EHR data detects patients at a higher risk of AFib, stroke or heart failure
Researchers have developed a brand new prediction model capable of identifying patients at a heightened risk of atrial fibrillation (AFib), sharing their findings published in Circulation: Arrhythmia and Electrophysiology.
The team’s new AFib risk score was designed to be straightforward, using data commonly found in a patient’s electronic health record (EHR). The score, known simply as EHR-AF, showed potential—but it still needed to be properly confirmed before implementation could be recommended on a larger scale.
With this in mind, the group tested its score with EHR data from more than 4.5 million patients seen at a variety of facilities from January 1999 to December 2019. The mean patient age was 62.5 years old, and 56.3% were women. All data came from the IBM Explorys Life Sciences dataset.
Overall, the scores generated by EHR-AF “demonstrated good predictive accuracy” for AFib within five years. The EHR-AF score also outperformed other existing risk scores for AFib.
In addition, the authors noted, the scores were able to identify patients at an increased risk of AFib-related complications, stroke and heart failure. When the patient had a history of stroke or HF, however, the scores were less accurate, “indicating that improved methods of prediction may be necessary in these individuals.”
“Within an EHR-based sample including over 4.5 million individuals, we demonstrate that the EHR-AF score predicts incident AF risk accurately, comparing favorably to other established risk scores,” wrote lead author Shaan Khurshid, MD, Broad Institute of the Massachusetts Institute of Technology & Harvard University, and colleagues. “Importantly, EHR-AF was well-calibrated to observed AF risk, suggesting the ability to provide accurate AF risk estimation either individually at the point of care, or at the population level using readily ascertainable EHR data. Future work is warranted to assess whether routine deployment of EHR-based AF risk estimation to guide interventions to diagnose or prevent AF leads to improved outcomes.”
The full analysis is available here.
Michael has more than 18 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.