anticoag for atrial fib

am acad of neurology just published new guidelines on stroke prevention in patients with nonvalvular atrial fib (see afib anticoag neuro guidelines 2014 in dropbox, or DOI 10.1212/WNL.0000000000000145). their findings:

    --in patients with cryptogenic stroke, most studies used Holter monitoring, followed by serial EKGs, event monitors, etc, with monitoring duration from 24 hrs to 30d. pickup of nonvalvular afib (NVAF) was 0-23%, average 10.7%. increased pickup with increased duration of monitoring.

    --INR goal for warfarin therapy is 2-3

    --non-warfarin therapies: dabigatran is "probably more effective" than warfarin. hemorrhage risks similar, though less intracranial hemorrhage and more GI bleeds. rivaroxaban is "probably as effective as warfarin", no diff in risks of bleeds (also more GI and fewer intracranial). apixaban "likely more effective than warfarin", with decreased bleeding and reduced mortality. oral anticoag is superior to clopidogrel plus aspirin, though less intracranial bleeding with clopid/aspirin.  combo of warfarin and low dose aspirin increases hemorrhage and no evidence that more effective than warfarin itself. apixaban "likely more effective than aspirin" with similar bleeding risks. clopidogrel plus aspirin better than aspirin but more hemorrhage.

    --in those over 75yo, warfarin (INR 2-3) better than low-dose aspirin with half the incidence of stroke and similar extracranial hemorrhage rates.

    --in pts with chronic kidney disease, warfarin (INR 2-3) better than aspirin or low-dose warfarin. apixaban better than aspirin. warfarin-related bleeding risks higher in patients with CKD.

so, recommendations:

    --clinicians "might obtain cardiac rhythm studies for prolonged periods" (1 or more weeks) instead of shorter (24 hours) in pts with cryptogenic stroke without known NVAF

    --if anticoagulation seems appropriate to clinician and patient, clinicians should choose warfarin (target INR 2-3), dabigatran 150mg bid if CrCl >30, rivaroxaban 15mg/d if CrCl 30-49 or 20mg/d otherwise, apixaban 5mg bid if creat 1.5-2.5 and body wt <60kg and/or >80 yo. can also use triflusal 600mg (not avail in US) plus acenocoumarol (target INR 1.25-2)

    --"clinicians should administer dabigatran, rivaroxaban or apixaban" if patient at higher risk of intracranial bleed, and "might offer" apixaban" if higher GI bleeding risk. in fact, they seem to prefer apixaban overall: "clinicians should offer apixaban to patients" unsuitable for or unwilling to take warfarin; and "where apixaban is unavailable, clinicians might offer dabigatran or rivaroxaban"

    --in terms of elderly pateints with frequent falls or advanced dementia, they point out that data are insufficient as to safety of anticoagulation, but a decision analysis modeled that an elderly person would need to fall 295 times in 1 year to offset the stroke benefits of warfarin.

    --insufficient evidence for recommendation for patients with end-stage renal disease

so, one concern here is that lots of people have NVAF as they get older, picked up especially with prolonged monitoring. not sure how to interpret shorter runs of afib (esp if <30 seconds) -- are these significant???  these runs tend to get longer as people age. and, given the potentially profound implications of a recurrent disabling stroke, and given the high rate of stroke even in patients who have only had a TIA at 10%/year, the better part of valor would be to treat (this of course assumes that the clinician and patient feel that the benefits outweigh the risks in individual cases). as per previous emails/blogs, i am concerned about these newer agents, esp dabigatran, and the variability of blood levels/risks of bleeds, and have been using warfarin for years without significant problems (monitoring patients closely, recognizing the drug-drug interactions and decreasing doses necessary in elderly....). also, in terms of evaluating patients, if normal EKG, should we start with a Holter for 24 hours and then proceed to longer monitoring if the Holter is negative?  to me, that seems more reasonable and less intrusive than jumping into a more prolonged and often unnecessary 7- or 30-day monitoring.