Predictors of atrial fibrillation on implantable cardiac monitoring for cryptogenic stroke

BACKGROUND: Since the CRYSTAL-AF trial, implantation and usage of implantable loop recorder (ICM) after cryptogenic stroke (CS) for detection of atrial fibrillation (AF) has increased. However, it is unclear which CS patients would most benefit from long term ICM monitoring. This study aims to determine the risk factors in patients that would confer maximum benefit from ICM placement following CS. METHODS: A Columbia University Institutional Review Board (IRB) approved retrospective analysis of medical records of 125 patients with CS followed by implantation of ICM was evaluated. Univariable and multivariable time-to-event analyses were performed on demographics, hours of activity and variability (HRV), stroke location, thrombosis etiology, and CHA(2)DS(2) − VASc score. The primary outcome was presence of ICM-detected AF defined as AF lasting at least 2 min. RESULTS: One hundred twenty-five patients (mean 67.6 years ± 2.4 years, 60% male) were followed for at least 3 months. Twenty-two patients (18%) were found to have clinically verified detected AF; median of time to detection was 95 days. Upon univariable demographic analysis followed by multivariable Cox regression analysis, individuals with age 75 or older (HR: 3.987, p = 0.0046) or LVEF 40% and lower (HR: 3.056, p = 0.0213) had significantly higher risk of AF. Diabetics also had a lower AF detection in multivariable analysis (HR: 0.128, p = 0.0466). CONCLUSIONS: Age 75 or older and LVEF ≤40% were the factors on multivariable analysis that predicted AF detection. Diabetes is a possible significant factor which should be evaluated further. CHA(2)DS(2) − VASc score was notably not predictive of AF detected on ICM. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s10840-021-00985-1.