Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study

Eur Heart J. 2012 Jun;33(12):1500-10. doi: 10.1093/eurheartj/ehr488. Epub 2012 Jan 13.


Aims: The impact of some risk factors for stroke and bleeding, and the value of stroke and bleeding risk scores, in atrial fibrillation (AF), has been debated, as clinical trial cohorts have not adequately tested these. Our objective was to investigate risk factors for stroke and bleeding in AF, and application of the new CHA(2)DS(2)-VASc and HAS-BLED schemes for stroke and bleeding risk assessments, respectively.

Methods and results: We used the Swedish Atrial Fibrillation cohort study, a nationwide cohort study of 182 678 subjects with a diagnosis of AF at any Swedish hospital between 1 July 2005 and 31 December 2008, who were prospectively followed for an average of 1.5 years (260 000 years at risk). With the use of the National Swedish Drug Registry, all patients who used an oral anticoagulant anytime during follow-up were identified. Most of the analyses were made on a subset of 90 490 patients who never used anticoagulants. Risk factors for stroke, the composite thromboembolism endpoint (stroke, TIA, or systemic embolism), and bleeding, and the performance of published stroke and bleeding risk stratification schemes were investigated. On multivariable analysis, significant associations were found between the following 'new' risk factors and thromboembolic events; peripheral artery disease [hazard ratio (HR) 1.22 (95% CI 1.12-1.32)], 'vascular disease' [HR 1.14 (1.06-1.23)], prior myocardial infarction [HR 1.09 (1.03-1.15)], and female gender [HR 1.17 (1.11-1.22)]. Previous embolic events, intracranial haemorrhage (ICH), hypertension, diabetes, and renal failure were other independent predictors of the composite thromboembolism endpoint, while thyroid disease (or hyperthyroidism) was not an independent stroke risk factor. C-statistics for the composite thromboembolic endpoint with the CHADS(2) and CHA(2)DS(2)-VASc schemes were 0.66 (0.65-0.66) and 0.67 (0.67-0.68), respectively. On multivariable analysis, age, prior ischaemic stroke or thromboembolism, prior major bleeding events, and hypertension were significant predictors of ICH and major bleeding. Heart failure, diabetes, renal failure, liver disease, anaemia or platelet/coagulation defect, alcohol abuse, and cancer were other significant predictors for major bleeding, but not ICH. The ability for predicting ICH and major bleeding with both bleeding risk schemes (HEMORR(2)HAGES, HAS-BLED) were similar, with c-statistics of ~0.6.

Conclusion: Several independent risk factors (prior ICH, myocardial infarction, vascular disease, and renal failure) predict ischaemic stroke and/or the composite thromboembolism endpoint in AF, but thyroid disease (or hyperthyroidism) was not an independent risk factor for stroke. There is a better performance for CHA(2)DS(2)-VASc over CHADS(2) schemes for the composite thromboembolism endpoint. While both tested bleeding risk schemes have similar predictive value, the HAS-BLED score has the advantage of simplicity.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Aged
  • Anticoagulants / therapeutic use
  • Atrial Fibrillation / complications*
  • Atrial Fibrillation / drug therapy
  • Atrial Fibrillation / epidemiology
  • Cohort Studies
  • Female
  • Hemorrhage / epidemiology
  • Hemorrhage / etiology*
  • Humans
  • Male
  • Middle Aged
  • Risk Assessment
  • Risk Factors
  • Stroke / epidemiology
  • Stroke / etiology*
  • Sweden / epidemiology
  • Thromboembolism / epidemiology
  • Thromboembolism / etiology


  • Anticoagulants