percentage time in afib and stroke risk

A recent retrospective cohort study found that the incidence of stroke in patients with atrial fibrillation was dependent on the percentage of time they were in afib (see afib stroke risk inc with more afib jamacardiol2018 in dropbox, or doi:10.1001/jamacardio.2018.1176)

Details:
-- 1965 eligible adult patients, from the Kaiser Permanente Real-World Heart Monitoring Strategy Evaluation, Treatment Patterns, and Health Metrics in Atrial Fibrillation (KP-RHYTHM) database, who had paroxysmal atrial fibrillation during 14 days of continuous ambulatory EKG monitoring, from 2011 to 2016. None were on anticoagulation
-- mean age 69, 45% women, 75% white/5% African-American/14% Asian, 20% diabetes/9% heart failure/62% hypertension/3% prior ischemic stroke/2.5% catheter ablation during the prior 12 months, 24% eGFR<60,1% had anticoagulants in the past 90 days, 39% anticoagulants in the 30 days post-monitoring
-- mean CHA₂DS₂-VASc score 2.6, ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) risk score was 4.3 (ATRIA provides more discrimination than CHA₂DS₂-VASc in several studies, with a range of 0 to 15, where 0 to 5 is low risk of thromboembolism, 6 is moderate risk, and ≥​7 is high risk)
-- afib burden 4.4%, total duration of atrial fibrillation 710 minutes, the median duration of the longest episode of atrial fibrillation 171 minutes, varying from 49 to 590 minutes.

Results:
-- there were 29 validated thromboembolic events (19 ischemic strokes, 8 TIAs, 2 other arterial thromboembolic events) with an unadjusted thromboembolism incidence of 1.51 per 100 person-years.
-- patients with higher atrial fibrillation burden were less likely to be women or of Hispanic ethnicity, though had more prior cardioversion attempts
-- those in the highest tertile of atrial fibrillation burden (≥​11.4% of the time), had more than a threefold higher adjusted rate of thromboembolism while not taking anticoagulants, HR 3.13 (1.50-6.56), as compared to a combination of the 2 lower tertiles
-- adjusting for CHA₂DS₂-VASc or ATRIA did not materially affect the results
-- the duration of the longest observed episode of atrial fibrillation was also not associated significantly with thromboembolism

Commentary:
-- this is a useful study because the question has been largely unanswered: whether there is a meaningful threshold of the amount of time in atrial fibrillation as a predictor of thromboembolic events in patients without implanted devices such as pacemakers, cardioverter defibrillators or resynchronization therapy devices
-- there are also conflicting studies in the literature about the relative role of paroxysmal atrial fibrillation vs more sustained a fib, with several study showing at least equal risk. Other studies, however, have found a higher risk in those with non-paroxysmal afib but not on anticoagulants.  One concern about paroxysmal afib is that fewer patients with paroxysmal afib tend to be on protective anticoagulants (hence the importance of these studies/blogs…)
-- Another evaluation (see Boriani B. Eur Heart J 2014; 35(8):508) analyzing 5 perspective studies of patients who had implanted electronic devices but did not have permanent atrial fibrillation found that with median follow-up 24 months, 43% of 10,016 patients experienced at least one day with the least 5 minutes of atrial fibrillation burden, and one hour of atrial fibrillation was associated with the highest hazard ratio of 2.11 for ischemic stroke. Five minutes, the lowest of the prespecified cutpoints, was associated with a hazard ratio of 1.76
-- there are a few inherent concerns about this study which might limit its generalizability:
    -- the minimal longest duration of atrial fibrillation in this group with paroxysmal afib was 49 minutes, much more than what might be considered clinically important (see yesterday’s blog). This might explain why the duration of the longest episode of afib did not affect the results (all episodes were really long…)
    -- this was a pretty low-risk group: though the CHA₂DS₂-VASc risk score was slightly more than 2 (though the bottom cutpoint for “high risk”, the ATRIA metric was in the “low risk”). This might explain why adjusting for these 2 metrics did not materially affect the results. But this might limit generalizability to higher risk groups
    -- is it really the percentage of time in afib that matters? Or is it the percentage of time but also the length of afib episodes? (as noted in this study the minimum was 49 minutes, way above the 30-second or even the 5-minute threshold referred to in yesterday’s blog: so is a large time in afib but only lots of 5 second bursts really the same as a similar total time but fewer 49 minute bursts?? Would a higher risk group who had very high CHA₂DS₂-VASc or ATRIA scores influence the attributable risk of the time in afib?
    -- does the fact that 39% were put on anticoagulants post-monitoring in this study distort the results?
    -- and, this was a large data-mining study of primarily insured adults in California, and they did not have information on why these patients were receiving arrhythmia monitoring. And as an observational study one cannot determine causality (is atrial fibrillation a marker of stroke risk or causative?).
    -- I am also not sure from my review how conclusive a 14-day monitor is vs a 30-day ambulatory monitor.

so, this is another potentially important piece in understanding the role of atrial fibrillation in systemic embolization and especially in stroke/TIAs. Again, as with the blog yesterday, this adds to the impetus to perform longer rhythm monitoring in those with paroxysmal atrial fibrillation (though the 30-day monitor may make more sense than the 14-day one in this study), yet add % of time in afib is another potentially useful, measurable quantity. And, as one of the studies noted yesterday, and as with most laboratory measurements, the results should probably be taken in the context of the overall patient condition (eg, pretest probability).  ie, we should also add in some assessment of baseline thromboembolic risk (CHA₂DS₂-VASc or ATRIA ​)


yesterday’s blog: http://gmodestmedblogs.blogspot.com/2018/05/rivaroxaban-or-aspirin-for-cryptogenic.html reviews several of the studies on paroxysmal afib and stroke risk
http://gmodestmedblogs.blogspot.com/2018/05/resolved-atrial-fibrillation-and-stroke.html for blog on importance of continuing anticoagulation in those with “resolved” afib and high CHA₂DS₂-VASc risk score
http://gmodestmedblogs.blogspot.com/search/label/atrial%20fibrillation for a slew of blogs on afib, as well as critique/concern regarding the direct oral anticoagulants.