Anticoagulation in people with atrial fibrillation

BMJ 2011; 342 doi: http://dx.doi.org/10.1136/bmj.d530 (Published 31 January 2011) Cite this as: BMJ 2011;342:d530
  1. Margaret C Fang, associate professor of medicine
  1. 1Division of Hospital Medicine, University of California, San Francisco, CA 94143, USA
  1. mfang{at}medicine.ucsf.edu

Risk prediction tools help, but treatment must be tailored individually

A cornerstone of managing patients with atrial fibrillation is deciding whether or not to prescribe an anticoagulant. Despite strong evidence supporting the efficacy of anticoagulants in preventing thromboembolism related to atrial fibrillation,1 many people with atrial fibrillation—even those with multiple risk factors for stroke—are not prescribed these drugs.2 To help clinicians decide on which management path to choose, several tools have been developed to estimate the risk of stroke on the basis of clinical factors.3 4 The performance of two such risk tools is described in the linked cohort study by Olesen and colleagues (doi:10.1136/bmj.d124).5

One widely used risk scheme is the CHADS2 index, an acronym representing Congestive heart failure, Hypertension, Age over 75, Diabetes, and previous Stroke, which can easily be applied in general clinical practice.6 However, like other risk schemes, this index is only moderately accurate at separating patients into different categories of stroke risk.3 4 7 Large numbers of patients considered at high risk for stroke—in whom anticoagulants would be recommended—will not have a stroke. Conversely, some people deemed to have a low risk of stroke, who would therefore avoid anticoagulation, will have a stroke. Hence, more accurate ways of separating people who truly are at low risk from those who would benefit from anticoagulants are actively being sought.

Olesen and colleagues compared the performance of the CHADS2 index with the more recently developed CHA2DS2-VASc risk scheme by applying both schemes to a large nationwide registry of patients admitted to hospital with atrial fibrillation in Denmark.5 The CHA2DS2-VASc score differs from CHADS2 in that age is weighted differently. In addition, CHA2DS2-VASc considers female sex and a history of vascular disease to be significant risk factors for stroke.7

Although the C statistics, which measure a risk scheme’s ability to discriminate between people who will have an outcome and those who will not, were similar when the two risk schemes were tested as continuous point scales, CHA2DS2-VASc had unusually high C statistics when applied using categories of “low,” “intermediate,” and “high” risk groups. This might be related to the selected cut-off points of risk. The annual stroke rate in the CHA2DS2-VASc low risk group (0 points) was 0.78% at one year, compared with 1.67% in the CHADS2 low risk group. Although pulmonary embolism was included as an outcome and may have inflated the event rates, the core results did not seem to change when pulmonary embolism was removed from the analysis.

Should this study then encourage broader use of CHA2DS2-VASc? An expert task force of the European Society of Cardiology suggested that a CHA2DS2-VASc based approach to estimating stroke risk be used in patients with multiple “clinically relevant non-major” risk factors for stroke.8 Because age 65 or more and female sex are considered relevant risk factors, essentially the only people who meet the “truly low risk” category are men under 65. In Olesen and colleagues’ study, only 8.7% of the study population met the low risk CHA2DS2-VASc criteria, compared with 22.3% when using the CHADS2 criteria. This finding is similar to that seen in other assessments of the index, including a large UK study in which only 8.6% of patients with atrial fibrillation in general practice were considered low risk by CHA2DS2-VASc.4 If the CHA2DS2-VASc criteria were adopted most people with atrial fibrillation would be offered anticoagulation. In Olesen and colleagues’ study, 80% of the patients were considered high risk using CHA2DS2-VASc; fewer than half would meet high risk criteria if CHADS2 were used.

Should all women with atrial fibrillation be deemed at intermediate risk at least and therefore be considered for anticoagulation? Sex has an inconsistent association with stroke—some observational studies have found higher rates of atrial fibrillation related stroke in women,9 whereas others have not.4 A history of vascular disease has not been clearly established as a risk factor either.4

Nevertheless, CHA2DS2-VASc does seem to be better than CHADS2 at identifying those people who are at very low risk. An important consideration for future studies is whether the development of alternative treatments that are easier to use and potentially safer than vitamin K antagonists should induce clinicians to lower the threshold at which they prescribe anticoagulants. For example, the direct thrombin inhibitor dabigatran may result in fewer intracranial haemorrhages than treatment with warfarin in patients with atrial fibrillation.10 If anticoagulation becomes safer, it might be recommended for use in more patients with atrial fibrillation, particularly those at intermediate or even low risk of stroke. This is an area where consideration must be given to the patient’s preference, the balance between absolute benefit and risk, and the costs of care.11 Where these thresholds ought to be set is an area that needs further investigation.12


Cite this as: BMJ 2011;342:d530


  • Research, doi:10.1136/bmj.d124
  • Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declares: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Commissioned; not externally peer reviewed.