Whether to anticoagulate: Toward a more reasoned approach

BETTER RISK PREDICTORS ARE NEEDED

Prediction tools are available for estimating the risk of stroke in patients with atrial fibrillation without anticoagulation^2,3 and to estimate bleeding risk from anticoagulation^4–7 (Table 1). Both tools have limitations, but as Hagerty and Rich point out, the stroke risk scales are likely better than the bleeding risk scales.

For example, Fang et al⁸ note that the risk of intracranial hemorrhage increases significantly after age 85. The bleeding risk scales use lower age cutoffs than this, perhaps increasing their sensitivity but decreasing their specificity.

Although HAS-BLED5,6 includes anti-platelet drugs such as nonsteroidal anti-inflammatory drugs and aspirin as risk factors for bleeding, ATRIA⁴ and HEMORR² HAGES⁷ do not.

Other drugs such as macrolides, quinolones, and high-dose corticosteroids raise the international normalized ratio (INR). These are typically used short-term, but can cause major fluctuations in the INR that may not be detected by monthly INR checks. Incorporating the short-term use of such drugs into bleeding risk scales would be difficult if not impossible a priori. Yet clinicians should be aware that these drugs can affect bleeding risk.

As Hagerty and Rich note,¹ the bleeding risk scores were developed for warfarin, and their applicability to patients treated with novel oral anticoagulants is uncertain.

All three of the available bleeding risk scales consider prior bleeding as a risk factor, but the severity of the prior bleeding varies. Although it is understandable to include major bleeding as a risk factor since it carries an increased risk of death, minor bleeding can affect morbidity and quality of life. Only the ATRIA score⁴ considers both major and minor bleeding, while HEMORR₂HAGES⁷ does not specify bleeding severity, and HAS-BLED^5,6 considers only major bleeding. Clearly, there is a need to update these existing bleeding risk scores so that they can apply to novel oral anticoagulants and consider both major and minor bleeding.

As the authors note, for predicting the risk of stroke, the CHA₂DS₂-VASc score³ provides more precision than the CHADS₂ score² at the lower end of the benefit spectrum. Unfortunately, there is no similar screening tool to predict bleeding risk from anticoagulation with greater precision in the middle to lower part of the risk spectrum.

THE PATIENT’S PREFERENCES MATTER

The patient’s life expectancy and personal preferences are important independent factors that affect the decision of whether to anticoagulate or not. It is the responsibility of clinicians who care for older adults to make sure that these two important considerations are included in any anticoagulation decision-making for this group of patients.