Interobserver variability in EEG interpretation.

A random sample of 100 active electroencephalographers in the United States evaluated 10-second samples of 12 selected EEGs. The evaluations consisted of multiple-choice questions related to the age of the patient, EEG finding, artifact, and consciousness of the patient. The rate of reporting the "correct" response was examined in terms of various respondent characteristics such as EEG board certification, age, percent of time in clinical EEG work, and number of recordings interpreted annually. This study indicates that, even today, there is considerable variability in EEG interpretation, and that this variability is influenced by specific reader characteristics.