Predicting atrial fibrillation in ischaemic stroke.

Abstract

Background

Atrial fibrillation (AF) holds the dubious honour of being the most frequent cause of embolic stroke. Up to 30% of patients with embolic stroke of unknown source (ESUS) have demonstrable AF on monitoring. Oral anti-coagulants significantly reduce the risk of ischaemic stroke in patients with AF. However, current guidelines do not recommend the commencement of an anticoagulant without documented AF. AF is not always captured on traditional 24-hour telemetry, and loop recorders and Holter monitors, while sensitive, are invasive and costly. Given the pitfalls of accurately detecting AF, should the focus be on prediction instead?

Objectives/Aims

To conduct a systematic review of the literature to identify clinical predictor tools for AF.

Method

We conducted a systematic review of the literature to identify clinical predictors of AF. Databases interrogated included MEDLINE, CINAHL, ScienceDirect, Informit Health Collection, Directory of Open Access Journals, Research Starters, Cochrane Library, and PubMed, Embase, Web of Science, and Oxford Medicine Online. We included original research published in a peer-reviewed journal between May 2017 and January 2018, available in English.

Results

Our review of the available literature has identified 11 tools that reliably predict the development of atrial fibrillation, with a C statistic ranging from 0.713 to 0.90. Commonly included variables were age, hypertension and left heart strain based on ECG and/or echocardiography. Some of these tools also incorporate genetic features and biochemical markers. Surprisingly, none of these tools are routinely used in clinical practice.

Implications/Outcomes for Planned Research Project

We suggest that the risk of developing AF can be predicted and should be part of routine clinical assessment of ESUS patients. Moreover, the addition of more recently identified variables of AF risk to preexisting tools may improve their sensitivity.

Final Thoughts

The ideal tool would be applicable in both advanced and resource poor regions.