Semi-automatic software based detection of atrial fibrillation in acute ischaemic stroke and transient ischaemic
M N Nickelsen1, A Snoer1, A M Ali1
1Neurovascular Centre, Department of Neurology, Faculty of Health and Medical Sciences, Zealand University Hospital, University of Copenhagen, Copenhagen, Denmark.
Background And Purpose:
Paroxysmal atrial fibrillation (PAF) is often asymptomatic and increases the risk of ischaemic stroke. Detection of PAF is challenging but crucial because a change of treatment decreases the risk of ischaemic stroke. Post-stroke investigations recommend at least 24-h continuous cardiac rhythm monitoring. Extended monitoring detects more PAF but is limited by costs due to manual analysis. Interpretive software might be a reasonable screening tool. The aim was to validate the performance and utility of Pathfinder SL software compared to manual analysis.
Methods:
In all, 135 ischaemic stroke patients with no prior history of PAF or atrial fibrillation and who had done a 7-day continuous electrocardiogram monitoring (Holter) were included. Manual analysis was compared with Pathfinder SL software including a systematic control of registered events.
Results:
Seventeen (12.6%) patients were diagnosed with PAF (atrial fibrillation > 30 s). Pathfinder SL software including a systematic control of events registered 16 (94.1%) patients with PAF. Manually 15 (88.2%) patients were detected with PAF. Pathfinder SL had a negative predictive value of 99% and sensitivity of 94%.
Conclusions:
Pathfinder SL software including a systematic evaluation of events is an acceptable alternative compared to manual analysis in PAF detection following ischaemic stroke. It is less time consuming and therefore a reliable, cheaper alternative compared to manual analysis.
