By Cara Pellegrini, MD
Assistant Professor of Medicine, UCSF; Cardiology Division, Electrophysiology Section, San Francisco VA Medical Center
Dr. Pellegrini reports no financial relationships relevant to this field of study.
SOURCE: Turakhia MP, et al. Atrial fibrillation burden and short-term risk of stroke. Case-crossover analysis of continuously recorded heart rhythm from cardiac electronic implanted devices. Circ Arrhythm Electrophysiol 2015;8:1040-1047.
While atrial fibrillation (AF) is well-established as a major risk factor for ischemic stroke (five-fold risk), the temporal relationship between AF and stroke remains unclear. Also unknown is what AF burden, or dose threshold, is needed to generate sufficient risk to warrant the potential harms of long-term anticoagulation. This is a particularly relevant question with the growing number of implanted cardiac devices detecting asymptomatic AF, expanding the potential at-risk population. Similarly, the availability of novel oral anticoagulants with rapid onset of action has made possible the option of “pill-in-the-pocket” anticoagulation. If periods of higher risk could be better defined, then perhaps therapy aimed at stroke reduction could be more precisely targeted.
A cleverly designed case-crossover study by Turakhia et al takes aim at some of these questions. They identified almost 10,000 patients with implanted cardiac devices with atrial leads in the Veterans Health Administration, and from these, sub-selected 187 patients with an acute stroke and 120 days of continuous cardiac monitoring prior to the stroke. They used each patient as his own control, comparing daily AF burden in the 30 days immediately prior to the stroke (case period) to the AF burden 91-120 days pre-stroke (the control period). This technique matches all stable confounders and allows for assessment of risk of a time-varying exposure on an outcome with a brief latent period. They hoped to address AF threshold doses needed to increase risk by varying their definition for AF burden and temporal relationship of AF burden to stroke by examining risk at varying temporal proximities to the stroke event.
The researchers reported an incidence rate of 1.63 strokes/100 person-years in their virtually all-male veteran population, with 39% previously diagnosed with AF and a mean CHA2DS2 score of 3.2/CHA2DS2-VASc of 4.8. An AF burden of ≥ 5.5 hours in any 24-hour period correlated with a four-fold risk of stroke, after adjustment for warfarin usage. The odds ratio for stroke was highest in the 5 days immediately following an AF occurrence at 17.4 (95% confidence interval, 5.39-7.31), declining steadily thereafter, with a notable drop 20 days after the AF event to ≤ 3.5. Unfortunately, they were unable to define the dose-response of the amount of AF on stroke risk due to the small number of discordant cases (AF present in case period but not in control period). The authors concluded that multiple hours of AF had a strong but transient effect raising stroke risk, suggesting that a strategy of intermittent targeted usage of rapidly acting anticoagulants might merit further consideration by a randomized trial.
COMMENTARY
This study extends the findings of previous work by better defining the temporal relationship between an AF episode and stroke risk. At the same time, it corroborates previous findings that proximate AF is only a minority contributor to ischemic stroke incidence, even among those with a known diagnosis of AF. In this study, 31 patients with a history of AF had an AF event sometime during the 120 days preceding their stroke; the vast majority (28 patients) had an AF event in the 30 days prior to their stroke. However, an even larger number (42 patients) had a stroke and a history of AF but no AF events during the 4-month pre-stroke period. In ASSERT, another trial of patients with implanted cardiac devices as well as a history of hypertension but no previously diagnosed AF, only 50% of the patients with stroke had subclinical AF discovered and only 8% had AF within 30 days of the stroke. Likewise, in yet another study (TRENDS), while AF detected by an implanted cardiac device increased stroke risk, only about 25% of patients had an episode of AF within a month of the event.
Thus, while a period of AF clearly raises stroke risk in a now more defined way, there are at least an equal number of strokes occurring in people with a known diagnosis of AF with no proximate AF episode. Whether this is due to a longer prothrombotic state in the atrium following a previous AF episode or competing risk of stroke due to often-related comorbidities in AF patients, such as hypertension, is unclear. These questions have important implications not only for the usefulness of a “pill-in-the-pocket” anticoagulation approach, but also for the expected benefit of an interventional stroke reduction therapy, such as placement of a left atrial appendage occlusion device.
The relationship of anticoagulation use or lack thereof to the findings of the study was mostly unexplored. The annualized stroke rate of 1.6% was substantially lower than expected, based on the reported CHA2DS2-VASc scores and reported warfarin usage of about 50% among those with a history of AF. While anticoagulation usage might have been higher than appreciated, one has to consider the possibility that device-detected AF might concur a lower stroke risk than clinically detected AF, making the question of how aggressively to seek and treat this entity a bit unclear. Another interesting observation on which the authors did not provide comment was the decrease in warfarin use immediately prior to the stroke event (5.2%) as compared to during the initial control period (19.8%). With all available data taken together, I still remain skeptical about the ability to predict those at sufficiently low-risk based on AF burden to warrant intermittent anticoagulation. Nonetheless, this study does provide ammunition for discussions with patients with an intermediate-to-high CHA2DS2/CHA2DS2-VASc score who are reluctant to be on anticoagulation long-term to at least consider short-term usage for the 20-30 days following a known AF episode.
