Paroxysmal atrial fibrillation (Afib) detection in cryptogenic stroke is difficult but essential because it changes management. We describe a scoring system that discriminates between cryptogenic ischemic stroke patients with implantable loop recorder (ILR) that were and were not found to have Afib. Consecutive cryptogenic stroke cases from cardiology’s ILR registry for a 2-year period (7/2017-7/2019) were reviewed. We used standardized case report forms to perform chart abstraction. Cases were excluded if ILR was not placed after the index stroke event, stroke etiology workup was not available, or data was incomplete. Patients found to have Afib on ILR were compared to those without evidence of Afib on ILR. We devised a novel scoring system using variables associated with Afib detection and compared its ability to classify Afib detection against CHA2DS2-VASc and LADS. One hundred fifty-seven patients met inclusion criteria. Afib was detected in 12% of cases (9% at 6 months, 10% at 12 months). The median time from ILR placement to Afib detection was 110 days (IQR 37, 507). Median time from Afib detection to the start of anticoagulation was 3 days (IQR 0, 8). The PAL-CrISP score ranges 0 to 7: age (70=0, ≥70=4), history of antihypertensive medication (no=0, yes=2), PR interval (≤200msec=0, >200msec=1). Of those found to have Afib via ILR, 74% (14/19) had a PAL-CrISP score ≥ 6. PAL-CrISP performed better at predicting Afib detection in cryptogenic ischemic stroke patients with ILR (AUC 0.810, 95% CI 0.706-0.913) than CHA2DS2-VASc (AUC 0.650, 95% CI 0.525-0.774) and LADS (AUC 0.745, 95% CI 0.624-0.866). Using only age, home medication review, and an EKG, the novel PAL-CrISP score performs better at predicting Afib detection than the CHA2DS2-VASc and LADS scores in cryptogenic ischemic stroke patients with an ILR.
An unsolved issue: utility of implantable loop recorder monitoring in cryptogenic stroke patients
