Study Questions Atrial Fibrillation Stroke Rates Guiding Anticoagulation

Patrice Wendling

December 12, 2016

BOSTON, MA — Reported stroke rates cited in various guidelines to direct anticoagulation in lower-risk patients with atrial fibrillation vary widely even within the same cohort, a recent systematic review shows[1].

Among the 34 cohort studies analyzed, the annual North American stroke rate was less than one-third that of the European rate (mean 1.30% vs 4.14%; P<0.0001), reaching a miniscule, study low of 0.45% in the US Women's Health Initiative.

Stroke rates were also a third lower between prospective vs retrospective cohorts (1.22% vs 3.80; P<0.0001).

Among Asian cohorts, mean yearly stroke rates ranged from 1.18% in China's Yunnan Province to a study high of 9.38% in Hong Kong.

"These results call into question the generalizability of current leading guideline recommendations that CHA2DS2-VASc scores of 1 and 2 points should serve as the threshold for use of anticoagulant therapy," according to the investigators, led by Dr Gene R Quinn (Beth Israel Deaconess Medical Center, Boston, MA).

"I'm not trying to beat up the guideline writers; I was a guideline writer myself at one point. You try to produce something that is scientifically correct and easily applicable, and there are booby traps along the way. The problem is I think people didn't pay enough attention to the variability in the stroke rates that were reported," senior investigator Dr Daniel E Singer (Massachusetts General Hospital, Boston) told heartwire from Medscape.

The CHA2DS2-VASc scoring system is recommended in the European and the 2014 American Heart Association (AHA)/American College of Cardiology (ACC)/Heart Rhythm Society (HRS) guidelines to estimate stroke risk, with a yearly stroke risk of 1% to 2% generally agreed to yield a net clinical benefit with oral anticoagulant therapy.

The guidelines, however, cite stroke rates from only a few cohort studies of patients off anticoagulants and assume that CHA2DS2-VASc point-scores correspond to fixed stroke rates, Singer argues.

"For the AHA/ACC/HRS guidelines, the stroke rates highlighted are not actually observed rates, they are imputed; some people would say made up," he quipped.

Singer said US guideline writers were aware to some extent of the variability in stroke rates—acknowledging so only in a table footnote that actual stroke rates may vary from cited estimates—and worries physicians may not realize how uncertain the correspondence is between CHA2DS2-VASc scores and the risk of ischemic stroke.

Commenting to heartwire , Dr Gregory Lip (University of Birmingham, UK) said it's "common sense that event rates vary by study population and clinical setting as well as ethnicity. Even patients with a CHA2DS2-VASc score of 1 are a heterogeneous population."

He added, "The obsession with what stroke rate is associated with a particular score is simply unrealistic. The default should be to offer [oral anticoagulants] for stroke prevention unless they are shown to be truly low risk—for example, with a CHA2DS2-VASc score of 0 in men or 1 in women."

European guidelines recommend all AF patients with a CHA2DS2-VASc score of 1 in men or 2 in women be prescribed oral anticoagulation therapy.

CHA2DS2-VASc-Specific Analysis

The study, published recently in Circulation and discussed by columnist Dr John Mandrola, includes a second analysis of 17 of the 34 studies that reported stroke rates based on CHA2DS2-VASc scores. Singer said it provides particularly compelling evidence for the wide differences in stroke rates across cohorts, even those with strikingly similar demographic features. For example, the Swedish Atrial Fibrillation Study reported yearly stroke rates of 0.2% and 0.6% at CHA2DS2-VASc score 0 and 1, while rates were three times higher, 0.78% and 2.01% at the same scores, from the neighboring Danish National Patient Registry.

"To me this variation indicates that the differences in rates are largely due to differences in study methods rather than true differences in stroke rates," he said. "The choice of one table in the guidelines or highlighting one or two cohort studies in the ESC or AHA guidelines was to some extent cherry picking."

Commenting to heartwire , Dr Ethan Weiss (Cardiovascular Research Institute, University of California San Francisco) said the study shows that "there is a great deal of variability in stroke rates in these big cohorts."

But he added, "Without a good sense of what the actual bleeding rates, are it's hard to talk about net clinical benefit."

Weiss highlighted the AVERROES trial in which oral anticoagulant therapy with apixaban (Eliquis, Bristol-Myers Squibb/Pfizer) reduced absolute stroke rates from 3.4% with aspirin to 1.7% per year without increasing major bleeding in AF patients unsuitable for warfarin with an average CHADS2 score of 2.

"I don't think most of us in clinical practice look at the CHA2DS2-VASc score as like a tattoo that you are going to have a 2% stroke risk over the next year. I think we all see it as a tool, an imperfect tool," he said.

Singer said there's a scientific imperative to learn more about differences in stroke rates at the individual patient level, most likely through registries like ORBIT-AF and GARFIELD-AF rather than large administrative databases, and to have more shared decision making at lower CHA2DS2-VASc scores, maybe even going up to a score of 3.

"There's all this emphasis to anticoagulate patients at low CHA2DS2-VASc scores, but from a population perspective, the emphasis should be to make sure patients with higher CHA2DS2-VASc scores are identified and anticoagulated."

Quinn was supported by the Harvard Medical School Fellowship in Patient Safety and Quality and reports no relevant financial disclosures. Singer was supported in part by the Eliot B and Edith C Shoolman Fund of Massachusetts General Hospital. Singer reports consulting for Boehringer Ingelheim, Bristol-Myers Squibb, Johnson & Johnson, Medtronic, Merck, Pfizer, and St Jude Medical; and research support from Boehringer Ingelheim, Bristol-Myers Squibb, and Medtronic. Disclosures for the coauthors are listed in the paper. Weiss reports no relevant financial relationships.

Follow Patrice Wendling on Twitter: @pwendl. For more from, follow us on Twitter and Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.