News

New risk scheme to predict stroke in AF

Major Finding: With the new stroke prediction scheme, 46% of AF patients were classified as having an extremely low (less than 1%) annual risk of developing ischemic stroke.

Data Source: A study in which the ATRIA stroke risk assessment scheme was created using a derivation cohort involving 13,559 California adults with AF, and the scheme’s accuracy was then verified in a validation cohort involving 33,247 such patients.

Disclosures: This study was supported by the National Institute on Aging; the National Heart, Lung, and Blood Institute; and the Eliot B. and Edith C. Shoolman Fund of Massachusetts General Hospital. Dr. Singer reported ties to Bayer Healthcare, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Johnson & Johnson, and Pfizer.


 

FROM THE JOURNAL OF THE AMERICAN HEART ASSOCIATION

A new scheme for assessing stroke risk, based on data amassed in the Anticoagulation and Risk Factors in AF, or ATRIA, cohort, performed better than the schemes currently recommended in leading clinical guidelines, according to a report published online in the Journal of the American Heart Association.

In a derivation study in which the ATRIA scheme was developed and a separate validation study in which its accuracy was confirmed, the ATRIA scores identified a substantially larger proportion of atrial fibrillation patients – 46% – as being at low risk for stroke, compared with other risk schemes currently in widespread use. This would allow clinicians to consider forgoing anticoagulant therapy in nearly half of AF patients.

The new scheme was particularly useful in calculating stroke risk in primary prevention patients, "the large group whose stroke risk is the most uncertain and where personalizing the anticoagulation decision is most pressing," said Dr. Daniel E. Singer of the clinical epidemiology unit, Massachusetts General Hospital, Boston, and his associates.

In addition, the ATRIA scores were especially good at predicting severe strokes, "the category of stroke that we are most interested in avoiding," the investigators noted.

Dr. Singer and his colleagues first used large health-plan databases to identify 13,559 California adults diagnosed as having nonvalvular AF in 1996-1997 and followed through 2003. These study subjects accounted for 33,497 person-years of observation on warfarin and 10,927 person-years of observation off warfarin.

There were 685 thromboembolic events, including 643 ischemic strokes, in this derivation cohort. Patient age and personal history of prior stroke were the two factors found to exert the greatest effect on future stroke risk.

From these data, the investigators identified eight highly predictive variables to incorporate into their risk assessment model: age, prior stroke, female sex, diabetes, heart failure, hypertension, proteinuria, and end-stage renal disease or an estimated glomerular filtration rate less than 45 mL/min per 1.73 m2.

This risk-prediction scheme differs from existing schemes primarily in that it uses a broader range of age categories; makes age, prior stroke, and their interaction the predominant risk factors to weigh into the calculation; and adds new high-risk factors such as female sex and renal dysfunction into the calculation.

When this ATRIA risk scheme was used, many more study subjects were accurately classified as low or high risk than with other schemes, the investigators said (J. Am. Heart Assoc. 2013; 2013[doi: 10.1161/?JAHA.113.000250]).

The ATRIA stroke risk score improved on the CHADS2 (congestive heart failure, hypertension, age at least 75 years, diabetes, stroke [doubled]) stroke risk score by 26%, primarily by correctly upgrading many patients from moderate-risk to high-risk categories. And it improved on the more recently identified CHA2DS2-VASc (congestive heart failure or left ventricular dysfunction, hypertension, age at least 75 years [doubled], diabetes, stroke [doubled]-vascular disease, age 65-74, sex category [female]) stroke risk score by 27%, exclusively by correctly downgrading many patients from high- or moderate-risk categories to the low-risk category.

The researchers then confirmed the accuracy of the ATRIA stroke risk scheme by testing it in a validation cohort of 33,247 adults who were newly diagnosed as having AF in 2006-2009. These study subjects accounted for 26,263 person-years off warfarin and 25,306 person-years on warfarin.

There were 496 thromboembolic events, including 466 ischemic strokes, in this validation cohort.

When the new ATRIA stroke risk scheme was used, the distribution of patients into low-, moderate-, and high-risk categories was "remarkably similar" to that in the derivation cohort. Similarly, the ATRIA scores were much more accurate than those in current standard use at determining which patients were at low risk and which were at high risk for stroke.

In particular, 46% of patients in both the derivation and validation cohorts were categorized by their ATRIA score as having a less than 1% per year risk of stroke. This designation would be extremely helpful to clinicians in deciding which patients can safely forgo anticoagulation therapy, Dr. Singer and his associates said.

Similarly, the ATRIA score was markedly better than the other two risk schemes at discriminating risk for severe, as compared with minor, stroke events. This would be very helpful to clinicians in deciding which patients are most in need of anticoagulation therapy, they said.

Recent research indicates that certain biomarkers now also appear to be promising predictors of stroke in AF patients. If this proves to be true, such biomarkers can easily be added into the ATRIA scoring scheme to further improve stroke risk assessment, the investigators added.

This study was supported by the National Institute on Aging; the National Heart, Lung, and Blood Institute; and the Eliot B. and Edith C. Shoolman Fund of Massachusetts General Hospital. Dr. Singer reported ties to Bayer Healthcare, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Johnson & Johnson, and Pfizer.

Next Article: