June 26th, 2014

To Screen or Not for AF: Is That the Question?

Up to one third of ischemic strokes are of uncertain etiology or “cryptogenic,” meaning that there is no overt explanation such as severe carotid stenosis, intracranial stenosis, or documented atrial fibrillation (AF). For some of these patients, intermittent AF may be present and could explain the ischemic stroke. This issue has been addressed by two recent studies published in the New England Journal of Medicine.

In the EMBRACE study, researchers randomized 572 patients aged 55 or older with cryptogenic stroke to either 30-day electrocardiogram (ECG) monitoring with a portable event monitor or to a standard 24-hour ECG recording. The primary outcome was newly detected AF lasting 30 seconds or more within 90 days of study entry. This endpoint was detected in 16.1% of the prolonged monitoring group, compared with 3.2% in the control group (P<0.001). By 90 days, significantly more patients in the prolonged monitoring group than in the control group were placed on anticoagulants (19% vs. 11%).

In the CRYSTAL AF study, investigators randomized 441 patients aged 40 or older with cryptogenic stroke to receive an insertable cardiac monitor (ICM) or undergo conventional follow-up. AF lasting more than 30 seconds was detected in 8.9% of the ICM patients versus 1.4% of the control group by 6 months (the primary endpoint; P<0.001). At 12 months, the rates were 12.4% vs. 2.0% (P<0.001). The median time from randomization to the first detection of AF was 84 days with ICM and 53 days with conventional follow-up.

Comment: Atrial fibrillation is the leading cause of cardioembolic stroke and is associated with severe strokes and high mortality. Therefore, identifying its presence in patients with stroke or transient ischemic attack is important. At face value, screening for AF in patients with strokes of uncertain cause would seem sensible.

In these two studies, brief episodes of AF were fairly common, in the range of 12% to 16% over a period of 1 to 12 months after the stroke event. The number needed to screen ranged from 8 to 10. However, some relevant uncertainties remain. Is a brief episode of asymptomatic AF, especially if detected months after the stroke, causally related to the stroke? How long must an AF episode last to be clinically significant, and when should it trigger initiation of anticoagulant therapy? Which stroke patients would benefit the most from AF screening? Screening for AF in patients with cryptogenic stroke likely should increase, but these questions must be addressed to determine how often we should screen for and treat AF in such cases.

–Seemant Chaturvedi, MD

Dr. Chaturvedi is Professor of Neurology at Wayne State University School of Medicine and Director of the Wayne State/Detroit Medical Center Comprehensive Stroke Program.
Reprinted from NEJM Journal Watch Neurology

3 Responses to “To Screen or Not for AF: Is That the Question?”

  1. David Powell , MD, FACC says:

    This is an ongoing issue at our hospital. The cause and effect relationship between afib and the prior CVA is tenuous. Indeed, one study found no temporal correlation between embolic events and detection of atrial fibrillation.
    How timely must a TEE be performed to achieve high sensitivity for atrial thromboembolism as the source of a recent CVA? If a TEE done a few hours after a CVA shows neither thrombosis nor signs of atrial stasis, what is the utility of prolonged monitoring to detect a few minutes if afib a month later? The high rate of subsequent prescription of anticoagulants is actually worrisome to me in such a scenarion.
    After all, the studies are uncontrolled without endpoints. There is an obvious bleeding risk. Patient selection is key, as the initial neurological referral is not always preceded by a careful assessment of the stroke category (or even more concerning TIA). We should not be doing even TEES for lacunar strokes.
    The ability now to implant a monitoring device in 15 minutes is only accelerating a premature treatment strategy.

  2. Lakshmanan Sekaran, MD FRCP says:

    I am concerned with what iam going to do with my patient this morning in my office.
    The questions raised above are valid. There’s no clear definition for paroysmal atrial fibrillation. Consensus oipnion varies from 5 seconds or more. If present it may not be the cause of index event but may contribute to further events. For the time being we may have to discuss these facts with patient in front of us and decide on anticoagulation with agents whose bleeding risk is comparable with antiplatelets while we search for evidence from future trials.

  3. Dr Charurvedi reports on two just published trials of prolonged monitoring for aFib in patients with cryptogenic stroke. Both studies show that with prolonged monitoring more afib is discovered. However, I would like to echo his concerns regarding causality, which is truly the key question. Short episodes of afib may not necessarily cause CVAs and indeed may not even be associated with strokes (association, not causation, could be a result of a similar underlying process). While these studies are quite interesting, it is not clear that they should change practice at this time, especially with regards to anticoagulation. While anticoagulation does decrease thrombo-embolism in afib, it does have significant risks also.