If the machine says 'Normal' on the ECG, can I put the patient in the waiting room?: a study on the reliability of ECG software classification

Autor: Afnan S. Sharourou, Omar T. Idrees, Ahmed L. AlManzalawi, Amro M. Tayeb, Thamer A. AlGhamdi, Anas Fouad Hamam
Jazyk: angličtina
Rok vydání: 2023
Předmět:
Zdroj: Saudi Journal of Emergency Medicine, Vol 4, Iss 1, Pp 001-005 (2023)
Druh dokumentu: article
ISSN: 1658-8487
16732049
DOI: 10.24911/SJEMed/72-1673204949
Popis: Background: Electrocardiograms (ECGs) are one of the most basic and fundamental screening tools used in the emergency department (ED). Previous studies have shown machine diagnosis of ECG to be unreliable, the ECG machine does provide a simpler classification of: 1- Normal, 2- Otherwise normal, 3- Borderline, and 4- Abnormal printed on the ECG. We aim to investigate if machine classification could be used reliably as a screening test for triage. Methods: This cross-sectional study was conducted from 1 to 14 June 2019 of ED at King Abdullah Medical Complex using electronic medical records. The ECGs were put into sets of 25 traces/set and then presented to 21 board-certified emergency medicine attending physicians (EMPs) to assess and decide on one of the actions: Put in the waiting area, see in triage, or admit immediately. The responses were analyzed for inter-subject correlation coefficient kappa (k). Results: Of the 3,149 patients, 452 had ECGs done from which 200 ECGs were chosen at random. The inter-subject correlation coefficient was found to be 0.315 ± 0.187 denoting a fair to moderate correlation. From the ECG traces classified as "Normal" by the ECG machine, only 46% ended up in the waiting room. While almost 15% were admitted immediately to an ED bed. In contrast, 27% of those labeled as "Abnormal" ended up in the waiting room, while 44% were admitted to an ED bed. Conclusion: The machine classification of the ECG traces unfortunately failed remarkably to predict the EMP's decision. As such, the assessment of the attending EMP remains a necessary and essential part of the assessment. [SJEMed 2023; 4(1.000): 001-005]
Databáze: Directory of Open Access Journals
Externí odkaz: