Accuracy of electrocardiogram interpretation by cardiologists in the setting of incorrect computer analysis

J Electrocardiol. 2006 Jul;39(3):343-5. doi: 10.1016/j.jelectrocard.2006.02.002.


Background: Overreading of 12 lead electrocardiograms (ECGs) is required to circumvent errors of computerized ECG interpretation. The accuracy of the overreading physician's interpretation of ECGs that were incorrectly read as atrial fibrillation by a computer algorithm has not been systematically examined.

Methods: A total of 2298 ECGs with the computerized interpretation of atrial fibrillation from 1085 patients were analyzed by 2 electrophysiologists, who identified 442 ECGs (19%) from 382 patients (35%) that were incorrectly interpreted as atrial fibrillation. Charts were reviewed to determine the interpretation of the ECG by the ordering physician (primary reader) and the overreading cardiologist.

Results: Cardiologists as primary readers more often corrected the misinterpreted ECGs as compared with internists, emergency physicians, or other specialists (94% vs 71%, P < .001). Surprisingly, interpretations by cardiologists as primary readers were more accurate than the interpretation provided by overreading cardiologists (94% vs 72%, P < .001).

Conclusion: Knowledge of an individual patient on whom an ECG is ordered may result in a more critical rhythm assessment and might account for the higher accuracy of rhythm interpretation by the cardiologist as compared with the interpretation by the overreading cardiologist who is lacking relevant clinical information.

MeSH terms

  • Atrial Fibrillation / diagnosis*
  • Atrial Fibrillation / epidemiology*
  • Cardiology / statistics & numerical data*
  • Diagnosis, Computer-Assisted / statistics & numerical data*
  • Diagnostic Errors / prevention & control
  • Diagnostic Errors / statistics & numerical data*
  • Electrocardiography / statistics & numerical data*
  • Humans
  • Prevalence
  • Professional Competence
  • Quality Assurance, Health Care / methods
  • Reproducibility of Results
  • Sensitivity and Specificity
  • Specialization / statistics & numerical data*
  • United States / epidemiology