Claude Nguyen, MD

Kishore A, Vail A, Majid A, Dawson J, Lees KR, Tyrrell PJ, and Smith CJ. Detection of Atrial Fibrillation After Ischemic Stroke or TransientIschemic Attack: A Systematic Review and Meta-Analysis. Stroke. 2014

Detection of atrial fibrillation (AF) is of great importance, as this often provides the necessary evidence for starting anticoagulation therapy. This task can be hindered by the mode of detection, as well as the duration applied. Kishore et al. from the University of Manchester performed a systematic review and meta-analysis of 32 prospective studies dealing with detection of AF on the various ways to detect AF after stroke or TIA. They sought to determine the overall rate of detection of AF, comparing stroke patients who were selected for detection versus those who were not. They also aimed to determine the optimum time of monitoring for these patients.  

Overall, they found that the detection rate varied greatly, with an overall detection rate of 11.5%, with a high heterogeneity index. They also found that detection rate was higher among selected patients compared to unselected patients (13.4% versus 6.2%), and that AF was found in 15.9% of cryptogenic strokes. When looking exclusively at inpatient detection, AF was detected to a higher degree in those selected for monitoring compared to those who were not.  They also found that prolonged monitoring increased the detection yield; however, these studies used varying time frames and monitoring modalities. Ultimately, the group concluded that prolonged periods of monitoring were more beneficial. The analysis was limited by the heterogeneity of populations and monitoring characteristics, largely because these studies did not report their criteria for diagnosing AF. 
Kishore et al. rightfully conclude that patient selection and more prolonged non-invasive monitoring tends to increase AF detection yield.  They also emphasize that in order to better define monitoring characteristics in the future, standard definitions of PAF should be used.  This would certainly make sense, as it is obviously important in any search strategy to define one’s target, narrow the search window, and ensure an adequate length of search.  Only then, can we potentially lower the percentage of cryptogenic strokes, many of which are sure to be cardioembolic.