%.17 Although there are different subcodes for paroxysmal and permanent AF, these codes are rarely used. Therefore, we report the prevalence of all types of AF as diagnosed clinically by physicians all over the country.
The prevalence of AF among patients with stroke was analyzed in the relation to CHA2DS2-VASc and HAS-BLED scores before stroke onset. CHA2DS2-VASc18 score is a widely used point-based scoring system, originally established to estimate individual risk for stroke among patients with AF where points are given for age, previous stroke, heart failure, hypertension, diabetes mellitus, vascular disease, and female sex.
HAS-BLED19 assesses the risk of bleeding with oral anticoagulation for patients with AF. It is based on 9 clinical indicators: age, hypertension, renal, disease, liver disease, stroke history, previous bleeds, medication, alcohol use, and labile international normalized ratio (ie, a measure of the quality of anticoagulant treatment with vitamin K antagonists). We did not have data about quality of anticoagulant treatment and could, therefore, not count points for labile international normalized ratios and, therefore, calculated a modified HAS-BLED score.
Dispensed Drug Register
Information about the use of anticoagulants and other medication was obtained from the Dispensed Drug Register, which stores details about every prescription that is handled in every pharmacy in Sweden since July 1, 2005. The register is ≈100% complete because information is electronically recorded automatically whenever a drug is dispensed. The only registered oral anticoagulant in Sweden during the study period was warfarin, with phenprocoumon as an alternative on special license for a small number of patients intolerant to warfarin.
The Swedish Population Register was used to obtain information about the date of death of patients.
Baseline characteristics are presented descriptively, and differences are tested with t test, Mann–Whitney U test and χ2 test as appropriate. P values <0.05 were considered significant. For the multivariable analyses, we used binary logistic regressions. The cofactors used are specified under each table. All analyses were performed in SPSS 22.0 (IBM SPSS Statistics; IBM Corporation, Somers, NY).
Approval for the study was obtained from the regional ethical committee in Stockholm (EPN 2010/852–31/3).
During the 5-year inclusion period, Riks-Stroke recorded 105 364 ischemic stroke events (ICD-10 code I63) among 94 645 unique individuals. Information about age and sex, derived from the civic registration number, was lacking for 562 patients without Swedish residency. Thus, 94 083 patients remained for the study. The mean age was 76.2 years, and 48.8% were women.
Prevalence of AF Among Patients With Ischemic Stroke
Among 94 083 patients with ischemic stroke, 26 913 (28.6%) had AF according to Riks-Stroke, which does not differentiate between previously known and newly diagnosed AF. According to the Patient Register, 20 797 (22.1%) patients had AF who were previously diagnosed, and 7623 (8.1%) additional patients had AF who were newly diagnosed. When data from both registers were combined, the number of patients with AF, previously known or newly diagnosed, was 31 428 (33.4%).
Age and Sex
Patients with ischemic stroke and AF were older than those without AF (80.8 versus 73.9 years; P<0.001; Table 1). The prevalence of AF increased with age from 8.6% at <60 years to ≥50% at >90 years (Figure 1). Ischemic stroke was more often AF related in women than in men (37.0% versus 29.9%; P<0.001), largely because of higher mean age among women than among men (78.8 versus 73.7 years; P<0.0001). After adjustment for cofactors, including age, the likelihood that stroke was AF related was 6% higher for women than for men (odds ratio, 1.06; 95% confidence interval,1.02–1.09).
Factors Associated With Underlying AF
Patients with stroke and AF had more comorbidity than patients with stroke and without AF (Table 1). Among patients with stroke and heart failure, valvular disease, or renal failure, the likelihood that there was underlying AF was ≥50% (Table 2).
The stroke risk score CHA2DS2-VASc, as estimated before the stroke event, showed a direct relationship with prevalence of underlying AF (Figure 2). With a score of ≥5 CHA2DS2-VASc points, 52.1% had underlying AF. The criterion age ≥75 years in combination with heart failure identified a large subgroup in which 70.1% had underlying AF (Table 2).
Multivariable analyses showed significant associations between the presence of AF and heart failure (odds ratio, 3.3), valvular disease (odds ratio, 2.03), hypertension (odds ratio, 1.50), and several other risk factors (Table 2).
Anticoagulants and AF-Related Ischemic Stroke
Of all 31 428 patients with AF and ischemic stroke, only 12.0% had redeemed a prescription for warfarin within 3 months before the stroke, and 16.2% had done so within 6 months. If the analysis was restricted to include only patients whose AF was known before the stroke event, and thus possibly could have been prevented if anticoagulation had been used, the proportion of patients who had received warfarin within the past 3 and 6 months was 16.2% and 22.0%, respectively.
After hospital discharge, 8248 of 23 586 (35.0%) surviving patients AF and ischemic stroke received warfarin from a pharmacy within 3 months. The proportion of patients who received warfarin was higher among patients with a first ever stroke than among patients with recurrent stroke (36.4% versus 28.2%; P<0.001). The higher the risk for new ischemic stroke, according to CHA2DS2-VASc score, the lower was the likelihood to receive warfarin (Figure 3). There was a similar, inverse relationship between age and likelihood of warfarin treatment; 63.8% of patients aged <60 years received warfarin within 3 months, but only 6.6% of patients aged >90 years. The higher the bleeding risk, as assessed by HAS-BLED, the lower was the likelihood of warfarin treatment after ischemic stroke (Figure I in the online-only Data Supplement). Other factors associated with absence of warfarin were female sex, dementia, diabetes mellitus, renal failure, smoking, and alcohol abuse (Table 3).
By adding information from administrative registers to a nationwide stroke register, we have shown that AF is considerably more common in unselected patients with ischemic stroke than what has been reported in most previous population-based studies.1,6,20–24 Swedish health registers are nationwide and carry information about AF diagnosed at previous hospital contacts throughout the country. These registers are not limited to what is found in local medical records at hospitals where patients with acute stroke have been admitted. Nor do the registers rely on the ability of stroke patients to give correct and coherent information about their medical history in the acute setting.
The mean age of the patients in our study was 76.2 years, which is higher than in previous studies, where mean ages ranged between 65.9 and 74.9 years.20–24 Although the prevalence of AF among patients with stroke almost doubles between ages 65 and 75 years (Figure 1), it is natural that a study that includes more elderly patients also reaches a higher prevalence estimate. Furthermore, these other studies were made 10 to 20 years ago and life expectancy has continued to increase in Sweden and in many other countries since then. A study of cross-linked nationwide administrative registers may also, possibly, be more inclusive toward the older than studies with other designs.
The overall prevalence of AF among patients with ischemic stroke (33.4%) is lower than was found in a regional study in Western Sweden, where 38% (4565) ischemic strokes were AF related.8 In contrast to our study, the regional study also had access to information about AF that had been diagnosed in primary care.
AF may also pass unnoticed as silent AF. In a recent screening study of 75- and 76-year-old individuals in the general population using prolonged intermittent ECG monitoring, 49% more AF was found than had previously been known (increase from 9.6% to 14.3%).9
The true prevalence of AF among patients with ischemic stroke is, therefore, probably higher than our estimates. The likelihood for AF among patients with stroke was directly correlated with the CHA2DS2-VASc score. Therefore, the CHA2DS2-VASc scheme, although originally developed for another purpose, could also be potentially useful for identification of patients among whom screening for AF is likely to give high yield. With the criterion age ≥75 years plus heart failure, with ≥70% likelihood of AF, one could even contemplate interim anticoagulant treatment until careful examinations have ruled out silent AF. Prospective studies will have to clarify this question.
AF is one of the stroke risk factors for which there is available effective interventions to lower the risk. Such interventions presently seem to be underused. We found that only 16% of all patients with AF hospitalized for ischemic stroke had been treated by oral anticoagulants during the months preceding the event. The corresponding proportion reported from France 2000 to 2006 was 22%.23
In our study, only 35% of patients with ischemic stroke and AF had received warfarin from a pharmacy within the first 3 months after discharge. According to the Riks-Stroke annual report 2012, two thirds of patients with ischemic stroke and AF <80 years are prescribed or are planned to be prescribed anticoagulants at discharge from hospital.14 However, the Dispensed Drug Register shows that <50% actually received this treatment. One reason for this discrepancy may be that initiation of treatment is delegated to primary care physicians with the intent that a later initiation will lower the risk of hemorrhagic transformation. It is only natural if primary care physicians are reluctant to initiate treatment if they get the impression that the hospital specialists are so afraid of the bleeding risk that they do not want to do it themselves.
Disappointingly, the patients who had the highest risk for a new stroke had the lowest use of warfarin. This could reflect a hesitation to prescribe oral anticoagulation to patients with multiple risk factors, despite the fact that these patients may have the most benefit.
The low use of warfarin both before and after stroke must, however, be seen in the context of what was the recommended practice during the years 2005 to 2010 when the study was made. At that time, both international25 and Swedish26 guidelines listed aspirin as an option for patients with AF at moderate risk for stroke. New evidence has shown that the protective effect of aspirin is small, whereas the bleeding risk is similar to that of the more efficient alternatives and consequently both European and local Swedish guidelines no longer recommend aspirin for stroke protection in AF.27 Another reason for the low use of warfarin has to do with patient preferences; frequent international normalized ratio checkups may be too inconvenient for elderly or with functional disabilities. Some patients do not like warfarin because they (erroneously) think that warfarin makes it dangerous to eat vegetables, whereas others think that bleeding strokes are just as likely with warfarin, as embolic strokes are without warfarin, and therefore refuse treatment. Since the study was performed, new oral anticoagulants (dabigatran, rivaroxaban, and apixaban) have been introduced. This has given patients and doctors more alternatives to choose from, which increases the likelihood to find an acceptable, safe and efficient drug for secondary stroke prophylaxis.
Limitations and Strengths
The validity of data in Riks-Stroke is high and cover all hospitals in the country admitting patients with acute stroke.14 Because this is a total population study with nationwide covering, the number of patients with stroke is much larger than in previous studies.
Assessment of individual CHA2DS2-VASc scores from register data is likely to give lower scores than the true situation. The extent of underdiagnosis is largely unknown because assessment requires screening of the general population and, in addition, is likely to change over time.
Our information about the exposure to warfarin assumes that all patients who purchased warfarin also used it, which rarely is the case. The share of patients treated with warfarin before and after stroke events are, therefore, probably even lower than our estimates.
It has to be pointed out that not all patients with a previous AF-related stroke should have oral anticoagulation. For patients with severe bleeding risks, as well as patients in palliative care at the end of life, anticoagulation is of no benefit. Thus, the goal should not be that 100% of all patients with AF and ischemic stroke should have anticoagulation.
In patients with AF, an ischemic stroke is not necessarily cardioembolic. For example, hypertension and vascular disease are risk factors for development of AF just as they are risk factors for stroke. Clinically, this distinction is of minor importance, because all studies about stroke and stroke prophylaxis in AF have been done with real-world patients who often also have these other risk factors.
Access to nationwide register data shows that AF is more common among patients with ischemic stroke than previously reported. Few patients with stroke and AF had anticoagulant treatment before the event, and few received it after the event.
CHA2DS2-VASc could be a useful monitoring tool to intensify efforts to diagnose AF among patients with cryptogenic stroke.
Sources of Funding
The study was supported by the Swedish Heart and Lung Foundation, The Stockholm County Council, and the Board of Benevolence of the Swedish Order of Freemasons.
Outside of the present work Dr Friberg has received research grants or given lectures for Bayer, Boehringer Ingelheim, Bristol-Myers-Squibb, and St Jude Medical. Dr Rosenqvist has received research grants or given lectures for Bayer, Boehringer Ingelheim, Pfizer, Medtronic, Zenicor, and St Jude Medical. Dr Terént has received a research grant from AstraZeneca. Dr Lindgren has received honoraria for lectures or participated in advisory board meetings for Boehringer Ingelheim, Pfizer, Sanofi, and Bristol Myers Squibb. The other authors report no conflicts.
Guest Editor for this article was Jeffrey L. Saver, MD.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.006070/-/DC1.
- Received May 8, 2014.
- Revision received May 31, 2014.
- Accepted June 17, 2014.
- © 2014 American Heart Association, Inc.
- Stewart S,
- Hart CL,
- Hole DJ,
- McMurray JJ
- Asberg S,
- Henriksson KM,
- Farahmand B,
- Asplund K,
- Norrving B,
- Appelros P,
- et al
- Björck S,
- Palaszewski B,
- Friberg L,
- Bergfeldt L
- Engdahl J,
- Andersson L,
- Mirskaya M,
- Rosenqvist M
- Doliwa Sobocinski P,
- Anggårdh Rooth E,
- Frykman Kull V,
- von Arbin M,
- Wallén H,
- Rosenqvist M
- Kishore A,
- Vail A,
- Majid A,
- Dawson J,
- Lees KR,
- Tyrrell PJ,
- et al
- 14.↵Riks-Stroke. Annual report 2012 (in Swedish). http://www.riks-stroke.org/content/analyser/Riks-Strokes_Arsrapport%202012.pdf. Accessed February 12, 2014.
- Glader E-L
- Schneider AT,
- Kissela B,
- Woo D,
- Kleindorfer D,
- Alwell K,
- Miller R,
- et al
- Grau AJ,
- Weimar C,
- Buggle F,
- Heinrich A,
- Goertler M,
- Neumaier S,
- et al
- White H,
- Boden-Albala B,
- Wang C,
- Elkind MS,
- Rundek T,
- Wright CB,
- et al
- Fuster V,
- Ryden LE,
- Cannom DS,
- Crijns HJ,
- Curtis AB,
- Ellenbogen KA,
- et al
- 26.↵National Board of Health and Welfare. National gudielines for strioke care—summary. http://www.socialstyrelsen.se/nationalguidelines/nationalguidelinesforstrokecare. Accessed February 2, 2014.
- Camm AJ,
- Lip GY,
- De Caterina R,
- Savelieva I,
- Atar D,
- Hohnloser SH,
- et al