Dr John Pfeifer Outlines Necessary Steps for Preventive Action in Patients With Atrial Fibrillation

November 18, 2019

By identifying patients at very high risk for atrial fibrillation, preventive measures and screening strategies can be implemented for heightened patient care, said John Pfeifer, MD, MPH, cardiologist at Geisinger Medical Center.

By identifying patients at very high risk for atrial fibrillation, preventive measures and screening strategies can be implemented for heightened patient care, said John Pfeifer, MD, MPH, cardiologist at Geisinger Medical Center.

Transcript

Based on the study results showing that artificial intelligence (AI) predicted incident atrial fibrillation directly from electrocardiogram traces, can you discuss the significance of a deep neural network as an accurate predictor for preventive treatment of atrial fibrillation?

Atrial fibrillation is the most common abnormal heart rhythm. Some studies suggest that up to 1 in 3 people will develop atrial fibrillation at some point in their life. This is important because atrial fibrillation is associated with a number of bad outcomes—chief amongst them being stroke. Strokes can obviously be devastating–they can lead to death, they can lead to loss of function ability, permanent residence in a nursing home. So, we have a great interest in preventing strokes if we can. Now 1 of the tricky things about atrial fibrillation is it is often asymptomatic; patients do not know they have it. Even in those asymptomatic patients, they are still at risk for stroke. We know that we can reduce the risk of stroke in atrial fibrillation, but that requires medicines called anticoagulants. Those medications have risk associated with them, primarily bleeding. So, we don’t want to give them out indiscriminately–we need an accurate diagnosis of atrial fibrillation before we prescribe the anticoagulant. This is where our model really comes in.

People have been interested in screening atrial fibrillation for a long time, but we’ve not really figured out the right way to do it. Now, part of the problem is that even though the prevalence of atrial fibrillation increases as people age, the year-to-year incidence remains pretty low in most age groups. So, any screening strategy needs to take that into account. Furthermore, the other thing that makes atrial fibrillation really difficult is the fact that it’s often paroxysmal. People can have episodes that are relatively short-lived but still increase the risk of stroke. So, the question then becomes, how do we capture those patients? There’s lots of different monitoring strategies out there—we can check a patient’s pulse when they come for the doctor’s visit, we can do an EKG at a single point in time. More sophisticated strategies look at patches or monitors that people wear continuously for 2 to 3 weeks at a time. We even have devices that can implanted in the body which can stand for over a year for long-term monitoring of atrial fibrillation.

Now, the problem with any of those strategies is 1, they are costly—2, in implantable monitors and invasive procedures, although the risk is small, there is some risk associated with putting them in. So, 1 of the things that we want to do is identify the very high-risk populations and then target our screening strategies around them and that is exactly where our model comes into play. We identify the patients that are very high risk for the development of atrial fibrillation within the next 1 year. Now at that point, there are a number of screening strategies that we can employ in that population to try to find the afib [atrial fibrillation], prevent the stroke.