There is nothing simple about atrial fibrillation; it is a complicated, often overwhelming disease, both for patient and physician. One question that invariably comes up early on is the question of prophylactic anticoagulation for prevention of stroke. Who should receive anticoagulation? Which anticoagulant? How should anticoagulation be handled around the time of surgical procedures, or before and after ablation or cardioversion? How should anticoagulation be monitored? How should it be modified in patients with kidney or liver disease? Should anticoagulation be used in patients who have increased bleeding risks? Just the topic of anticoagulation in atrial fibrillation is overwhelming! Too much for a short blog post. We’ll have to narrow this down further. Let’s talk about using risk scores to decide who should be placed on anticoagulation therapy.
Atrial fibrillation risk scores were designed to assess stroke risk in patient populations with atrial fibrillation “without valvular heart disease.” I quoted that because “without valvular heart disease” is not well defined for this purpose. Certainly these risk scores don’t apply to patients with prosthetic heart valves, or with rheumatic mitral stenosis, but beyond that in practice these scores seem to be used even in patients with mild to moderate non-rheumatic valvular disease. The CHADS2 score is very simple, but has become passé in recent years. It is too gross a measure; people with low scores can still be at significant risk for stroke. It has been replaced by the CHA2DS2-VASc score in recently published guidelines. This score makes it much harder to achieve a score of 0 and escape anticoagulation. Using this risk score, both the 2012 European Society of Cardiology (ESC) and 2014 American Heart Association/American College of Cardiology/Heart Rhythm Society (AHA/ACC/HRS) atrial fibrillation guidelines recommend no anticoagulation if the score is zero, and full anticoagulation if it is 2 or greater. Where there is some hesitation, if not disagreement, is when the CHA2DS2-VASc score is 1. Anticoagulate or not? The previous iteration of the guidelines leaned strongly toward anticoagulation for a CHA2DS2-VASc score of 1. The latest sets of guidelines are more equivocal. How to handle a score of 1 is particularly important when one realizes that female sex, on its own, is a risk factor in CHA2DS2-VASc with a point value of 1. Yes, half the people on the planet are born with a CHA2DS2-VASc score of 1 and by the old guidelines would require anticoagulation just on the basis of their sex.
A Swedish study published in 2012 sheds some light on this issue. The study concluded that, while female sex is a risk factor for stroke in atrial fibrillation if other risk factors are present, by itself, in women less than 65 years old without other risk factors, female sex does not confer a significant risk of stroke. The implication is that a CHA2DS2-VASc score of 1 that is only due to female sex does not warrant anticoagulation.
The results of this study were directly incorporated into the 2012 ESC guidelines (I note that Dr. Gregory Lip is a coauthor of both these guidelines and the Swedish study). Thus the recommendation by the ESC is full anticoagulation (aspirin and aspirin + clopidogrel are relegated to remote second-line therapy) for CHA2DS2-VASc score of 1 or higher, after excluding females with no other risk factors and age < 65 years, who (as with men with the same criteria) do not need anticoagulation.
The AHA/ACC/HRS 2014 atrial fibrillation guidelines are more vague than the ESC guidelines when the CHA2DS2-VASc score precisely equals 1. Cardiology guidelines are presented using a sort of quantified equivocation, with recommendations classified as I (should do it), IIa (reasonable to do it), IIb (you can consider doing it) or III (don’t do it). Not quite orthogonal, there are 3 levels of certainty as well: A (data derived from multiple randomized clinical trials), B (data from one randomized clinical trial), or C (“expert” opinion). Given this, it is interesting that anticoagulation for a CHA2DS2-VASc score of 2 or more is a class I, A level of evidence recommendation, and no anticoagulation for a score of 0 is a class IIa, B level of evidence recommendation. For a CHA2DS2-VASc score of 1 there is complete equivocation, with the following class IIb recommendation:
For patients with nonvalvular AF and a CHA2DS2-VASc score of 1, no antithrombotic therapy or treatment with an oral anticoagulant or aspirin may be considered. (Level of Evidence: C)
Addressing the possibility of a exclusion for females with a CHA2DS2-VASc score of 1, the guidelines state (again equivocating):
"In a study of Swedish patients with nonvalvular AF, women again had a moderately increased stroke risk compared with men; however, women younger than 65 years of age and without other AF risk factors had a low risk for stroke, and it was concluded that anticoagulant treatment was not required. However, the continued evolution of AF-related thromboembolic risk evaluation is needed."
This all creates a problem for physicians, patients (females especially), and also for the physician-programmer writing an app such as EP Mobile that calculates these risk scores and attempts to make recommendations. At present EP Mobile simply uses the old recommendations, as do most of the web-based online risk score calculators I surveyed (e.g. here and here). A user of EP Mobile pointed out to me that its recommendations are out of date. Trying to fit such complexity into a small dialog box on a smartphone screen is challenging. Nevertheless I will be updating the app so that its anticoagulation recommendations more precisely match current guidelines — at least until the next set of guidelines comes out.