cryptogenic stroke and atrial fibrillation

background: stroke is common, but 20-40% of strokes and 50% of TIAs have no evident cause after work-up (called "cryptogenic"). since afib is a common and treatable cause of stroke (and the treatment of afib with anticoagulants is different from that of other strokes with antiplatelet agents), and since those with untreated afib have higher likelihood of recurrent stroke, it seems reasonable to look hard for afib. 2 recent articles in NEJM tried to assess this --

the CRYSTAL AF trial (see stroke and cryptogenic afib1 nejm2014 in dropbox, or DOI: 10.1056/NEJMoa1313600) was an RCT of 441 patients >40yo (mean age 61.5) with cryptogenic stroke in prior 90 days (negative workup including 24-hr Holter) randomized to a long-term insertable cardiac monitor (ICM) vs conventional followup. primary endpoint was time to first detection of afib lasting >30 sec within 6 months. secondary endpt was time to first detection of afib within 12 months.  Mean baseline CHADS2 score of 3 points​. results:

    ​--by 6 months: afib detected in 19 patients in ICM group (8.9%) vs 3 pts (1.4%) of controls

    --by 12 months: afib in 29 ICM pts (12.4%) vs 4 control pts (2%), mean rate of detection of afib was 84 days after randomization. detection of afib in ICM group continued to increase after 36 months

the EMBRACE trial (see stroke and cryptogenic afib2 nejm2014 in dropbox, or DOI: 10.1056/NEJMoa1311376) was an RCT of 572 patients >55yo (mean age 72.5) who had cryptogenic ischemic stroke or TIA (63% had stroke) within past 6 months of undetermined cause (including negative 24-hr Holter as well as other standard tests), randomized to additional ambulatory EKG with either a 30-day event-triggered recorder (intervention group) or another 24-hr Holter (control group). Mean baseline CHADS2 score of 3 points. Primary outcome was newly detected afib lasting 30 seconds or longer within 90 days after randomization.  secondary outcome was episodes of afib lasting 2.5 minutes or longer and anticoagulation status at 90 days.  results:

    --atrial fib > 30 seconds in 45 of 280 pts (16.1%) in the intervention group and 9 of 277 (3.2%) in control group.

    --atrial fib > 2.5 min in 28 of 284 pts (9.9%) in intervention group vs 7 of 277 (2.5%) of controls; oral anticoag prescribed for 52 of 280 pts (18.6%) in intervention group vs 31 of 279 pts (11.5%) in controls

although these results are interesting (and made it into the popular press), i do not think they provide actionable conclusions. not exactly shocking that prolonged monitoring picked up more episodes of afib. but for me, the primary issue is not so much whether there were small episodes of atrial fibrillation only detectable with prolonged monitoring but whether these episodes had anything to do with the antecedent stroke, or even if the increased monitoring provides any utility in determining whether the patient is at risk for another stroke. these types of afib episodes are common, especially as people get older (estimates of 4% of those >60yo have paroxysmal afib, defined as at least 2 episodes of afib  >30 sec long in 7 day period. 90% are asymptomatic. episodes can vary greatly in length). so, it would have been useful to have a real control group (matched group without stroke) to see if the incidence of afib picked up on prolonged monitoring was higher in the group with stroke than in non-stroke controls. and, the real test: whether treating patients with afib found on prolonged monitoring (with warfarin, for example) actually decreased the incidence of subsequent strokes. these studies would help sort out such issues as well as: does a 30 second episode of afib matter? or a 2.5-minute one? what is the appropriate frequency and/or length of these afib episodes that confer a higher stroke risk?  also, as a matter of perspective, 25%+ of strokes (>3 million strokes) are cryptogenic and this aggressive monitoring still misses 85% of them (ie, the issue of cryptogenic strokes remains huge even if it turns out that by finding/treating rare afib events that benefits>risks -- esp since anticoag therapy is not exactly benign, both medically and socially)).