Who benefits most from screening for atrial fibrillation?

A new study finds insufficient evidence that AF screening helps more than it harms. It provides an immense opportunity to learn the limits of screening for disease

There was big news in the AF world: JAMA has published a major paper on screening for atrial fibrillation.

The 16-person United States Preventive Services Task Force (USPSTF) authored the review and concluded that the current evidence is insufficient to assess the balance of benefits and harms of screening for AF. USPSTF is a notable group because they are independent experts in critical appraisal. Think Neutral Martians not cardiologists looking at the evidence.

USPSTF assigns grades of preventive services on an A-B-C-D-I system. For example, A = high certainty of net benefit and doctors should offer or provide the service while D = high certainty of no net benefit and harms outweigh benefits thus doctors should not offer this service. ‘I’ means insufficient evidence.

My colleague Andrew Foy, from Penn State University, and I wrote an editorial on the review in JAMA-Internal Medicine. We mostly concurred with the USPSTF group, but emphasized that a major trial published after their review suggested that screening for AF actually gets closer to a D grade.

Wait a minute…

The idea that screening for AF does not improve outcomes seems counter-intuitive. Proponents of looking for AF in people without symptoms cite three facts: a) most AF is asymptomatic, b) AF increases the risk of stroke, and c) we have oral anticoagulants that reduce stroke.

It makes sense therefore that picking up AF before it causes a stroke would be a positive.

I italicized “makes sense” because those two words should always raise suspicion. It made sense to suppress PVCs after heart attack and it made sense to cool survivors of cardiac arrest. But when studied in proper trials, the sensical proved wrong.

So it goes with AF screening.

The USPSTF looked at 26 studies and found no convincing evidence that screening for AF reduced important outcomes. They did find that over short time periods screening will find more AF episodes and lead to more prescriptions for anticoagulants. But the evidence that this reduces stroke or death was not there.

Our editorial explains why AF screening does not deliver net benefits.

The super-important LOOP trial

We first reviewed the LOOP trial, which was published after the USPSTF review. LOOP represents a best-case scenario for screening.

LOOP compared AF screening in older adults with an implantable loop recorder to standard care. The loop recorder is an amazing AF detection device; it is always on, requires no action from the person, and detects AF with decent accuracy.

Indeed, in LOOP, the group with the recorder had three-fold more AF discovered and nearly three-fold more anticoagulant prescriptions. Yet all that extra AF discovered and anticoagulant prescribed did not lead to a statistically significant reduction in stroke.

LOOP Is important because implantable loop recorders are the best AF screening tool we have. And if you can’t show a reduction in stroke with this device, less robust devices would be unlikely to work.

I want to anticipate a criticism of my interpretation of LOOP. People who look closely at the results might say, hey Mandrola, the group who had the monitor actually had a 20% reduction in stroke. While it did not reach “statistical significance,” it was still a reduction. I would counter that by saying that major bleeding was 26% higher in the monitoring arm.

Should we be surprised that AF screening doesn’t work?

Foy and I argued that the lack of benefit seen in the LOOP is expected.

The first reason is that we don’t completely understand the relationship between AF and stroke. We wrote this:

Consider how AF screening compares with breast cancer screening. The goal of early detection of breast cancer is to reduce the death rate from breast cancer, and because it is a leading cause of death in women, to reduce overall mortality. Early AF detection is different because AF is but 1 cause of stroke. In fact, the majority of the strokes in patients with screen-detected AF may not be sensitive to anticoagulants.

Another reason why we were not surprised by LOOP is that we don’t know how much AF warrants oral anticoagulation. In the old days, anticoagulants were proven to have a net benefit (stroke reduction > bleeding increase) in patients with “clinical” AF—that is, AF that prompted a doctor’s visit.  

But now we have devices that can detect AF lasting seconds to minutes. We don’t know if treating these short-duration episodes will reduce stroke more than it increases bleeding. Fortunately, there are two ongoing randomized trials looking at this question.

Another reason AF screening could fail is the harms of downstream testing. A person with a smart watch who discovers an irregular heart beat may meet a calm wise doctor who does just the right amount of testing as well as counseling on reduction of AF risk factors, such as obesity, alcohol excess and sleep apnea. This would be a positive.

But the asymptomatic (previously healthy) person with screen-detected AF may also meet a doctor who orders a slew of unnecessary tests—any of which could lead to harm. Doing oodles of testing is sometimes warranted but it nearly always medicalizes a problem.

AF Screening is mostly out of the hands of doctors

While the USPSTF findings apply to AF screening by doctors, the reality is that rhythm monitoring is increasingly done with consumer devices, such as watches and smartphone apps. It does not matter what I or any other doctor thinks about AF screening. It’s happening whether we order it or not.

Which leads me to believe that the most important part of the USPSTF review is as a thinking exercise about the value of screening for disease. Here is what Foy and I concluded:

We will increasingly see patients present with the surrogate marker of screen-detected AF. As with any screening test, the question is whether this knowledge can guide management decisions that ultimately improve outcomes. The evidence to date suggests the answer is no.

It is crucial that people understand the limits of early detection of surrogate markers of disease—because the digital revolution will only lead to more detection of more dubious surrogate markers of disease.

The good news is that it is possible—likely even—that in the coming decade(s) many of the uncertainties today will become clear. These devices will teach us, for instance, how much irregular rhythm is actually a normal part of aging.

Until then, though, it is clear that--on average--AF screening will benefit doctors, hospitals, and device makers far more than consumers.

Caveat emptor.